Skip to main content Skip to navigation

Abstracts for WEB-EM 03

Submission time: 30/11/2006 10:46
Name: Nicolas Pope
Library Card Number: 0322756
Provisional Title: A pure definitive language for Empirical Modelling

Abstract: The objective of this paper is to evaluate a new purely definitive language as a possible Empirical Modelling tool. This will involve comparing and contrasting the features of this language with the features of existing EM models and the EM principles on which they are based. A brief introduction to the language itself will also be included to show how to create basic models as well as the underlying concepts in the language. There are a significant number of differences revolving primarily around the use of recursive definitions and how this relates to agents and procedural aspects in existing tools which will be discussed in depth. The language itself has already been designed as part of a third year project, however there are various issues remaining that this paper will attempt to resolve to produce a cleaner and more useful system. Most of these issues relate to how the language is evaluated, specifically to do with time, concurrency, atomicity and so forth. This is where the comparison to EM tools and principles will be useful. To go about making this comparison it will be necessary to construct a program to run this language and then construct a few simple models in it that have already been done in tkeden. In addition to this a more comprehensive model is being constructed by fellow students using this new language to further explore its capabilities and usefulness to Empirical Modelling. Finally, the potential applications and future directions for this new language will be outlined and discussed as well as its other potential benefits or disadvantages, including distribution, file system replacement and integration into an operating system. Hopefully at the end of the paper it will be possible to say one way or the other if this language is useful for Empirical Modelling. Just to clarify, the user interface to the tool and the modelling environment itself will not be discussed in this paper, other than to say that wrapping an IDE around this system would not be a great technical challenge but a HCI one.

Model description: The modelling aspect of this project will involve recreating some existing models in the new language which itself will involve creating a tool to run the language. How many models are converted and how complete they will be depends a little on time available, however at least jugsBeynon1988 and roomYung1989 will be done along with some other models to show how the system could be used outside of Empirical Modelling, as a file system for example. Possible models include:


A spreadsheet, similar to spreadsheetRoe2002


As mentioned in the above abstract another model is being constructed in parallel by fellow students, this may prove useful in my work also. The primary modelling work involved in this project will be the creation of the tool itself, this presents many complex issues and requires significant time but does introduce me to the issues relating to Empirical Modelling. At the time of writing this proposal a significant amount of this tool has been written but much still needs doing. The remaining issues to be resolved for this project as of the 30th November are:

  • Order of evaluation is currently incorrect and sometimes results in out-of-date values being used.
  • Textures, custom models and display lists in the opengl extension.
  • 2D shapes and line drawing, although windows could be used
  • Ability to get the key object so that definitions can be general purpose/relative.
  • Optimisation of number objects and garbage collection
  • Array index optimisation for large quantities of data.


References: D.K.Gehring. Spreadsheets and programming. Proceedings of
the First Psychology of Programming Interest Group Postgraduate Student
Workshop, 1996.

D.K.Gehring et al. Higher-order constructs for interactive graphics in a definitive programming framework. Proc. 14th Eurographics UK Conference, 179-192. 1996.

W.M.Beynon. Definitive notations for interaction. Proc. HCI'85, ed Johnson and Cook, Cambridge University Press, 23034, 1985.

A. Borning, R Duisberg. Constraint-Based Tools for Building User Interfaces. ACM Transactions on Graphics Vol 5. No. 4, 345-374. 1986

M. Garrett. J. Foley. Graphics Programming Using a Database System with Dependency Declarations. ACM Transactions on Graphics Vol 1. No. 2, 109-128. 1982.
Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)
Uni ID no: 0322756
usercode: csudek
user name: Nicolas Pope
user email:

Submission time: 30/11/2006 10:59
Name: Nicholas Lee
Library Card Number: 0315420
Provisional Title: Definitive Logic Simulation

Abstract: This paper will explore the possibilities and problems associated with modelling digital logic circuits in a definitive language. This paper was inspired by the authors work on Open Logic Sim 2 (OLSv2) - an open source digital logic simulator written in Java and previously used in the "Computer Organisation and Architecture" module of the 1st year computer science degree at The University Of Warwick. The author made many improvements to OLSv1.36 in his second and thrid years so has a broad understanding of OLS's implementation.

The paper will attempt to discover whether a circuit can be modelled with a definitive notation e.g. it is hoped that wires in a circuit can be modelled as dependencies in Eden - the output from the wire (and therefore input to the next component) is the output of the previous component. However, modelling the wires simply as depencies could lead to problems with modelling feedback - affecting the input of a component based on its output. One of the simplest 'interesting' digital circuits is the flip flop - used for storing information and constructed with pairs of NAnd gates that feedback to each other. However, due to the use of dependencies, this would probably be a difficult (if not impossible) feature for a definitive language to have since each time the output is set it would trigger another setting of the output leading to an infinite loop (this will be discussed in the paper). It is possible the feedback problem may be overcome by using discrete timing steps. However, this could involve not utilising the advantages of empirical modelling dependencies and moving back towards the threads & buffers implementation used in OLS. The methods used by the model (and possibly the efficiency of a dependency based simulation) will then be compared and contrasted with the methods used in the OLS solution.

Model description: The model would attempt to visualise the circuit in the same way that OLS does. That is, drawing the components and wires on screen, coloured based on what their current state is. After the main model has been written, there are several directions which could be taken to extend functionality and/or improve ease of use. A simple graphical interface could be designed and implemented for educational use. A pair-oriented learning feature could be implemented using distributed TKEden. One student would create a circuit (using the interface talk about previously) which the second student would be watching on another terminal. When the first student was finished, the second would 'break' some of the components (which the first student would not be able to view) and it would then be the first students task to 'debug' the circuit to find the broken chips. Broken chips could take several forms - outputting random data, constantly on or off, inverting the output, ignoring one of the inputs etc. When the circuit is being debugged, the wire colouring may have to be turned off (since this may make it too obvious where the defect lies) in favour of using a 'logic probe' so only one value can be observed at a time. It may be possible to adapt to the model to use analog logic rather than digital - that is, each input, output and wire would have a voltage rather than a logic level. This could then be used to investigate effects of noise. However, this may lead to the necessity for using (possibly) complicated electronic principles for instance resistance and the voltage drop in a wire - the longer the wire, the more voltage is lost therefore affecting the effects of noise. Seperate (and possibly preceding) the above mentioned possibilities, is the creation of a new notation. The notation would be used for the creation of circuits and would be written in AOP. If a new notation were to be created it would probably be created before the interface was programmed so the interface could use the new notation to perform its operations. (OLS : I have tried searching the DCS website for it but it seems it is no longer used in COA. If you would like a copy of OLS (v2 (jar and/or source) and/or v1.36 (jar and/or possibly source)) please ask me.)

References: Sorry, haven't looked yet.
Weighting: Paper - 30/Model - 70 (Paper not exceed 3 pages)
Uni ID no: 0315420
usercode: csudcy
user name: Nicholas Lee
user email:

Submission time: 30/11/2006 11:12
Name: Andrew Giles
Library Card Number: 0304384
Provisional Title: Robots in the Home: Learning to do everyday tasks.

Abstract: Autonomous robots in the home have long been the dream of computer scientists and science fiction fans alike. The thought of having someone or something to complete the everyday mundane tasks around the home is very attractive. The problem is that for a robot to be able to do even the simplest of tasks, such as pick up a dirty plate, requires a lot of lines of program code if conventional procedural programming methods are used. For more complex tasks the amount of code required is very large. If a computer also wants to learn things (e.g. the location and layout of rooms in a house) then the task becomes quite large. Humans overcome this later problem by developing mental models in their brain. For example, when I want to go to the toilet, I know where the bathroom is in my house from the mental model in my head. Humans make and use mental models so naturally that we're not aware that we are doing it. What if we could get robots to make these mental models? In this paper I want to propose that Empirical Modeling (EM) could be used to create such mental models within robots. EM is a different approach to computer based modeling where the modelers make a construal from their experience in terms of three main concepts; observables, dependency and agency. In the first section of the paper I will look at the benefits of using mental models in robots to aid their learning in a situation. I will then consider how EM tools could be used in a robot's controls to create mental models of houses and rooms in which they need to do tasks. I will do this with reference to a model I shall create (using EM tools) of a robot cleaning a room (including vacuum cleaning the floor, dusting surfaces, cleaning windows etc).

Model description: I will be modeling the cleaning of a room by some autonomous robot servant. I will begin by creating a model of an empty room within a house. I will then place a robot within the room. In a separate window will be the robot's mental model of the room. This will start off blank and empty. As the robot moves about the room and begins to see things like walls, these will be added to the robot's mental model view. The robot will then be able to clean the room according to it's mental model. I will not be concerned with how the robot cleans the room (e.g. the robot's motors which need to be moved). Neither will I be concerned with the robot's sensors. I will just assume it has the capability of detecting objects a certain distance from itself. For this model, I will make extensive use of the Eden, Donald and Scout definitive notations. I will then extend this basic model by adding some furniture in the room. The robot will then have to clean around, underneath and of top of some of these objects. Other things which require cleaning could also be added to the room e.g. windows. I could also add things into the room which do not require cleaning by the robot such as pets or people. I could further extend the model by adding multiple connected rooms thus creating a house set up. The robot would then be able to make a mental model of the whole house. I could then build a command component so that the robot could be commanded to clean a specific room. The robot would then have to use it's mental model of the whole house to navigate between rooms.

References: W.M.Beynon, S.B.Russ, The Interpretation of States: a New
Foundation for Computing? 1995
W.M.Beynon. Empirical Modelling and the foundations of AI. 1998
W.M.Beynon. Radical Empiricism, Empirical Modelling and the nature of knowing. 2003
W.M.Beynon, A.Harfield, S.Chang. Alternative model-building for the study of socially interactive robots. 2005
K.Dautenhahn. Getting to know each other - Artificial Social Intellience for Autonomous Robots. 1995
K.Dautenhahn, A.Billard. Bringing up Robots or - The Psychology of Socially Intelligent Robots: From Theory to Implementation. 1999
K.Dautenhahn. Robots We Like to Live With?! ? A Developmental Perspective on a Personalized, Life-Long Robot Companion. 2004
K.Dautenhahn What is a Robot Companion ? Friend, Assistant or Butler? 2005
K.Dautenhahn How May I Serve You? A Robot Companion Approaching a Seated Person in a Helping Context. 2006
B.Graf, M.Hagele. Dependable Interaction with an Intelligent Home Care Robot. 2001
B.J.Kuipers, Y.Byun. A Robust, Qualitative Method for Robot Spatial Learning, 1988
R.A.Brooks. Intelligence without representation. 1991
K.Dautenhahn. The Art of Designing Socially Intellligent Agents - Science, Fiction and the Human in the Loop. 1998
Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)
Uni ID no: 0304384
usercode: csudgs
user name: Andrew Giles
user email:

Submission time: 30/11/2006 12:06
Name: Martin Gibson
Library Card Number: 0207146
Provisional Title: The Applications of Empirical Modelling within the scope of Engineering Design.


Abstract: This paper intends to apply the ideas of modelling by animation to the concept of modelling engineering systems by Finite State Machines. The latter uses states and transitions, protected by guarded rules, to define the movement and exhibited behaviour of a system of states. In the past, Finite State Machines have been used [1] to model system behaviour from the point of view of testing a FSM model against certain logical theorems to prove that the system exhibits the behaviour stated by the theorem. In contrast to this, the concept of animating and defining a previously designed Locking System[1] will be undertaken to discern whether the Empirical Modelling approach yields a) a deeper understanding of the system b) a more accurate construal of the realworld system c) a better way of conveying the model to third parties, through the use of interaction. The comparison of languages SDL and LSD[3] will be extended to look at John Rushby's SAL language, as was used in [1] to model FSM behaviours. By drawing comparisons to Beynon's toy example[2], the model will be critiqued under the light of the practicality of modelling concurrent interactions, as well as the ramifications of assuming instantaneous transition between observables, whilst illuminating the danger of using variables which already encapsulate implicit understanding of their behaviour: for example the locking system's variable 'locked' which implicitly assumes that there is an instantaneous transition between 'locking' and 'locked', and that once 'locked' no external environments can modify it - for example drilling the locks and opening the door. Once a model is made, a critical evaluation between the two methods of modelling will be compared, which is expected to highlight that Empirical Modelling is more adept at coping with incomplete modelling; that both methods cannot solve the frame problem of clipping the scope of interaction with the real world (the locking system may be destroyed by a falling comet should this be modelled, or represented with the variable 'unexpected'?). It is also expected to conclude that whilst Empirical Modelling will test the designer's understanding of the model, it lacks the next step of taking the model and implementing it / creating additional value from it.


Model description: We wish to build a car alarm and central locking system. The system will consist of a key fob with one button to lock/unlock, the alarm itself, and some locks that can check their actions. When we press the key fob (we'll call this a `plip' action as that's the sound the fob makes when pushed) we want the car to attempt to either lock or unlock depending on its current state, and ensure that the locks don't seize during the process, possibly leaving the car thinking it's locked when in fact one of the locks is stuck open or vice versa. Between the states of locked and unlocked there needs to be at least one check. If the car is locked, plipping it will start the process of unlocking. At this point the car must check the locks; if the car's locks are up (unlocked) instead of proceeding to unlock, it should go no further and revert to a locked state as a failsafe. If the locks are down (like they should be) the car can proceed to attempt an unlocking action. This action will need to be checked again; if the car is still locked it should go back to a locked state to prevent itself trying to loop over lock and unlock actions. If the locks have unlocked (locks are up) then everything has worked and the car can consider itself in a state of unlock and can switch off the alarm sensors. If the car is unlocked, plipping it will start the process of locking. At this point the car must check the locks; if the car's locks are up then it should proceed to attempt a lock. If the car's locks are down (locked) already it should still attempt to lock - if the owner does have a seized door locked he will wish to get the car into a locked state so any other doors that are correctly unlocked lock and that the alarm arms.[1] This model was initially specified in a colloquial manner, as to emulate the human understanding of the system before modelling had begun. In the paper[1], this was converted to a FSM: and was tested by specifying some Linear Temporal Logic to see whether the system would always be able to lock itself regardless of component failure. Modelling this in tkEden will allow the system to be interacted through animation, by setting the lock pins, plipping the system, and seeing what the alarm does in response - the model behaviour above will be embedded into the system. From this stage, further examples of reallife observables will be discussed: separating the 'locked' state from a statement of the system to the expected outcome the system expects to find itself in, and how one could go on to modify the system to cope with such discrepancy.

[1]An attempt to formalise some system requirements - M Gibson
[2]Definitions for modelling and simulating concurrent systems - M Beynon
[3]Comparison of SDL and LSD M. - T.Norris.
Weighting: Paper - 60/Model - 40 (Paper not exceed 6 pages)
Uni ID no: 0207146
usercode: csucik
user name: Martin Gibson
user email:

Submission time: 30/11/2006 12:07
Name: Andrew Knight
Library Card Number: 0305767
Provisional Title: An Investigation into the Sasami Notation

Abstract: The Sasami three-dimensional modelling notation was developed in 1999 by Ben Carter for a third-year project. It provides an interface to the OpenGL API, allowing cross-platform rendering of graphics. It is, however, merely a "thin" layer to OpenGL, so it cannot interact with the underlying model, only display a representation of it. This paper will investigate the Sasami notation's ability to visualise EDEN models, looking at integration with existing models and usability aspects. In [2] a definitive notation for geometric modelling is described, with examples of modelling using set-theoretic operations on primitives. Sasami provides no facility for this sort of graphical modelling, known as 'Constructive Solid Geometry'. The paper will discuss how this could add to Sasami and what this would mean in terms of modelling. The topic of interactive graphics has been documented [1], and this paper will assess the relevance of this to Sasami. Along with the associated model, it will investigate how the notation could be altered to include interactive elements. This paper will also review other techniques currently available for visualising EDEN models in three dimensions, including the Java-based VRML tool, as used in the Clayton Tunnel model. This performs a similar role to Sasami, but is external to the tkEden tool. The as-yet unimplemented CADNO notation will be discussed, as will the Empirical Hyperfun project. The relationship between these techniques and Sasami will be detailed, as well as how they could interact.

Model description: The model will be used to demonstrate various issues discussed in the accompanying paper. An example program will be written to show the Sasami notation in its current form. It will demonstrate how an existing model can be visualised in three dimensions using the notation. Additionally, the Sasami notation will be extended, allowing the creation of basic primitives, such as cubes, spheres and cylinders within the software. This will allow the creating of basic models without needing to use external modelling tools. An example program will also be created based on an existing EDEN model to demonstrate these features. Finally, the inclusion of interactive elements will be investigated and programs created to demonstrate the uses and applications of this to modelling with definitive scripts.

References: [1] W.M.Beynon. Evaluating definitive principles for
interactive graphics. New Advances in Computer Graphics, Springer-Verlag, 291-303, 1989.

[2] W.M.Beynon. A definitive programming approach to the implementation of CAD software. Intell. CAD Systems II: Implementation Issues, Springer-Verlag, 126-45 (Appendix with A.J.Cartwright), 1989. A definitive notation for geometric modelling, Proc. 2nd Eurographics ICAD Workshop, CWI Amsterdam, April 1988.

[3] W.M.Beynon, V.D.Adzhiev and A.A.Pasko. Interactive Geometric Modelling based on R-functions. Proc CSG'94: Set-Theoretic Solid Modelling: Techniques and Applications, Winchester, Information Geometers, 253-272, 1994.

[4] V. Adzhiev, R. Cartwright, E. Fausett, A. Ossipov, A. Pasko, V. Savchenko, HyperFun project: A framework for collaborative multidimensional F-rep modelling. In Proceedings of Implicit Surfaces '99, 59-69, 1999.

[5] W.M.Beynon. Computational Support for Realism in Virtual Environments. In Proc 11th International Conference on Human-Computer Interaction (HCII 2005): Volume 10 - Internationalization, Online Communities and Social Computing: Design and Evaluation, Las Vegas, NV, 22-27 July 2005 (7pp. - published on CD).

[6] R. Cartwright, V. Adzhiev, A.A. Pasko, Y. Goto, T.L. Kunii. Web-based shape modeling with HyperFun. Computer Graphics and Applications 25,2 (March 2005), 60-69, 2005.
Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)
Uni ID no: 0305767
usercode: csudbc
user name: Andrew Knight
user email:

Submission time: 30/11/2006 12:48
Name: Michael de Selincourt
Library Card Number: 0300110
Provisional Title: Distributed Model of Early Naval Gun Control

Abstract: In this paper, I shall study the application of the principles of Empirical Modelling to the creation of a model of an electro-mechanical gunnery control system common on United States naval vessels around World War II. The Mk37 gunnery control system, for which my primary source is Mindell's "Between Man and Machine," has several interesting traits that lend it to modelling using the tools and techniques of Empirical Modelling. Most significantly, it is distributed, and has human operators at various points constantly working to provide data for a vessel's guns which is dependent upon observations made of the enemy's relative position and movement. The established and traditional view of computing systems developed with (and thus is based around) single-threaded procedural programmes, with a user entering input and awaiting results. Although today's software reaches far beyond this simple level of interaction and complexity, the "paradigm" or way of thinking persists, and this formal, mathematical approach has to be 'scaled up' to include concurrency, constant human interaction, etc. Distributed concurrent systems are widely recognised as some of the hardest systems to create using conventional procedural programming techniques, especially as 'conventional' programming approaches which treat the concept of an individual 'agent' as fundamental or near-fundamental are rare. Empirical Modelling, by contrast, is a research area into creating a "new foundation" for computing, one of the core concepts of which is the interaction of agents, and the constant, rather than 'alternating' interaction of human and machine. This paper will discuss the merit of an Empirical Modelling approach to creating an 'artifact' representing the Mk 37 and its predecessors, with specific reference to the advantages EM may confer in handling the distributed, concurrent nature of the system, and the need for contant human interaction. The Mk 37 naval gunnery control system provided an excellent early example of machines and humans working together on shared computational task. Given the nature of the task, it would not, perhaps, be unthinkable to consider the system as a whole, both humans and machines, to be performing a crude form of perception. If the modelling proves successful, this will provide another illustration of EM's applied uses. The model will also, hopefully, have value as an informative and engaging demonstration both of Empirical Modelling, and of the 'Rangekeeper' system it models!

Model description: The Mk 37 was, although not widely renowned, a significant part of the history of automated computing. Over the previous three decades, the difficult problem of gunnery control on naval vessels was undergoing the change from entirely human-operated to largely electro-mechanised. The change began with the simple introduction of a calculating machine to speed human computation, but by the Mk 37, the entire system, from 'input' (observation tower) to 'output' (guns!) was part of a complex electro-mechanical system. The Mk 37 was composed of three main sections, which I would model as connected entities, using the Distributed TKEden tool. Each of these sections consisted of both human operators and mechanical components. The first component were manned "directors," or spotting towers, which converted information from radar and visual observation into electronic signals. This information was passed to an armoured plotting room below decks, containing the Ballistic Computer (and, in earlier systems, its forebears) where the machine and gunnery officers would use the information from the observation towers to form a mathematical model of the situation. Results from this model would then be provided to the last section, the guns, to enable them to fire accurately. I hope the sections of the model to have a simple graphical interface, so that users can take the roles performed by humans onboard such a vessel, and see the results. The Mk37, as mentioned above, was one in a long line of progressively more automated gun control systems, and the model's development might mirror this progression, resulting in a small series of models, starting with ones in which most work must be done by humans, progressing to largely automated ones. Ideally, these would mirror the advances of the original system's capabilities, but sufficiently detailed historical records may not be available.

References: Mindell's "Between Human and Machine,"
The LSD Notation for Communicating Systems (Beynon RR87)
Ian Bridge's Cruise Control model & documentation,
The distributed Clayton Tunnel Railway Accident model & documentation,
...and as many relevant papers as I can locate in the publications folder!
Weighting: Paper - 50/Model - 50 (Paper not exceed 5 pages)
Uni ID no: 0300110
usercode: csudab
user name: Michael de Selincourt
user email:

Submission time: 30/11/2006 17:04
Name: Sandro Krumbein
Library Card Number: 0633176
Provisional Title: Distances between cars in traffic - influences, dependencies and outcomes

Abstract: The rules, which are provided by driving schools to apprentices for driving licences, for finding the right distance to the car driving in front are fairly simple. It equates down to something like 'keeping a distance of about half the reading of the speedometer in meters'. Though this might be good to have as a guideline and might describe something like the 'ideal distance', it hardly describes, what is happening on the roads in real world traffic situations. The actually observable distances in the real world are created by dependencies between factors, which are deeply rooted in human psychology, hard to grasp fully and hidden to most automobile travellers. Many different factors such as weather conditions, social norms, traffic regulations, former experiences of the drivers, the current mood of the drivers and even the properties of the actual cars are likely to have an influence on the distance a driver will finally apply. This paper will therefore describe the attempt to visualize and intuitively convey a model of dependencies as it was undertaken in the accompanying modelling study. The suggested dependencies are to be seen basically as qualitative in nature. The application of more detailed empiric research might be necessary to achieve more realistic predictions. The nature of the model itself and the approach to the visualization are well suited for application of definitive principals. Furthermore it will mention possible applications of the provided model in areas of adult education (e.g. in driving schools) as a means of making the drivers more aware of their doing, inducing more conscious driving behaviour and possibly reducing the likelihood of accidents due to too small distances. Finally it will look out on possible extensions of the given model to improve its use even further.

Model description: This modelling study is an attempt to create an environment, in which the situation of driving behind another car can be experienced in a safe and secure manner and factors, which influence the distance, which will be kept, can be explored more deeply. The basis for this is a model, which establishes quantitative dependencies between certain factors, influencing the 'real world distance'. These dependencies are then visualized with particular emphasis on the amount and direction of influence a certain observable has towards the final outcome in a certain constellation of a factors. For this a tree-like structure is used and several attributes of the nodes and edges indicate the influences. The attempt has been made to base these on facts about the perceptual abilities of the human vision to improve intuitive understanding. The awareness, which was potentially created in the mind of the user by visualizing those dependencies, should then be used create an even stronger educational impact. This can be done giving meaning to them by associating the abstract dependencies with the actual real world situation as closely as possible. Therefore it is necessary to experience the influence those observables will have on the distance and also the potential dangers of being negatively influenced by them. An accident happening due to too little a distance is the worst case, which is to be considered. Ideally this would be done by driving real cars, on real roads in real traffic situations. As this is not recommended due to health and safety reasons, another possibility would be to do so in a more secure environment in non public areas with cars especially equipped to increase passenger safety. The final approach taken in this study is it to create the experience in the form of a visualization of driving behind another car and the outcome of the applied distance in case of an emergency. This will ideally be done in (simple) 3D graphics, due to time restrictions a 2D visualization might be used as well.

Hopefully rather scientific, psychologic studies on the topic can be found.
Don't have any yet, though

Weighting: Paper - 30/Model - 70 (Paper not exceed 3 pages)
Uni ID no: 0633176
usercode: csugcw
user name: Sandro Krumbein
user email:

Submission time: 30/11/2006 18:13
Name: Robert Donovan
Library Card Number: 0323941
Provisional Title: Synesthesia For All ? Environmental Effects on Musical Experience

Abstract: There are two main areas of interest upon which this study is based. The first of these is the effect of environment upon an individual's experience of a piece of music, in particular through vision. To illustrate this, I recently retrieved a perfect example of this phenomenon from a fellow student. He told me of a drive he recently made at night from Coventry to London on which he was listening to a particular song which he had heard numerous times previously and did not find particularly emotive of late. However, he told me of how a dramatic section change within the song coincided with the lights on a lorry, motionless in a lay-by, coming on all at once, sending shivers down his spine. These two events on their own have little emotive force, but put together they clearly had a profound effect on this individual. The second area of interest is to do with the relationship between music and colours. Finnish composer Jean Sibelius had a form of synesthesia which caused him to see musical notes as colours. Synesthesia is described as a rare neurological condition which causes different sensory perceptions to become coupled and confused. However, it is not rare for people to be affected differently by a song based on the shades of their surroundings. Certain pieces of music affect me more profoundly at twilight for example, when shades of white grey are in abundance. The affects of visual perception on musical experience is of particular interest to the music video industry. The idea behind music videos if to enhance the listener's experience by providing synchronised images relating to a specific song. This principle works both ways, however, with music being used significantly in the film industry to enhance the viewer's experience of certain scenes or moments. Stave notation is the extremely familiar method for modelling pieces of music and is used by musicians to map symbols to notes. 'With four simple letters I can refer to a composer who hasn't existed for more than 200 years - not to a set-theoretic model of him, but to his very heart and soul'. This notation can be likened to a set of instructions on how to build the Cutty Sark. One may read them and get a vague picture of what the ship may be like but observing the finished product is an entirely different, and presumably far more emotive, experience. This study is therefore geared towards examining the affects of other sensory perceptions (as opposed to hearing), but particularly vision, on an individual's experience of a piece of music. The focus will largely be around the affects of modelling a song in colours but using different methods to do so. For example, one method may be to assign a colour system to individual notes and display the song as a series of colours relating to the dominant melody while the song is playing. I am also interested in the relevance of the choice of dependency between the music and colours, and how these dependencies may have different affects with different people.

Model description: The model I am proposing is similar to that of a common visualisation found in application such as Windows Media Player but will be more tailored to the experiences of the observer. The model will most likely be based on "Erlkonig" by Franz Schubert in order to build on previous work. I propose two ways in which to model a song. The first of these is to have one colourscape that refers to the dominant melody in the song at any given time. This may not refer to a single instrument, since emphasis often alters during the progress of a song, during solo sections for example. My general assumption is that listeners pick out certain melodies in songs in a similar way to the extraction of voices described by the cocktail party effect. However, the emphasis that listeners place on melodies may be completely different from person to person. This is another area for investigation. The visualisation for this method will be a screen of colour that smoothly changes based on whatever the dominant melody is doing. The second method is to separate the parts of the song that are easily distinguishable and display them in different parts of the screen. In the case of Erlkonig, this will most likely be the different hands of the piano and the vocals. These will have separate colourscapes similar to that described above. It may be interesting to see how the affect of these separate visualisations have on the observer. For example, one person may normally listen carefully to the vocals of a song, but if they focus on the visualisation of the left hand of the piano, they may be more likely to pick out this part of the melody.

References: Papers:
Meurig Beynon - Mathematics and Music - Models and Morals.
W.M.Beynon, Steve Russ and Willard McCarty - Human Computing: Modelling with Meaning.
And probably many more

Music Program - David Wai
The Erlkonig model (name?) - Meurig Beynon
Weighting: Paper - 70/Model - 30 (Paper not exceed 7 pages)
Uni ID no: 0323941
usercode: csudeu
user name: Robert Donovan
user email:

Submission time: 02/12/2006 00:08
Name: Robert Newman
Library Card Number: 0324442
Provisional Title: Investigation into the practicality using DOSTE for
empirically modelling physical situations

Abstract: This paper and the associated model are concerned with the practicality of using the prototype DOSTE modelling software for modelling real-world, physical situations.

The model is primarily concerned with buildings and structures. A structure is made of a series of 'girders', which can be attached to other girders or anchor points using the mouse. It will then respond to gravity and other girders as it should (gravity can drag it down, other girders can support it or add weight, and so on). Other environmental factors can also come into play.

This model will be created in Nicolas Pope's prototype DOSTE empirical modelling software. DOSTE is rather unique in that it does not allow features such as subroutines; everything must be expressed as a dependancy. The paper in part describes how practical DOSTE is for this kind of model; what parts of the model it handles well, and what it handles less well.

Model description: My idea for the model is to build a (simplistic) model of a physical structure, as might be made by an engineer. As the abstract describes, the way I thought of doing this would be using 'girders' to build the structures. The individual girders would be affected by other girders attached to them, gravity, and possibly other environmental factors (wind, for example). They might also respond to weight; if a heavy weight was put on top of whatever structure (or the structure was badly designed), the downward force may overcome whatever 'bonding' force holds the girders to each other and the structure would collapse.

This kind of thing has been done before; the idea is very roughly based on a game called Bridge Construction Kit (, the idea of which is to build a bridge that doesn't fall into the water when a train crosses it. This kind of thing has also been used in recent games with physics engines (Half-Life 2 being a notable example). All of these examples are, however, strictly procedural.

This would (hopefully) be done in Nick's DOSTE modelling program. I described the model to him, and he seemed fairly certain it would be possible (this is apparently roughly the kind of problem he designed his program to be able to model). I'm not completely convinced, but in a worst case, if it doesn't work, I should at least be able to talk about why I couldn't make it work...



Weighting: Paper - 40/Model - 60 (Paper not exceed 4 pages)
Uni ID no: 0324442
usercode: csudew
user name: Robert Newman
user email:

Submission time: 04/12/2006 00:31
Name: James Dowling
Library Card Number: 0225540
Provisional Title: Potential for 3D empirical modelling in dost

Abstract: This paper would explore the possibilities of the new empirical modelling tool ?dost? currently under development by Nick Pope, how well this copes with expressing a 3D model in an empirical manner and explore its potential use as a tool in this regard. I would also like to investigate the Tamworth rail disaster present in the model not only from an dost experiential point of view but also look at it in terms of its usefulness in analysing railway accidents of this nature, how well does empirical modelling lend itself to this sort of application?

The paper would focus on both the accident itself, looking into how using an empirical model in this manner could be of use in the design of railways and transportation systems in general, if the designers had had a real world example model to play with that simulated their designs, could they have spotted the circumstances that lead to the accident? How did dost perform during the process from a modellers point of view? Was it as intuitive as tkeden? The main bulk of the work would be contained in the model.

Model description: These topics will be explored through the creation of a 3D model of the Tamworth railway disaster. Given that this will be one of the first models produced in the system there will be feedback given during the modelling process to help improve the system further.

The modelling activity would involve creating a 3D representation of the Tamworth railway station, tracks, river, bridge, trains?etc. that are present in the model and then being able to run the model in a similar fashion to the Clayton tunnel model, where the trains can be controlled and the simulation can be view and controlled from the viewpoints of the various agents involved, as well as an arbitrary viewpoint (bird?s eye). To facilitate the modelling activities I would have to write a script to export the file format of a 3D editor to the format expected by dost which will be a little task in itself.

The model would operate from an empirical perspective, with the trains position dependant on many variable, its current momentum, force applied by the engine at the power level set by the agent (or force applied by braking), what position the track switch it set to?etc. The model will be based on what values the user sets for it to play out rather than a scripted single run ?accident? approach.


References: Empirical Modelling for Educational Technology
Definitive principles for interactive graphics
Human Computing: Modelling with Meaning

Weighting: Paper - 30/Model - 70 (Paper not exceed 3 pages)
Uni ID no: 0225540
usercode: csuchy
user name: James Dowling
user email: