Skip to main content Skip to navigation

WEB-EM-8 abstracts and feedback

Name: Jess Nickson
Library Card Number: 0958330
Provisional Title: Blackball in JsEden


A game of blackball, or 'pool' as it is more commonly known, is a game where players aim to pot their balls in the pockets of a pool table while avoiding pocketing their opponent's balls. At the same time, a player must not pocket the black (eight) ball until all of their balls have already been pocketed.

The initial break of the balls in a game of pool can be a deciding factor on how the game will progress. A good break might cause all of the balls to separate out with the player's balls close to the pockets and his opponent's being clustered together in the middle of the table, which automatically gives the player an advantage.

However, there are a number of factors to take into account when making the break shot, which can massively change how the break turns out. These include the initial direction of the cue ball and how much force was put into the break shot. If the break is not good enough it can end in an illegal move, which allows the player's opponent to take the break shot.

Due to the complexity of the break shot, and how even tiny changes in the variables can have massive affects on the outcome of the shot, I have decided to try and model this initial break shot for my project.

My project will expand upon ideas present in previous billiards models in order to generate an improved model, which can account for all sixteen pool balls.

Model Description:

My project aims to model the break shot in a game of blackball (pool). The model will consist of sixteen balls (1xWhite, 1xBlack, 7xYellow and 7xRed) and six pockets on a pool table. The user can decide which of two arrangements to use for the racked balls i.e. red or yellow first. The balls will then be set up in the rack, which will be triangular in shape. The model will enable the user to set the location of the white ball to anywhere on the table (even illegal moves). The user can then line up where the pool cue should hit the white ball, and how much force should go into the hit.

Once these variable values have been chosen, the break shot will take place. The user will be able to view how the location, direction and force of the white (cue) ball affects the racked red, yellow and black balls. Different amounts of force may mean that the same hit (based on direction and location of the cue ball) will cause the racked balls to break in different ways, perhaps pocketing more balls or causing an illegal state.

The model will play out the shot by continually checking each ball for any collisions with other balls, and calculating momentum for each ball based on this. If balls hit the sides of the tables then they will rebound with the correct momentum. Also, if a ball is pocketed during the break shot, then scores will be calculated and the user will be assigned a colour in accordance with blackball rules. In this case, the player will be assigned the colour of the first ball their hit pocketed.

Once all of the balls' momentum has reached zero, the model will check the locations of each ball and indicate to the player if the break shot made ended in a legal move, in accordance with WEPF (World Eight-Ball Pool Federation) rules.

Once the legality of the shot has been determined, the user will be able to reset the model.

The model will be a 2D representation of the game. Momentum and friction will both be taken into account for ball movement. Some form of trial and error may be required to determine the most realistic co-efficient of friction to use, what range of force to provide for hitting the cue ball and how best to represent ball movement (e.g. moving a coloured circle or use graphics to make spin more obvious).

Further improvements to the model could be to continue the game past the break shot or introduce a front-on view of the cue ball so that users can line the cue up exactly how they want. This would include the ability to aim at the top or bottom of the cue ball, which can affect the spin on the ball. This addition may require gravity to be introduced into the model, as changing the vertical alignment of the cue can increase the chances of balls bouncing on the table.


Similar models:


A Definitive System for the Browser - Tim Monks


Weighting: Paper - 60/Model - 40 (Paper not exceed 6 pages)

This is an interesting topic for which you seem to have identified most of the key practical resources. You may also find it useful to consult - and critique - are CS_RR_260 and EM paper 48. Key issues that have been raised in previous discussions mentioned in these sources are: whether the construal will take full account of acts of intervention, such as obstructing the cue ball in motion, removing a ball from the pack before the impact of the cueball; whether the step-size in computing collisions will be adapted to take account of the speed of the cue ball; to what extent it may be appropriate to employ a pre-computing-then-replaying strategy. Whatever you decide in these respects, implementing your construal with JsEden will be novel and instructive - for instance, where coping with a dynamic observables is concerned.

Name: Yu Pang Yip
Library Card Number: 1060969
Provisional Title: Modelling a Battleship puzzle


This paper explores how Empirical Modelling and Artificial Intelligence can be used in constructing an improved version of a well known guessing game - Battleship. Battleship is a popular pencil-and-paper game that requires two players to play. We will be constructing the game for single player against AI. The EM approach of developing a game involving AI will be explored during the development process of the model.

Model Description:

The two player version of battleship game consist a battlefield of 10x10 cells. Ships are laid out either horizontally. The size of the ship is represented by numbers of cells it occupied. There will be one ship of size 5, one of size 4, two of size 3 and one of size two. Each player does not know the location of other player’s battleship. Each turn, players can choose one cell to fire at. Cells that have been fired will be marked in different color, if it was occupied by a ship it will mark by another color. The first one lost all the ships lost the game. We will construct the model based on these rules but will include an AI driven computer opponents. On each round of the game, there is a decision to be made for the AI, either it’s a trial shoot to randomly shoot at a cell to seek for a ship or a sinking shot to continue sinking.


CS405 lecture notes,
Notation documentations on EM website
External resources for implementing the AI algorithm.

Weighting: Paper - 30/Model - 70 (Paper not exceed 3 pages)

As we have discussed in the lab sessions, there is a considerable overlap between your proposal and that of Yile Liu (see below), and it will be important for you to make sure that your approaches are sufficiently independent. One way to do this would be for you to use JsEden. Your reference to "exploring the EM approach to development" gives more the flavour of what would make this proposal interesting than your chosen title "modelling a game of battleships". In some way, it's important to make more your submission more than 'a standard modelling exercise' - giving that the game itself is conceptually simple. Possible sources relevant to your proposal are wumpusCole2005 and a model of Mastermind from WEB-EM-7 [0705387 James Ribi / Inferences About The Game of Mastermind]. It would be a good thing to work on improving your English by discussing your text with a native speaker - for instance "Ships are laid out either horizontally" is not a complete sentence etc.

Name: Matt Cranham
Library Card Number: 1063531
Provisional Title: Parallel Parking Simulation


The purpose of this Empirical Modelling study is to model the process of parallel parking a car into a space between two other cars. Within the model, the user takes the role of an external observer who is able to investigate the effects of altering the values of various dependencies on the car performing the manoeuvre. This study also behaves as an investigation into whether the js-eden environment is currently mature enough to support the intended model.

Model Description:

The model I envisage producing will have two parked cars with a space between in which a third will attempt to parallel park into. The model will a two-dimensional aerial observation of the parallel parking manoeuvre. The car which will be performing the parking manoeuvre will be subject to dependencies within itself and its interaction with the environment it will be operating in.

I will also attempt to model Ackerman steering for the turning wheels, which is the turning of the outer wheel of a turn less than the inner wheel to observe the effect of the outer wheel travelling a further distance. This in itself is an interesting, and potentially difficult, Empirical Modelling exercise as it require will inevitably require a large number of dependencies to model the correct behaviour of Ackerman steering.

Another interesting aspect which would be useful to try and model would be that wheels generally only rotate in the direction they are facing, so the car shouldn't be able to travel in a direction against the direction of the wheel.

The user of the model will also be able to interact with the model to observe what different settings will make to the process of the car trying to park. Altering the rate of turn of the car, or the position as to where the turn starts to occur, will affect the potential for a successful manoeuvre into the parking space.


McHale, J. "The Car-parking Simulator", 2003
Monks, T. "A Deefinitive System For The Browser", 2011

Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)

In a model of this kind, the agent perspective you have in mind has a critical bearing on what you do. Your references bring this out quite vividly: "A sensor-guided parking system" / McHale's driver-oriented car-parking simulator with side and rear mirror views / Gardner's F1 observation of a track that focuses on abstract location at which to accelerate, turn and decelerate. Since all the above models abstract away from the mechanics of steering the vehicle, and you have drawn explicit attention to this, it probably makes sense to focus your attention on the observables that relate to the mechanics of steering e.g. to give a visual representation of how the steering wheel is turned as the parking manoeuvre is carried out. As far as the steering mechanism itself is concerned, a useful - but flawed - source may be Salman Shahid Butt 0961493 Car simulator model from WEB-EM-6. If you choose to use JsEden, this will indicate some of the Donald operators that you will need to implement in some way.

Name: Yile Liu
Library Card Number: 1155566
Provisional Title: An Empirical Approach to Airplane Fighter Game


This paper of this empirical modeling study is to model the game of guess the position of every plane in map. considering not only the model itself but also the efficient strategy to win the game. It will firstly discusses the approach to airplane fighter game with Empirical Modeling. One player will provide the layout of three planes in the 10 by 10 board, then they will not reveal on the board. Another player will find out every plans with inference. In addition, this model will give the every possible position of the head of plane after any action. With this prompt function, second player will find out the position of planes more efficient.

Model Description:

Firstly, we will implement the interface of the game. There is a 10 by 10 board, and many button such as start game, hide plane and rearrange. There is also a display board which output the possible position of current plane head. It will display related coordinate. Secondly, the process of playing game is designed as follow:

Firstly, first player provides layout of every plane. The shape of plane is like "士". The wings of plane will need 5 checks and body of plane will need 4. The empennage needs 3 checks. After arrange every plane, player one can choice hide planes or rearrange which will remove every plane and allow player one rearranges them.

Secondly, player two click start game button. Then player two input the coordinate of check which he supposes there have a part of plane. If there have nothing, it will no change about check, otherwise the check will display the part of plane. If the play two find out the head of plane directly, the plane will appear on the board. The game will be finished untill every plane be displayed on the board.

The main part of this model is figuring out every possible position of plane head after the player find out any part of plane. We presume what player two found is every possible part of plane, and depending on that, a corresponding position of head will be provided after computation. Player could win the game according to prompt information.The prompt board will change every time after the action of player two.


Hunt the Wumpus: an Empirical Approach

Beynon,W.M. Joy,M.S.:Computer programming for Noughts-and-crosses New frontiers.

Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)

This modelling exercise is very much the same as that discussed by Yu Pang Yip (see above). He is intending to implement his model using JsEden, and a suitable way to ensure independence between your models might be for you to use tkeden. Provided that you don't copy each other's work to closely, some cooperation may be welcome - for instance, you could include a section in your papers about the difference between a JsEden and a tkeden implementation, or write about the nature / your experience of collaborative development with EM principles and tools. Perhaps you could also consider (or at any rate discusss) how you might set up a game between your models (even if this had to be human-mediated). It is worth noting that the language you use for your model description is very procedural in character - you talk about implementing sequences of actions, but not about key observables and dependencies. For instance, you can surely define a dependency that identifies squares that might be part of a plane given that a certain pattern of squares has been exposed (or for instance identify when a particular pattern of exposed squares is inconsistent with the rules). Incidentally the title "Aircraft fighter game" is not very appropriate (it suggests something different by way of a game activity - you should also try to identify a purpose relating to EM for your modelling exercise, and convey this in the title. It would be a good idea to get a native speaker to review your English.

Name: Lung Li
Library Card Number: 1157177
Provisional Title: Idealist Timetabling


There are many software available to help in timetabling nowadays. Most of the existing software adopted the realist perspective and serve as a automatic machine with certain algorithm which require user to set the constraint preliminary, and the result generated has zero flexibility. However, these softwares ignore the fact that the environment is always changing and affect the decision of user at the same time, which leads to amendment in generated timetable.

A new way of timetabling is needed in order to reflect the reality. Instead of using fully automatic process, a mode of user-computer cooperation may be more suitable. User can always involve by adding new constraint to the current generated timetable, until a timetable which best meet the requirement of the user is resulted eventually. This is the idea of idealist timetabling.

This paper try to use Empirical Modeling to achieve such timetabling method. The model simulates the situation of oral presentation time slot allocation. There are students who need to present their project, and teachers who act as their supervisor. Certain classrooms are available for presentation. Each teacher, student and classroom has their own available time and is always subject to change. These changes are represented as agencies and affect the dependencies of certain observables, which is a certain time slot in this case, and changes the behavior of these observables. In this way, the timetable will always reflect the latest changes and shows the available time slots for certain student.

The ultimate goal is to obtain a timetable which meets everyone's requirement, that is all student can register a time slot for presentation in one of the classrooms with their supervisor available.

Model Description:

A smaller scale of class and teachers instead of full scale will be used in this model. The class size will be 10, and there are 5 teachers and 2 classrooms.

User can select as a certain student and classroom to view their timetable for oral presentation. The availability of certain time slot depends on the student’s supervisors and the classroom they selected. If the time slot is not available, it will be shown in red color, otherwise, green color will be used. User can register any available time slot for oral presentation and the selected time slot will be shown in orange color. Once a time slot for certain classroom and teacher is registered, it is no longer available to other students and will be reflected on their own timetable. Teachers and students can always change their mind, and the availability of classroom is subject to change at the same time. If the changes affect the availability of the registered time slot, the student registered for it has to select another time slot.


Weighting: Paper - 60/Model - 40 (Paper not exceed 6 pages)

This is an interesting and challenging exercise. As your cited sources indicate, we did once have a very effective timetabling model that was used for timetabling final year project orals with as many as 80 students. The source of this is not very well-captured in Keen's model from 2000, but is buried in the /dcs/acad/wmb/public/projects/misc/timetable/NEWTT/TT20xx directories, from which I've tried to resurrect it myself several times (and came close in 2010!). You may also find some helpful material in Rungrattanaubol's PhD thesis (see Chapter 4 p.114-125). EM paper 058 (from which you've borrowed the 'idealist timetabler' concept) should be amongst your references. Making a small-scale model of the original is a good idea - one of the main problems in the original model was the complexity of the mechanism used to construct the timetable interface, which had many hundred Scout windows. Even understanding enough about the original 'Temposcope' model to recreate a mini-version would be a considerable achievement. Ideally, a re-engineered model would make use of the EDDI notation (the dependency relations between timetabling data items in the Temposcope were all expressed in raw Eden, but could readily have been based on relational algebra operators). You might also find it easier to use Angel to model parts of the interface. Construal comprehension on a model of this scale would be enough to sustain a good WEB-EM-8 submission. One thing to be careful about: be sure that you give evidence in support of claims about the nature of current software applications for timetabling - some scholarship should be devoted to the theme of the kinds and characteristics of timetabling application that are out there, though I appreciate that this is a very large topic in itself.

Name: Zhetong Fang
Library Card Number: 1160398
Provisional Title: linear programming and optimisation


Linear programming and optimisation is widely used in computer science and business area, it is one of the best way to gice the best result with a given goal and several constraints. A 2D graph of Linear programming and optimisation is always used for teaching, in this paper I decide to make a model to help people to learn linear programming and optimisation, some part of non-linear programming may be considered.

Model Description:

There will be a user interface that people can change the goal or add/delete/modify the constraints, and 2D frame of axes will automatically update the information to the graph, and the best result will also be automatically calculated and shown on the screen. People may have the chance to see the non-linear best result (for example, the result must be integer).

If there is enough time, I will try to do a 3D model to make it more wildly used.


Antony Harfield, Meurig Beynon, "Graphics Presentation using 3D Room", Department of Computer Science, University of Warwick, 2007 TBC

Weighting: Paper - 30 / Model - 70 (Paper not exceed 6 pages)

In modelling exercises of this nature, there is an obvious way in which you get a product that has an "EM-like" quality. For instance, you can imagine an environment in which you set up a linear programming problem, display an optimal solution and then demonstrate how changing some of the key parameters of the problem changes the result "as an illustration of dependency". (The beamdetector model that was the focus of one of the construal comprehension exercises is a good example of this.) What is important is to be sure that you don't merely write a conventional solving program S e.g. in procedural Eden, then use this as a function in a "script" of the form s = S(x,y,z). What you need to aim for here is the kind of observation and experiment that might lead a learner to understand the principles of linear optimisation more fully. For instance, I have a vague memory of being guided through the principles of the Simplex algorithm, where you navigated from one vertex to another of a polygon seeking for an optimal value. Ideally, in the spirit of the heapsort construal demonstrated in the lectures, you should be trying to model the kind of environment that the lecturer would set up to explain how (e.g.) the Simplex algorithm works. This is more difficult than what you propose, as you will not be able to simply translate the code of an existing implementation into Eden (cf. the heapsort construal). In general, you need to put more emphasis on what EM brings to this exercise, both in your title (which is not really a title at all at the moment), and your abstract (which is really describing the narrow functional model that I'm critiquing above). I do like the idea of "making things more wildly used" though - this may not be what you meant to say, but it captures the spirit of EM well!

Name: Robert Schiller
Library Card Number: 1133024
Provisional Title: Sport is not just a game


Sport becomes more and more important for people all over the world. This has different reasons. On the one hand side people want to earn much money, on the other hand side it is just to come down after a stressful day, to establish relations or just to live healthy. The last reason touches most of the human beings. Some studies have shown, that the popularity gets fatter and fatter these days, because they don't move a lot, eat lots of fast food and sit always in front of a TV or a computer, but not everyone accepts its body and wants to do something to get thin. Thus the sport spectrum becomes broader, you might be able to find a variety of sport gyms, pitches and halls in each town. However, the prices rises enormously and many people haven't enough money to get their way into a gym or to book a pitch.

One of the cheapest way to do sport is definitely Running. You can go out off the house and can run wherever you want to. Nevertheless there are many different kinds of running. You can run fastly, you can do jogging, you can walking. Each kind depends on your mood, your fitness status etcetera.

That is why I want to do a model about Jogging. You are supposed to touch the kids early. They will be involved into a world of sport and health and can make their own opinions. It is not just a model for kids, but also for adults who don't know many about its. What I want to try is to involve all generations in a playful way. They shall realize, for example, how important it is to drink a lot or in general to take care about their bodies.

Model Description:

The keystone of my Model is to show how the dependencies of different agencies and observables are. You can, for example, choose between different fitness states(like fat or fit) and how the pulse, speed and thus the mileage might be changed. Also the mood plays a big role in sport, thus I want to have a choice between three different kinds of the mood, maybe depressive, normal and motivated. During the run it could be happen that you are injured, so for this random event there will be a button to cure its. There will be also a “Thirsty”-Button that you might be push if you are thirsty. How often you push both of these buttons depend on the position of the running style. In my model you can choose between three different kinds. You can walk, jog or you can run fastly. These are the most important agencies for me, because everything depends on the position of the style-button. There is also a reanimation button you can press, but just when the life of the man draw to an end. I want to implement some constraints, which shall include a maximum of pressing the liquid, reanimation and injury button. Thus the little game will finish at some point and your goal is it to reach the most mileage.

The design of my model will be simple and easy. In the middle of the screen a picture of a man is shown. Around this picture there are all the buttons. On the right hand side will be the the reanimation-button, the injury-button and the thirsty-button. On the left hand side there will be all the agencies like the style-button, the mood-button and the fitness-button and on the top of the screen you can see the observables like speed, pulse and an Timer.


Weighting: Paper - 40/Model - 60 (Paper not exceed 4 pages)

Your abstract addresses a promising theme, but doesn't give any indication of what your modelling exercise has to do with EM - though this does emerge more clearly from your Model description. Your references (!) are similarly non-specific: it might be appropriate to look at models like the lungs model from WEB-EM-1 ("EM and its Application to Human Biology") and the follow up paper in WEB-EM-5 ("Furthering the Application of Empirical Modelling in Teaching Human Biology"). There's obviously an issue about how far your data can be strictly faithful to experimental evidence, but it is important to recognise that even making simple construals is significant in health education. For instance, it is useful just to identify the key components of our body, the broad functions they perform ("the heart is an organ that pumps blood around the body") and the constraints within which they operate ("it is possible to have a heart attack"). If you genuinely want to construct a model that serves an educational function and encourages poeple to take up running, the challenge is to make a model in which the observables and actions are not so artificial that they don't connect with the real activity. On that basis, I'm not sure whether buttons to repair injury are appropriate! Perhaps better is something to indicate how long it would take to recover from an injury, what the consequences of the injury might be for health (e.g. how much weight would you put on if you couldn't run for 6 months!) and whether there would be permanent implications for fitness and performance. Developing that idea, it might be good to focus on symptoms that you may encounter when running: e.g. feeling faint on account of dehydration, muscle strains of various kinds, chest pains, so-called "stitch" in the side and what are responsible reactions to these. There are a number of strange quirks in your English - I think you mean "populace" where you write "popularity" for instance, and "run fast" not "run fastly". "Many about its" is also an odd construct - perhaps you mean "much about it"?

Name: Sebastian Krämer
Library Card Number: 1132361
Provisional Title: Empirical Model Simulation of Sport Betting Odds Calculation for Live Football Games


Sport betting is a huge market in the United Kingdom, but also within other European countries and especially in Asia. Millions of Pound are turned over every single day in the accounts of the millions of customers playing against hundreds of different bookmakers all around the globe, not even considering the vast amounts of money traded on betting exchanges like UK-based "Betfair", where people bet against other people without a real bookmaker calculating odds.

Markets are still evolving as some European countries like Germany are in a process of deregulating markets, driven by judgements of the European Court of Judgement. The sports with the highest turnovers are football, tennis and horse racing. This paper will focus on football bets; however basic theory described in it can also be adapted to tennis. Horse racing is slightly different so it wont be considered in this paper.

While in the beginning of the century, sport bets were mostly placed by phone, the development of the Internet and the general improvements in digital communication had their impact on the sport betting markets as well as in most other fields of life. Customers were more and more able to place bets online, providing new opportunities for bookmakers but also for customers, including "live betting", which is discussed later on.

In general, it is a common estimation that around 95% of all players lose against the bookmaker in long term or in life time. Nevertheless, with a rapid growth of online bookmakers, the general market competition increasingly tightened as some customers became aware of the possibility of calculating odds themselves to pick "good" odds ("value bets") or even outsmart different bookmakers with comparing odds of different bookmakers and betting on all outcomes of an event, creating outcome-independent, sure returns ("sure bets"). So, these are basically the only two ways to win with sport betting in long-term.

This paper will only briefly discuss the abstract roots of "value bets" and its impact on betting and won't consider "sure bets" by any means. Generally, a bet can be considered as "value", when the estimated probability of the outcome is higher than the odd of this outcome implies. That"s why the calculation of the probabilities of outcomes of certain events is the main business of bookmakers and the basic source of their income. Only players that can calculate odds better than bookmakers can win long-term with usual "punting" (placing bets on one outcome).

The aim of the paper is an approach on an empirical model that gives an abstract insight into the procedures and dependencies occurring in the process of calculating odds for live events. Those live events are not supposed to be happening in the real world yet, moreover the model is supposed to display a simulation of a football game.

Model Description:

Not all bookmakers are calculating their odds themselves, in fact, there are only very few that calculate odds on their own, the rest is just copying and slightly adopting other bookmakers odds. Most of the times, odds are initially published by large Asian bookmakers who are bigger than all European bookmakers, taking bets up to high limits. These high limit bets define the odds in the Asian markets (high bets on certain outcomes will lower the odd for this outcome while the odds of the opposed outcome rise), which are later on copied by European bookmakers. So what one can see in European bookmakers is mostly just of copy of the odds that "evolved" due to major bets in Asian markets. As these odds therefore are reflections of the bets of many players, including the best players (staking big amounts), the mechanism of crowd intelligence leads to the fact that most of the odds we see at bookmakers are very exactly calculated and most of the times appropriate to the real probability. So this doesn't leave much space for calculating better odds than bookmakers.

That's why this paper will focus on odds calculating for live betting.

Live betting is a relatively new field of betting which is still emerging. Customers can place bets while the actual event is going on, with odds continuously adapting to the events on the pitch. That makes it much harder for bookmakers to calculate exact odds, as the time for adapting and copying is missing, so it is easier for customers to beat the bookmaker by calculating better odds.

Odds are generally calculated considering various general "pre-game" parameters (observables in empirical modelling), as for example general strength of a team, recent results, performance on home games or away games, conceded and scored goals, historical results of certain fixtures, weather and injured players. Those parameters may not all be included, as that might complicate the model a lot.

Additionally, there are "live-bet" parameters for the actual events on the pitch that need to be considered, like the score cast, possible injuries, weather changes, bookings and substitutions and of course the time played. Some are constantly changing (e.g. time), some may be triggered manually by the user while the simulation of the game is running (e.g. score cast).

Between all those parameters there are certain dependencies existing, which should be defined and possibly manipulated in the model.

The process of using the model may start with the user (agent) adjusting the "pre-game" parameters. The odds are than calculated once. After that, the simulation of the game will start and the time will start running. So while the game is on, the user is able to trigger certain game events, like goals scored. After an event occurred, the odds are steadily recalculated contingent on the dependencies defined.

Finally, the agent can try to adjust dependencies on-the-fly, considering his own experience with the development of live market odds in real football games which might enable the agent to draw conclusions on how to weigh certain parameters within the dependencies in order to get as close as possible to the real markets and desirably find valuable bets.


Empirical Modelling website
Various websites about sport betting, mainly discussion forums 
as information about odds calculating is very rare
(basically due to the fact that there's a lot of money in it).

Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)

This is an extended abstract that gives some interesting context, but the actual statement of the aim of the paper in the last two sentences of the abstract is not very clear to me. The objective of "displaying a simulation of a football game" seems rather ambitious, and I'm not sure to what extent you mean to devote effort to this as well as the central model to assist live betting. So clarification of these two last two sentences is very important! And though I don't really understand what you have in mind here, I can see that this is where you are trying to refer to the relevance of EM. It is not a good idea to use the phrase "empirical model" though - clumsy as it is, I prefer the term EM model, which distinguishes what you have in mind from a product of empirical modelling in the general - and different - sense of that term. I think the idea of using EM in a live context is an excellent one because of the dynamic immediate quality of the modelling activity and the possibility - in principle, at any rate - of taking unprecedented events and considerations into account. The difficulty seems to be that it's hard to set up the empirical study without somehow having a resource to create the "live event", and for me your proposal: "while the game is on, the user is able to trigger certain game events, like goals scored" doesn't seem to have the appropriate contextual content. There's definitely a promising idea here for which you can get good credit without addressing all the technical issues in detail. What might be good here is to use previous data culled from real matches (such as "team X has not conceded a home goal in the last 20 minutes in the last 10 games") to inform your live assessment of a live event. For my money, cricket offers a much better prospect for analysis of this sort - for instance, it proves very easy to make an on-the-fly calculator of the required run rate in a one day match, but what is more difficult to take account of without proper preparation is knowledge about how fast the remaining batsman in a side normally score their runs and how economical the remaining bowlers normally are etc. There is potentially some interesting data relating to this theme from an MSc dissertation project I supervised in 2009-10, but of course learning the rules of cricket can be challenging!

Name: Jilong Qin
Library Card Number: 1161169
Provisional Title: Extension on Interactive Empirical Modeling of Aircraft Simulation


In this paper, I will improve the interactive simulation case designed with definitive notation and interactive principles of the Empirical Modeling. The DCS student of University of Warwick has developed a simulation model of aircraft by Empirical Modeling. So I need to add some new interesting functions to the model. In the computer science, building an accurate model and observing objects exist in the real world play a really important role in related research field. Empirical Modeling is a good way to build an experiential and exact interactive environment. With the Empirical Modeling, I can build the make simulation in the experiment, and use the experience to model the unknown iteration and situations. In the aircraft simulation, I will extend the more circumstances such as how the wind impacts on the aircraft, demonstrating the track of flight, how the aircraft turns Left/Right. These functions are lack in previous model. Based on the Empirical Modeling, constructing a real world situation becomes easier, the visual part of simulation is programmed in Scout and Donald, e.g. Aircraft Body, steering, meters, and the core functions will be implemented in the Eden, e.g. Record the tracks, count speed, detect changes of environment, departure, arrive. All interaction could be presented and constructed in the Empirical Modeling. Observables, dependencies, agencies will be emphasis in this model. In following paper, I will present observables, dependencies, agencies in my improved model based on the previous model which simulated the aircraft interaction and show how Empirical Modeling principles are used in the this model.

Model Description:

The previous model what I want to improve mainly describes the design of the interactive graphic environment with the EM. This empirical model simulates the simple behaviors of an aircraft. These behaviors represent the simple principles of flying for the aircraft, such as turn left, turn right, close gate when departure, landing. But the model only can display the how the wings change when aircraft want to go different direction. Furthermore, the aircraft is static graphic in the window. So the model is simple and I want to enhance. The first modification I do is to change the figure of aircraft so that the aircraft can really fly in the window, when we operate the aircraft to turn left/right, the head of aircraft do, unlike the previous model just display "Turn right/left" on screen. Secondly, after the aircraft moves, we define a weather situation, for example, windy, the wind has serious impact on the wings, the data which is impacted by wind will demonstrate on the screen. The last thing I want to extend is to make the airplane has a auto coordinate function so that the airplane can adapt to different kinds of weather. Above-mentioned functions are what I want to extend the existed model.


Interactive Empirical Model Design Simulation of Aircraft,
The Fourth Warwick Empirical Modelling Bulletin (WEB-EM-4)

Weighting: Paper - 40/Model - 60 (Paper not exceed 4 pages)

The term that has been used for EM models of this nature is Interactive Situation Model (ISM). (The term appears in several EM publications concerned with applications studied especially between 1995 and 2001 - see #069, #055, #052, CS-RR-352 and CS-RR0353. You should read some of these sources and include them in your references where appropriate. You may also find it useful to look at - and reference - other models that are somewhat related to the model you are proposing to build, such as cruisecontrolBridge1991) The model you propose to build on is indeed very simple, and is more limited than would be expected to get the 50% pass mark, so you need to make sure that you are being ambitious enough. You've focused here on introducing some more context for actions that shape the plane's behaviour - this is good, and doesn't need to be particularly "precise" so long as it helps the user to learn more about how the flight of a plane is controlled in general terms (cf. the discussion of Robert Schiller's proposal above). Introducing this elementary "educational" aspect is one way to extend the model. Other possibilities to consider are: introducing an EMPE context for the original model (cf. the construal comprehension exercise) so that you can add some commentary and pre-planned sequence of interactions to help the learner become familiar with the model. You might also consider how to port the model to JS-Eden - and discuss what difficulties that might present. Note that you don't have to make complete models in order to be able to support such discussion - though you will need to elaborate the existing model significantly to justify a 60:40 model:paper weighting. Your English style is quite clumsy and I would recommend you to get a native speaker to help you to suggest ways in which your abstract could be improved in that respect.

Name: Sharon Secondus
Library Card Number: 1164291
Provisional Title: Using Empirical Modelling to show solutions of Decision- making


Decision- making is something we come across everyday at work, home, almost any where, the act of us deciding whether to go home from work or stop by the grocery store is an act of decision - making. Decision- making is an environment in which the process of choosing using cognitive process to weigh options for the benefit of something, decision- making can apply to various domains, management, Finance, Computer systems and much more, the art of decision- making using technology became popular in the 90's companies such as SAP, Sugar Customer Relationship Management, Google have showed the success in these systems. Decision - making has always been presented through Software Development, some features and the facet of some decision making software is based around Observables, Dependencies and Agencies which envelope the Empirical modelling approach. It will be a different approach using the construal to show how decision can be made between two partner oil companies. One company handling the distribution of oil products while the other company will be in charge of distribution as well as going to the field to withdraw the raw material. Both companies have different range of customers from petroleum outlets, government, etc the scenario is created between the partner companies where the partner company has a field close by to a customer that needs a large amount of distribution, a model will be designed to control customer relationship management using a decision- making model to show data sharing to benefit customers, retain competition levels and gain user satisfaction. This model will be used by both companies to keep contact of their customers, their distribution levels, oil stock markets and interactions from department in charge of field results using the Empirical modelling approach of Observables, Dependencies and Agencies to create these interactions, this model can also aimed at learning to show software development to Empirical modelling in decision - making aspect.

Model Description:

The decision making model is to be used to show decision- making combined with perception using the Empirical Modelling approach, decision- making has been represented in software development and cloud computing. Using the construal to produce a model similar to the software development approach using the principles of Empirical modelling ( Observables, dependencies and agencies), decision making is an aspect all domains must have to plan towards sucess, applying decision making towards the oil and gas domain is what this model will be aiming at showing how interaction between two partner oil and gas companies can be reflected on a model. This model will also go to the extent of solving CRM (Customer Relationship Management) between two oil and gas partner companies, the model will be built using the EDEN notation, the EDEN construal will allow the key features of sharing data between the two partner companies for a customer who will require a large distribution which will require the partner company to make the distribution. Another key feature will be to added in terms of using the model to show live stock markets and other management tools, EDEN was chosen because of the way definitive scripts can be built then added gradually to the model, using the three concepts of definition, function and action, a spreadsheet will be used to regulate and keep the customer details, EDEN will show the interactions between the spreadsheet from one part of the model and also translated into charts as the user wants. The objectives of this model is to show the learning can be done from the view of decision making software to Empirical modelling, other aspects such as perception and constructionism of the transformation , the other objective is drawn from aim of using Empirical modelling to solve a Customer Relationship Management scenario using this model.


Hersh, M.A.1999.Sustainable decision making: the role of decision support systems
IEEE Systems. vol 29. pp395 - 408.

Hullermeier, E.2005.Experience-Based Decision Making: A Satisficing Decision Tree Approach. 
vol 35. pp641 - 653

Meurig Beynon and Antony Harfield. Constructionism through Construal by Computer.
Constructionism 2010, The American University of Paris, August 2010

S. Rasmequan and S. Russ.
Cognitive Artefacts for Decision Support.
Proceedings of the IEEE SMC 2000
"Cybernetics Evolving to Systems, Humans, Organizations and their complex interactions",
Tennessee, USA, October 8-11, 2000.

W.M. Beynon, S. Rasmequan and S. Russ.
The Use of Interactive Situation Models for the Development of Business Solutions.
Proceedings workshop on Perspective in Business Informatics Research (BIR-2000)
Rostock, Germany, March 31- April 1, 2000.

Weighting: Paper - 70/Model - 30 (Paper not exceed 7 pages)

The idea of using EM to address decision-support from a cognitive perspective is a good one, and you seem to have found some good EM references for this (EM paper 061 is another related one that may be worth consulting]. I find it hard to understand your proposal fully in part because of the way in which you construct your sentences! There are some very sensible phrases here. For instance the idea that your model will be "used to show decision- making combined with perception using the Empirical Modelling approach" is good, as is the concept of targetting Customer Relationship Management. What is less clear is why you choose to combine phrases relating to very different ideas in one sentence - as when you add this clause at the outset of your model description: "decision-making has been represented in software development and cloud computing". A good way of making it easier for your readers would be break up your sentences into their individual pieces (for instance: the sentence "Another key feature .... as the user wants" could to my mind be broken down into 4 or 5 short sentences). From your weighting, I infer that you want to write more and are perhaps less confident about building a model - this is of course fine so long as you can express your insights and can do enough model-building to gain insight and illustrate key points. It is unclear whether you can find a way to do this without making a model that deserves more than 30% of the credit. You will need to identify quite a number of observables for instance, and ideally need to be able to show how you would implement live feeds etc. One possibility might be to put your focus initially on giving an LSD account of the relationship between the Oil and Gas companies, which will involve explicitly identifying the relevant observables and classifying these as state, oracle, derivate etc. The ISM concept (which appears in the title of one of your references - and is mentioned in connection with the proposals by Tie Chen and Jilong Qin above) is also relevant here. Please note that "Empirical Modelling" - unfortunately - needs its capitalisation, as it is otherwise not being properly distinguished from "empirical modelling".

Name: Philip Gibbs
Library Card Number: 0820850
Provisional Title: A study of Empirical Modelling to teach historical programming methods.


This paper will investigate how Empirical Modelling could be used as a way of teaching how historical computers were programmed, and to contrast this with existing Empirical Modelling teaching tools. Computer programming has changed over time with the evolution of computers, and to this day, many of the old programming methods are still in use - directly and indirectly. However, the very first mechanical and electronic computers used a very different, hands-on approach to programming, involving much interaction with the hardware itself. Programs would be fed into the computer using punch cards - physical cards that contained encoded information in the form of holes. Later computers would use similar mechanisms such as punched tape, or physical manipulation of the machine's own switches and cables. These machines are no longer readily produced or available, and as such it is not easy to have a practical investigation or experimentation of the methods that were used to program the very first computers. The paper will explore the feasibility of using Empirical Modelling to provide representations of the very first computers, to teach users historical programming methods and to allow them to experiment and learn from these models, without needing access to the hardware itself. It will also compare its effectiveness with existing Empirical Modelling teaching tools in the computer science field, such as basic programming and sorting algorithms.

Model Description:

The modelling study will be based on a simplified representation of the world's first computers, where programs were entered in the form of punch cards. Such computers include the Electronic Numerical Integrator And Computer (ENIAC), the Manchester Mark 1, and Charles Babbage's (theoretical) analytical engine. The study will focus on a teaching environment as to how these computers may have been programmed using punch cards and physical switches.


Constructivist Computer Science Education Reconstructed

Empirical Modelling as an Unconventional Approach to Software Development

Paradigms for Programming

A Computer-Based Environment for the Study of Relational Query Languages

Empirical Modelling for Educational Technology

Programming With Punched Cards (

bubblesortBeynon1998 (

heapsortBeynon1998 (

heapsortBeynon2008 (

Weighting: Paper - 70/Model - 30 (Paper not exceed 7 pages)

This is an ambitious theme which connects with several previous EM projects (both final year and WEB-EM), many of which have unfortunately not been preserved in a form that you can consult. It is definitely a good thing to consult Steve Russ in connection with this theme, as he used to teach related topics on the History of Computing module. You might also contact Martin Campbell-Kelly, who has written an EDSAC simulator either directly (as or via Steve. I notice that you don't cite either of the precedents for WEB-EM submissions in this sort of area: "An investigation into the Empirical Modelling of physical devices, in an educational context, illustrated with the Enigma machine example" and "Modelling Babbage's Difference Engine" (see WEB-EM-2 online). The challenge for your project seems to me to be deciding where to put your effort: for instance, if you realistically make simplified models of several early computers, the level of simplification may be such that there is nothing interesting that is being said about individual machines; if on the other hand you focus on modelling one machine in detail you will probably find that you're in danger of losing the big picture and getting ensnared in details. (Most previous efforts have ended up with models so incomplete that "learning from them" is not yet possible.) Perhaps there's a suggestion in the way you've written the abstract (and the weighting you propose) that you are looking to give a treatment that targets observables that are not at the machine level but directly relate to the programmer/user's experience - such a shift in perspective has sometimes proved to be a good way to exploit EM more effectively in what had previously proved to be a difficult area of application.

Name: Tie Chen
Library Card Number: 1164200
Provisional Title: Gas Station model


In our daily life, we can easily see the gas station in the road or next to the exit of supermarkets. For the drivers who are engaging in the transportation industry, it has become the second home for them. The gas stations not only help the drivers to solve the problem of fuel shortage, but also supply a short-term rest area for the car driver. I believe we all know the significance of gas station which does not need too much explanation. But for the working principle, except the drivers, a lot of passers may not know it. This paper focuses on the working principle of gas station and research about the dependencies between the objects of vehicles, tankers fuel trucks and payment. For example, we know that gas stations usually have two kinds of payment methods: by cash or card payment, we can through changing the fuel costs payment method to observe the changes in the other objects. In addition, I will develop an Empirical Modeling model to simulate gas station situation so as to help the people who can use my model to better understand the gas station working principle. In the other hand, through the process of developing this model; I can not only learn more about the Empirical Modeling but also improve my development ability.

Model Description:

The basic objective of this model is to display the working principle in the gas station. The main objects include: vehicles, tankers and fuel trucks. The work of vehicle is going into the gas station to add fuel, the tanker is the tool providing to pumping, and the truck is to provide fuel to the gas station. Assume all the vehicles' pumping time is the same; there are two kinds of fuels: petrol and diesel; there is only one kind of petrol: 95#; there are two kinds of fuel cost payment: by cash or by card; the card payment is faster than by cash. The vehicles' filler is on right hand side.

The process of the gas station: the vehicle goes into the gas station (assume the driver has known what kind of fuel he needs), select payment method firstly and then wait in the appropriate queue. After the vehicles which is in front of the queue has left, then the next vehicle moves to the pumping, after finishing pumping and making a payment, this car will leave the gas station, and so forth. If the tanker is locked, it means this tanker is unavailable, and then vehicles can not pump in this tanker. If the fuel trucks are working, all the tankers will be locked; until the fuel trucks work is finished. If time permit, I will cancel more restrictions in the gas station and make the model more similar to reality.


Beynon. W.M, Norris. M.T., Slade. M.D.,
Definitions for modelling and simulating concurrent systems, 
Proc. IASTED conference ASM'88, Acta Press, 1988.
M. Beynon Radical Empiricism, Empirical Modelling and the nature of knowing (2003)
Empirical Modelling in the warwick websit
Meurig Beynon. Definitive notations for interaction.
Meurig Beynon. Empirical modelling for educational technology.
Meurig Beynon & Chris Roe. 
Empirical modeling principles to support learning in a cultural context.
1st International Conference on Educational Technology in Cultural context, , 2002.

Weighting: Paper - 40/Model - 60 (Paper not exceed 4 pages)

For this project, I think it would indeed be a good idea to revisit the kind of modelling of concurrent systems (based on LSD and the ADM) that is discussed in the first reference you cite (the other references you cite seem a little less focussed as far as your specific theme is concerned). The advantage of your chosen scenario is that it can illustrate the distinction between a standard simulation (e.g. where there is an abstract queue of vehicles proceeding in an orderly manner in carrying out a rigid process) and a simulation that recognises the impact of real-world observables (as when cars dispose themselves on the forecourt in such a way that only some can get access to any pump for instance). As I've suggested to Jilong Qin (see comments and references above) a useful concept to be aware of is that of the Interactive Situation Model (ISM). A very important principle to apply in building up your model is to start small - e.g. introducing the process of filling up with reference to a single vehicle and petrol pump for instance, and then gradually embellishing your model to consider matters such as the location of the filler pipe etc and adding more vehicles, pumps and contextual elements (such as the garage layout). The description of the model that you give shows the influence of traditional process/algorithm oriented thinking: you are being very prescriptive about the procedure to be followed, which suits a certain kind of simulation, but doesn't match well with the idea that EM is first and foremost about modelling state and the initially quite open-ended potential actions of agents. It is not in the spirit of EM to make restrictive assumptions such as you have above "all the vehicles' pumping time is the same", "The vehicles' filler is on right hand side" etc - these are the kind of simplifications that enable traditional modelling of behaviours (which can otherwise get too complicated to grasp) but are not natural features of the garage state or of drivers in general (all of whom are different, and some of whom will be unlikely to follow a set sequence of actions in a robot-like fashion). There is credit to be had by drawing up a good LSD account of the agents in the scenario - you might use the railway station animation discussed in the lectures, or Neil Turner's five-a-side football animation ( as a source of ideas about how to draw this up. Unfortunately, there is no easy way to develop an ADM simulation from such an account at present, so the role of LSD will be primarily as documentation.

Name: Jona Gjini
Library Card Number: 1063531
Provisional Title: A study that shows the use of Empirical Modelling in Education


Throughout the years some research has been conducted on how Empirical Modelling (EM) is a new way of developing software for educational purposes. One thing that is going to be examined throughout this paper is some aspects of how EM can help in the process of learning how to use certain things that are related to education. In addition to all that another thing that is going to be looked at in this paper is how the development of an Abacus, which is based on EM, it is quite useful for educational purposes.

Model Description:

In order to prove that Empirical Modelling is a good approach when it comes to education a study on how users’ can learn to use well an Abacus that is developed in EDEN it is going to be presented. The basic idea of how an Abacus will be developed it is going to be given in the next few lines. First of all, there is going to be a graphical interface that represents an Abacus, with nine columns and ten rows. The rows will be divided by a line that will separate them in two parts, in the first part there are going to be three rows and in the second part seven rows. Furthermore, in every column there are going to be seven beans, two in the first part and five in the second one. Every bean will represent a separate observable, which will be connected with the other observables through dependencies. This way it is easier to calculate the number that the user wants. In addition to all that there is going to be a text box that shows every time what number is represented by the beans in the Abacus, so the user will know whether it is the number that wants or not. By repeating this process over and over again, the user will start learning how to use an Abacus which is the outcome that this study is trying to accomplish.


W.M.Beynon.(2206).Empirical Modelling for Educational Technology. 
Department of Computer Science. University of Warwick. 
Meurig Beynon. (2009). Constructivist computer science education reconstructed.
ITALICS Volume 8 Issue 2.
Computing technology for learning - in need of a radical new conception. 
Journal of Educational Technology & Society.(2009)

Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)

I like this topic, and I think that it has potential to be an iconic EM model (consider how SBR used a real abacus in his demonstration of "human computing"). To my mind, your title and abstract would benefit from going straight to the point, and put teaching the use of the abacus as a central concern. You should also refer to SBR's interpretation of the abacus - not least because it makes simulating the use of an abacus much more interesting than e.g. "simulating the operation of a clock mechanism" - in which human interaction and interpretation isn't directly and essentially involved. I appreciate that you may think it odd to put the focus on modelling an obsolete technology in this context, but this is actually saying something much tangible about EM than talking about "the use of EM in education" and "some aspects of how EM can help ... how to use certain things that are related to education", though it's quite OK for your paper to include some general discussion of this nature (preferably motivated by the study of the abacus). Your model description is promising - I especially like the fact that you don't put the emphasis on the mechanics of doing sums with the abacus in the first instance, but on the various states in which an abacus can be configured and how these can be interpreted. In the spirit of EM it would also be good in this context to highlight the states that can't be interpreted, and the interpretable states that are "equivalent" in the sense that they represent the same number - and how you transform from one state to another equivalent state, since this is at the heart of learning to use the abacus. And don't overlook the fact that helping the user "to learn how to use an Abacus" builds on something richer than that, which is "understanding how the abacus works". (Part of the point of the abacus is that you can learn mechanically how to use an abacus without understanding anything about arithmetic operations apart from how to present inputs and read outputs.) The heapsort construal ("getting a pigeon to make a heap by clicking on the red nodes") is a useful point of comparison here, and you might also find it tangentially instructive to look at other models of instruments with a similar "human computing" pedigree like the planimeter for measuring area (cf. the video clip at the "Tour" tag for planimeterCare2005).

Name: Michael Lau
Library Card Number: 1161118
Provisional Title: Using EM to teach the quick sort algorithm


This paper aims to assess if Empirical Modelling can be used to help educate in teaching the principles of algorithms such as a quick sort algorithm.

The quick sort algorithm is a divide and conquer algorithm in which the array of elements is partitioned by selecting an element which would become the pivot. All elements smaller than the pivot element, are moved before the pivot element in the array and all elements larger than the pivot are moved after. It is an algorithm that is one of the most popular sorting algorithms and therefore is one of the algorithms which are taught to new computer science students. For this reason it is important that the students understand how the quick sort algorithm works and have the correct understanding. In ensuring that the algorithm is understood a practical example can prove to be beneficial.

In this paper it is proposed that the use of Empirical Modelling tools to generate software to assist in educating and teaching the behaviour of the quick sort algorithm, through the creation of the model of the quick sort algorithm.

Model Description:

The aim of the model was to teach the users the quick sort algorithm by giving them a visual representation. The main focus of the model is the interaction and usability that the users have with it. To improve the chance of understanding, the model takes the user step by step through the series of actions that the algorithm takes upon an array of elements. It was decided that the ability to add and remove elements to the array was not a high priority and thus this feature is not apparent in the model.


1.	Greedy Algorithms and the Making Change problem 
2. Can Empirical Modelling Facilitate Learning?:
Modelling Graphs to Assist in the Teaching of AI Search Algorithms 
4. W.M.Beynon. Empirical Modelling for Educational Technology 
5. J. Myers - The color of learning: enhance memory and retention with a splash of color 

Weighting: Paper - 40/Model - 60 (Paper not exceed 4 pages)

This is a potentially interesting case-study, but one which may not be as straightforward as it might appear. It is surprising that you have not referred to the heapsort construal discussed in the lectures, as a successful model would ideally need to analyse quicksort in observational terms in a similar way. (The bubblesort construal used in the comprehension exercise is also relevant.) But whereas heapsort is a process that can be explained quite clearly by illustrating the key observables and action with very few elements, it's not so obvious that quicksort can be explained in this way. (For instance, the choice of a random element is best understood when there are a significant elements to select from.) Your model description doesn't seem to go far beyond "giving the user a visual representation" and "[taking them] step by step through the series of actions that the algorithm takes" which is not enough to engage fully with the idea of modelling the observables and dependencies underlyingthe conception of quick sort. (In fact, merely exploiting dependency to link visual elements on the screen to the values internal observables in a model is a fairly standard pitfall in "applying EM principles" - such a use of dependency does not usually do full justice to the observables and dependencies that need to be taken into account.) So for instance in quick sort one principle is that if you choose an element at random, the likelihood of "about half" the elements being bigger and "about half" being smaller is high, and you can imagine this being a good principle to illustrate with an interactive EM model that can then be elaborated towards a quicksort model. As I've suggested elsewhere in discussing the modelling process (cf. Tie Chen's proposal above), it is not appropriate in general to be specifying restrictions from the outset on what kind of interaction is possible - I don't understand why "It was decided that the ability to add and remove elements to the array was not a high priority and thus this feature is not apparent in the model" for instance. From the point of view of supporting the process of understanding the algorithm it may be very helpful to be able to consider the implications of discarding/adding elements. (The relevance of the J Myers article isn't clear.)

Name: James Michael
Library Card Number: 0801583
Provisional Title: A model of the Welsh sentence for learning to construct sentences.


This project will look at creating a construal that could be used in an educational environment to teach or demonstrate to new learners of the Welsh language about the structure of a sentence and how to correctly form sentences in the language.

Learning a new language can be difficult. One area that learners often find particularly difficult is in constructing sentences. The learner will have an idea of a concept they wish to convey in the language, but can have trouble understanding the intricacies of how the words fit together to form a valid sentence. When a learner forms a sentence with this understanding of the language, words can get muddled up, transposed, ignored or used incorrectly, resulting in incorrect sentences. A tool which could allow a user to interactively build up a sentence through experimentation using the individual building blocks of the language---such as tense, subject, or individual words---could enable the learner to gain a much more thorough understanding of how the elements of the language interact.

When a learner has a misconception of a concept, they can get confused when they see sentences or try to construct sentences. By constructing sentences and receiving feedback they are able to fix their misconceptions. A tool could help the learner to see how valid sentences are constructed, an allow the learner to resolve their misconceptions.

It can also be difficult for a learner to accurately convey the problems or misconceptions they have with a language, or for a teacher to correctly understand why the student is having problems. A tool could be used as an aid to help explain the problem a learner is having.

This project will use the empirical modelling toolset to create a model of a sentence in the Welsh language. This model can then be used to construct sentences, allowing a learner to actively participate in the construction of the sentence. The resulting construal will allow the learner to experiment with the basic structure of the sentence in order to boost their understanding and confidence with the language. The construal could also be used in a classroom environment to demonstrate the concepts discussed in lessons.

Model Description:

The modelling study will involve constructing a construal which represents an understanding of a Welsh sentence. Through interactions by a user, the construal will allow sentences to be constructed for the purpose of gaining a stronger understanding of the language or demonstrating basic concepts of the language. In addition to the construal, the study will look at the applicability of empirical modelling and the empirical modelling tool set to assist in solving problems like this by creating construals that allow experimentation and construction.

Given time limitations and scope of this project, a small yet representable subset of the Welsh language will be implemented. This will include a few dozen of the most common words in the language, such as nouns and verbs, together with some concepts which can be used to modify the sentence, such as tense or time.

The key observable of the construal that will be built will be a sentence that is constructed based on a users interactions with the model and displayed on screen to the user. Additional observables might conceivably include an illustration of the relationship between elements in the sentence, or an English language translation of the constructed sentence.

The dependencies will model the Welsh sentence, and allow a sentence to be constructed from the model based on the input from the user.

The main agent of the system will be a learner of the Welsh language, who will use the construal to construct sentences in order to improve their understanding of the language. An example interaction that such an agent might have could include the agent clicking on images representing the nouns or actions that form a sentence, which will influence the constructed sentence.

Other agents that might use the construal could include teachers in a classroom scenario, interactively demonstrating the concepts discussed in the lesson.


Empirical Modelling for Educational Technology
Constructivist Computer Science Education Reconstructed
Computing technology for learning - in need of a radical new conception

Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)

Dwyf fi ddim in gallu siarad Cymraeg yn dda, ond dwyf fi yn wybod fod hon yn astudiaeth ddiddorol iawn. It is interesting to note that - despite the fact that grammatical constructions in natural language represent the context in which the idea of 'construal' is most familiar - no previous EM study has addressed this topic. There are some examples of indirect references to language construction in EM literature though - it may be of interest to look at numbersRun-bol2002 which was developed as an illustrative example in conjunction with Rungrattanaubol's PhD thesis (see Chapter 2 p.65-68) and considers the way in which numbers are expressed in Thai; at "Variation 9" as displayed in the JUGS Theme and Variations poster at kaleidoscopeBeynon2005/posters/JUGSPoster.pdf (the source of which is in the kaleidoscopeBeynon2005/jugsVariations/ directory); and to reflect on some of the thinking around the agent-oriented parser (see the Docs directory in agentparserHarfield2003). One of the key challenges in learning Welsh is to handle mutations correctly and to appreciate the role of noun gender in this connection - also to recognise the primary forms of mutated words: it would be probably be helpful to indicate where such issues stand in relation to your proposal - something that is not at present entirely clear from your abstract. (Handling mutation in Welsh in fact looks like an excellent and rich example of dependency in action.) Please note that "Empirical Modelling" - unfortunately - needs its capitalisation, as it is otherwise not being properly distinguished from "empirical modelling".

Name: Robert Steele
Library Card Number: 0821769
Provisional Title: An Empirical Modelling Approach to Text Processing


Text processing remains on of the most widespread uses of computers in both professional and personal work. Software used ranges from the complex but general use Microsoft Word to software simpler, but perhaps sometimes unintuitive to the beginner, such as vi.

In the paper, I hope to investigate text processing from an Empirical Modelling point of view, considering methods to make working on text documents more intuitive for the user as well as considering the form of software that would assist in the development of further models to illustrate the uses of Empirical Modelling.

In the use of conventional text editing software, users can desire dependency linking but lack appropriate tools, leading to time consuming 'copy paste' behaviour or the requirement for extensive editing. The user may also use information in a variety of ways, wanting to integrate notes or unrefined content into the text editor without committing it to the final document. The paper will look at this from an Empirical Modelling point a view, considering how the model of the completed document is built by the user and ways to aid in this process.

The paper will also consider this problem in relation to specific uses of a general text processor, such as writing a report or paper, creating a letter to multiple recipients and software development. In particular, the paper will look at the programming and development involved in creating models in an Empirical Modelling context.

By understanding the relationship between the user and the software using Empirical Modelling, we can look at how to tailor the software to engage more with the user and their ultimate intentions.

I hope that this approach will also assist development of programs for EDEN and future Empirical Modelling projects, helping to generally define observables sharing large amounts of content without using a strictly object based structure; which may be unsuitable for Empirical Modelling projects.

Model Description:

Model description: The model will be a simple text editor that takes advantage of Empirical Modelling principles. This will hopefully allow the user to have greater control over the categorization of text and create word or term dependencies as required.

To illustrate the categorization of text, consider the rough outline of the report or program written in note form against the words and paragraphs that form the final product. Ideally the program will let the user have these at hand without using a separate text processor or written notes.

For the dependency structure, consider a letter or computer program. In a letter, the user may wish to write a general letter, perhaps for a mass mail like function, and refer to a recipient's name several times throughout the letter. Hopefully the model will provide functionally to allow the user to then use this information to produce the individual letters in a convenient form.

In the development of software, the user can often find themselves replicating large amounts of code with very minor changes. This can especially be a problem in an Empirical Modelling environment, where it is desirable to define observables distinctly. By using dependencies in this situation, creating models can hopefully be made less daunting for the user.

The model will be developed in EDEN, focusing on letting the user write programs and models in an environment that embodies more of the principles central to Empirical Modelling at the development stage than the traditional programming approach. The model will aim to let the user more easily develop and extend models for Empirical Modelling in a way which will be accessible to both people familiar with Empirical Modelling and people new to the principles and methodology of the field.

The model will focus on making the usage of the text processor as intuitive as possible, providing facilities for users used to different commands for standard functions. For example, deleting the last word or line.


Various EM papers on the interaction between the user and the computer
 and the user's understanding.
Papers concerned with the Subtext programming language,
which incorporates some of the ideas I hope to look at from the viewpoint of EM.

Weighting: Paper - 60/Model - 40 (Paper not exceed 6 pages)

The role of text in EM is the focus for much discussion. The central idea of using artefacts to convey understanding and assist communication in some sense undermines the central role claimed for language in some philosophical traditions (cf. the Experiential Framework for Learning). In practice, the whole culture of "modelling with definitive scripts" has given great emphasis to text in EM, and the way in which definitive notations are rooted in algebra also promotes symbolic aspects. Though Cadence aspires to marginalise text - or even dispense with text entirely - and instead exploit inspection and manipulation of explicitly displayed structural relationships, text still has an important place in current EM prototype tools. What would be most effective is some kind of hybrid technology that combines direct manipulation of structure with text processing. Examples of tools that work in this spirit include the dynamic geometry package GeoGebra and Karl King's drawscout (drawscoutKing2004). Whilst for the time being scripts that can be stored in files are the most effective ways of introducing persistence to models, the creation of subtle scripts should ideally involve something much less intimidating than generating reams of text using editors or macros. It might be useful for instance to have ways to select a word "XYZ" in a document, and automatically generate a set of distinct observables of the form a1, a2, a3, ... to correspond to all instance of "XYZ" in the document. By default, it might be that (e.g.) a1 = "XYZ", a2 is a1, a3 is a1, ... etc so that dependencies between substrings are established, and the text can be subsequently transformed by giving new definitions to a1, a2 etc. Other potential sources of ideas are texteditorYung1987 - the application for which EDEN was originally created, and the AOP parser agentparserHarfield2003. As a case-study in scaling up an EDEN model, you may find it useful to consult Jaratsri Rungrattanaubol's extensions to the original linesBeynon1991 model in the EM projects archive (see e.g. linesBeynon1991/jaratsriextensions/codelines.html).

Name: Anees Ahee
Library Card Number: 0810853
Provisional Title: An interactive model of the game Go for pedagogic use


I intend to use Cadence to construct an interactive model of the ancient board game Go (also known as 'weiqi' in Chinese) which could be used for teaching purposes.

This game can seem quite intimidating for beginners, thus I want to explore methods of visualising certain abstract concepts (such as 'influence') in the game by altering the graphics of the board and stones. This will help new players understand these concepts. Empirical Modelling provides me with a way of modelling these things which are vague and intuitive, in an exploratory manner.

Model Description:

The model will consist of a go board of standard size (either 9x9 or 19x19) which is just a grid. Stones will be placed on the intersections of lines on the grid and these can be represented by black and white circles. A user will be able to click on the board to place stones on the board in valid places.

The model will not allow illegal moves, for example placing 2 stones on top of each other at the same point on the grid or placing a stone on inside a square on the grid. It will automatically remove captured stones from the board. Each point on the board as well as each stone can change colour depending on its state.


Albert Lindsey Zobrist. 1970. Feature Extraction and Representation for Pattern Recognition
and the Game of Go. Ph.D. Dissertation. The University of Wisconsin - Madison

Cem Ozturan. 2011. Chess Mentor: a model of chess developed with Empirical Modelling concepts. URL:

Sensei's Library:
This is a reputable wiki written by the Go community

Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)

This modelling exercise seems to be focused on modelling a specific kind of dependency - one that relates influence to the position of the stones on the Go board. There is a danger here of interpreting the modelling of one (albeit very interesting) dependency relation as an exemplary illustration of EM (cf. the discussion of Zhetong Fang's proposal above). By using Cadence, you will eliminate the danger of giving a procedural specification of the influence function that can be used on the RHS of a single dependency relation in a somewhat pathological script - this is not an option in Cadence. You should be able to use cloning to model the board. The challenge will then be to express the relationships between different stones using dependency relations framed in Cadence. The representation of the board in Cadence may well preclude "putting two stones on the same square", but beware of prescribing other aspects of the management of the state of the board in a premature fashion (cf. my comments on Tie Chen's gas station model). To implement the behaviours you specify (e.g. to process user clicks and to automaticaly remove squares from the board) you may also need to exploit the simple agent concept that Cadence affords (for details see the "Cadence resources" tab in the LH panel associated with the CS405 webpage for 2010-11). It will be important to explore the feasibility of this application of Cadence on a very small scale initially, so that you can identify any possible problems. Ozturan's chess program is unlikely to ne a useful reference unless you after all opt to use Eden / JS-Eden instead. (A comparative study of different potential ways to implement a model of influence would also be an acceptable submission.)

Name: Isaac Lewis
Library Card Number: 0728164
Provisional Title: Exploring the Financephalograph: an analog simulation of the economy


The Financephalograph, also known as MONIAC, was an innovative analogue computer built in 1949 and designed to simulate the workings of the national economy of the United Kingdom. It used the flow of water around various tanks and pipes to simulate the flow of money between different economic sectors. A complex system of valves and other mechanical devices were used to simulate the complex dependencies between different economic variables.

The initial motivation was to create a teaching aid; students could experiment with tweaking economic parameters such as interest or tax rates to see the effect on the simulated economy and gain a conceptual understanding of the interrelationships of the real world economy. However, the Financephalograph also proved to be useful as an economic simulator in its own right, with machines being purchased by several governmental and corporate customers.

This project will look at recreating the Financephalograph in EDEN. The main focus will be on recreating the behaviour of the original device, but if time permits the increased complexity of the modern economic system will also be modelled in-part. The project should therefore provide an interesting tool for students wishing to learn more about how the modern economy functions, and also be an exploration into an analogue-based model of computing that is very different to the standard, digital paradigm.

Model Description:

The model will be based largely on the original Financephalograph. This used tanks to simulate the different aspects of the economy; for example, a large tank represented treasury income, which flowed into smaller tanks representing healthcare, education and so on, from there flowing into tanks representing consumer savings and investments, and eventually flowing back to the treasury as tax. These will be represented in my model by simple graphic depictions of such tanks, with the height of water of each tank represented by a floating point variable. This means that the simulation will not be truly analogue (impossible on a digital machine), but will be close enough for the purposes of the model.

Pipes allowed water to flow between the tanks, and taps could be used to control the rate of flow, for example simulating the movement of money from wages to government income via taxation. These can be represented by movable sliders in my model. The Financephalograph had to use some clever mechanical tricks to simulate more complex relationships between variables; simulating such relationships in software will be more straightforward.

If time permits, adding functionality to simulate changes in the economy (and our economic understanding) in the last half-century can also be added. For example, globalisation means the modern British economy is more influenced by international trends. The high performance of modern computers has also enabled the multi-trillion dollar market in complex financial instruments such as derivatives, which in recent years has had a huge impact on the real economy. Both factors would be interesting to simulate.


Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)

This is an interesting topic, and one that relates to a long-established tradition in applications of EM: modelling historical instruments and computing devices (cf. Philip Gibbs's and Jona Gjini's proposals above, and the resources cited in my comments). In using EDEN for this modelling, you will need to consider modelling of dynamic quantities of the sort that have been addressed in models like cruisecontrolBridge1991. An important element to be addressed here is feedback (such as is represented in the cruise control by the way in which the forces on the vehicle influence the speed which directly affect the forces on the vehicle by changing the wind resistance etc). This kind of process-like relationship is more naturally modelled in Cadence, and it may be worth making some reference to the comparative merits of these tools for implementation. One of the problems with electing to model a specific historical referent is that a full-scale realisation may prove to be beyond the scope of what is feasible in the WEB-EM context. On that basis, it would probably be a good idea to focus initially on a feasibility / methodology study where you consider to what extent - and if so how - you can make a viable model. Whether you then succeed in "completing" the model you can then achieve a good overall result for your submission by making simpler construals that establish proof-of-concept (or expose problematic issues). Giving a good account of the Financephalograph in terms of observables, dependency and agency would also be a valuable contribution, and you might find it useful to make construals that help someone without financial expertise to understand the concept behind the Financephalograph.

Name: Aastha Kakaria
Library Card Number: 1156267
Provisional Title: Mapping Mob Behaviour-With Emphasis On Football Riots


Football hooliganism, sometimes referred to by the British media as the English Disease, is characterised by unruly and destructive behaviour - such as brawls, vandalism and intimidation - by association football club fans. Fights between supporters of rival teams may take place before or after football matches at pre-arranged locations away from stadia, in order to avoid arrests by the police, or they can erupt spontaneously at the stadium or in the surrounding streets. Football hooliganism can range from shouts and small-scale fistfights and disturbances to huge riots where firms attack each other with deadly weapons such as sports bats, bottles, rocks, knives and pistols. (definition from The variety in the violence shown by football mobs is interesting to note as it's root cause is almost always consistent: the result of a football game. What I hope to explore is the effect factors such as security, popularity of the team being supported, the severity of the loss have on the level of violence. Wherever possible, I shall also look to include social background of the rioters.

Model Description:

The model I have in mind shall represent the mob as a group of black circles and the factors that might incite a riot as a row of coloured buttons. Depending on whether the factor being inputed has a positive or negative effect on a riot, the mob should glow more or less crimson. If the dots should cross a certain level of "redness", they disperse across the screen in imitation of a mob.


 Mob Psychology for Dummies amongst other books on mob behaviour.

Weighting: Paper - 60/Model - 40 (Paper not exceed 6 pages)

This topic is certainly interesting, but it may not be straightforward to apply EM concepts and tools. It's noticeable that you haven't made explicit reference to observables, dependencies and agency in your abstract, and it's not clear to what extent the detail you give (e.g. about "deadly weapons" can be made relevant to your study). Though it's important that you do emphasise the visual experiential nature of the model you propose, the exact colour conventions is not so important as to be the key element of your model description. A thorny question is whether there can be an appropriate way "to explore is the effect factors such as security, popularity of the team being supported, the severity of the loss have on the level of violence" within the scope of this modest coursework exercise - obviously, to address this question seriously, it would be necessary to study the histories of many instances of hooliganism and attempt to infer dependencies (which may or may not exist?). An appropriate way to approach this theme would be to venture LSD accounts for the behaviour of agents in a mob context, and animate these in the spirit of the Abstract Definitive Machine (ADM), as discussed and illustrated to some extent in lectures in Weeks 9 and 10. As the Ant Navigation model shows, getting interesting behaviours is a complex process, and the current EM tools don't give you as much support for multi-agent modelling as we'd like. If you find it difficult to model many agents, it might be worth considering some modest prototyping to support a comparison between an EM approach and other agent-based system approaches as an alternative theme. It may well be of interest to look at the work of Prof Li Tsai-Yen in NCCU Taiwan (see - particularly Virtual Crowd Simulation) - he heard my presentation on Ant Navigation in December 2009, and took the view that there was some significant connection between his research and EM (which again might be a good topic to explore).

Name: Lucy M H Davies
Library Card Number: 0802331
Provisional Title: Using Empirical Modeling to teach the basics of digital circuits


The purpose of this submission is create a tool to interactively teach the basics, and some advanced concepts, of digital circuitry. The paper will compare this tool to existing digital circuitry tools and to EM tools for learning in other domains.

Model Description:

The model is anticipated to be in the form of a series of EMPE presentations. These will cover topics such as basic logic operators (AND, OR, NOT, NAND, NOR, XOR and XNOR), truth tables, logic gates, K-Maps, and more complex components including adders, multiplexers, demultiplexers, encoders, decoders, latches, flip-flops, and counters.

Models within the presentations would include conversions between truth tables and circuits, conversion between logic expressions and their corresponding K-Maps, and simulations of individual logic gates and complex components, with the intention of allowing custom complex components to be defined in terms of other basic or complex components.

CS132 Lecture Notes 
Constructivist Computer Science Education Reconstructed
Empirical Modelling for Educational Technology
Weighting: 70:30 - Model:Paper

This is a topic that has attracted interest in the past - perhaps because of the influence of simulation tools used in the CS132 module on Computer Organisation and Architecture. The most well-developed attempt to address this theme was the "ELS ­ An Eden Based Digital Logic Simulator" submission #24 for WEB-EM-3. This submission is particularly interesting because of the strong criticisms it makes of EDEN as a tool for software development, and highlights the challenges that you may face where lack of object-orientation and issues concerned with modelling feedback with undirectional dependencies etc are concerned (the model-builder was an excellent programmer who had experience of building such a simulator with conventional tools). This suggests that you are right to focus on using the EMPE in a pedagogical role, and not attempt too much by way of complex circuit modelling. You may be able to borrow visual components from the ELS - these were very elegantly constructed, and it may be difficult to improve on them! It would be good to see whether you can find better ways to model the behaviour of components though. There are good precedents for the EMPE in educational applications - notably graphicspresHarfield2007 for instance. If you find that you need another angle on your paper, you might also consider what advantages Cadence might bring to the modelling study - cf. calculatorEvans2011 from the Virtual Showcase in Lab 6, and the discussion of object-like and process-like observables that Cadence affords (cf. Lab 1 and related lectures).