Feedback on your WEB-EM-9 abstracts
Some general advice
- Remember that the paper component of your submission is restricted in length and must conform to a specified format. It should have a theme that relates to EM, rather than consisting solely of model documentation. You are encouraged - though not obliged - to include supporting model documentation, especially where this will aid our assessment (we may request a demo from you if this is essential in order for us to evaluate what you have done). There is no restriction on the length or format of the model documentation.
- Don't be too concerned about whether your model proposal has a big enough scope in the first instance. It's different if you're making a proposal for a final year project or a dissertation, but this is after all only an exercise in EM that can attract a good mark for a wide variety of reasons (and doesn't have to be too mono-thematic).
- Danger signs in conceiving your submission are:
- a very broad idea that is difficult to relate to a specific context that you can imagine interpreting with reference to an artefact
- an idea that puts the spotlight on modelling a closed behaviour that can best be specified by abstract rules rather than by open-ended and ill-defined agent actions
- an idea that is primarily focused on achieving a clearly defined functional goal, rather than on making an environment in which to explore the agency that might ultimately (but not necessarily within the scope of this assignment!) lead you to achieve that goal.
- Bear in mind the potential for shifting your focus if the model-building or paper-writing prove to be too challenging e.g. considering feasibility issues, introducing a comparative study component (e.g. EDEN vs JS-EDEN, EM vs alternative implementation), and changing the emphasis, whether to something more specific or more discursive. It's generally helpful in producing a good submission to blend development with writing - possibly even directing your model-building towards fragments / exercises that suit your purpose in the paper, or introducing new sections in paper to reflect issues you encounter in modelling.
- If you proposing to use JS-EDEN, and have any technical concerns about feasibility etc, it might be a good idea to run your proposal past Ant Harfield (ant@dcs.warwick.ac.uk), and possibly Tim Monks and Nick Pope - all three will have a better grasp of the technical issues than Steve and myself. You are recommended to do this via the JS-EDEN google group cited in the JS-EDEN Rough Guide. Russell Boyatt (rboyatt) would also be someone who is well-placed to comment, as might Elizabeth Hudnott. Bear in mind that we don't have dedicated support for the tools these days, so it's a good idea to explore any technical issues upfront so far as possible. The Christmas period can also be a difficult time to get attention from people! Remember also that we recognise that JS-EDEN is a work-in-progress and is likely to pose problems that you may not be able to anticipate or resolve to your satisfaction. We shall make allowance for this in our marking.
- If your topic leads you to stray outside the principal areas we have mentioned in the Call for Papers, that's really no problem. In general, it would be fair to say that good novel topics have attracted higher marks than 'more of the same old thing'. (Plus - as an indication of how much we value new ideas - the prize for originality always goes to one of the more innovative submissions - typically addressing something we've not considered before.)
Individual feedback on your abstract submissions
To be read in conjunction with the pdf of all the abstracts that can be found here.
0907931 Jonathan James Cave Understanding cryptographic algorithms through experiential learning
This abstract is not very informative. Perhaps the key sentence is the second sentence, where you frame a proposition about EM and learning. It would be helpful to understand the distinction between the kind of visualisation that has been done by Tao et al and what could be done with EM and JS-EDEN. This issue alone would be a reasonable focus for your submission. One possibility is to take a very specific instance of the operation of one of your ciphers and build a visualisation around that. The model by Duggins from WEB-EM-4 may be a good precedent for this. Note also the Enigma theme has been considered before - in WEB-EM-2 - see: http://www2.warwick.ac.uk/fac/sci/dcs/research/em/publications/web-em/02/enigma.pdf Looking at this submission - which set out with rather unrealistic expectations - may be helpful in gauging what is an appropriate scale for your study. Your abstract does not give any information about the specific security algorithms concerned. One of the problems with a short paper will be to get relevant information across so that your model-building and discussion can be appreciated. The quality of model is going to be critical in supporting this communication. With reference to the theme of security, it would be very interesting to see some attempt to give an LSD account of the agency involved (no-one has done this before, I think, but it seems natural to conceive security issues in terms of observables known to some agents but not to others). You should note that in your paragraph about the modelling study, there is really nothing that is specific to your chosen theme. "The model will be created using JS-EDEN" doesn't give insight into the reasons and motivation behind your choice of case-study. It might be useful is you were to study the nature of the visualisation ventured in Tao et al's paper. If JScript is involved in this, it would be interesting to know whether you can take advantage of this in any way. Re-engineering an existing visualisation using JS-EDEN might offer a more clearly focused theme.
0910628 Menelik Collymore Connect 4
0907045 Greg Corbett Creating an interactive learning environment powered by a JS-EDEN model to aid in the understanding of vectors
I'm not sure how to interpret "The vector space model will initially be a sparse, simple, microworld (Papert, 1993)" - 'model' suggests something abstract, and 'microworld' something more contextualised. (Also not clear how "scaffolding" - which I tend to think of as an abstract pedagogical device - can be interpreted as part of the microworld.) The second para, where you allude to 'learning to program procedurally', 'other areas of mathematics' and 'sports coaching', suggest that the potential scope is quite broad. I'm not sure that it is right to think that all these topics can be treated in the same way - you may need to be careful to be discriminating about the nature of different learning tasks etc. Be careful about scope of ambition - the adaptive systems agenda is a large one in itself - are you doing a relevant module on that theme also? Emphasis in an assignment of this scale needs to be on principles not product - so think in terms of proof-of-concept / demonstrating key aspects / "why EM?" agenda References from EMILE re learning the concept of vector may be helpful - cf. WMB's mystical efforts in this direction, which you can find in the directory /dcs/emp/empublic/emile/UsingEMPE/linearalgebra2dBeynon2011 (There is also a JS-EDEN version that may not have been made public.) This may be of interest - if only in relation to comparative study. I don't know whether the absence of Antony Harfield's PhD thesis is significant or an oversight - though clearly you are not short of material. Another reference of possible interest is the GeoGebra package. Beware of expressions such as: "The model hopes" ... I'm also not sure whether 'codenamed' is quite the right word here. Pretty ambitious agenda both written and practical here!
SBR - who proposed this theme, comments: You have done some useful background reading by the look of this. MEI is a good source. In your paper you need to give fuller references (publishers, details etc). By all means use any material or ideas from the EMILE project - but should reference that webpage for that of course. I think you have chosen about the right scope (areas to include). Certainly good to be thinking of microworlds, scaffolding and adaptive learning. Not clear whether you expect to include any applications, or assessment (exercises, or problems). Very nice idea to include some methods from sports coaching - if it's possible. Might find some more references for the adaptive learning side. By all means ask questions - I'll be very interested in what you can come up with
0913340 Jonathan Davies Modelling Aeroplane Boarding Procedures
This abstract does not make any explicit reference to EM, nor does it introduce a clearly defined "modelling study". Your theme needs development and discussion in order to frame best approach. I'm not sure that "The model will look at how these procedures work in a perfect world" is the right focus. Generally EM is better oriented towards modelling the imperfect world, where observables are personal and subjective etc. There are several directions in which you might elaborate your study. It might be worth conceiving an application that allows you simply to monitor visually the effect of boarding a plane according to some loose strategy (such as people in rows M to T now etc) - from the perspective of the staff responsible for managing the board, and then considering the implications of factors that may not be abstract at all - such as how slim, agile people are, how much luggage they have, how luggage distribution is most effectively managed. Possibly one purpose would be to call into question whether an abstract strategy that doesn't take account of factors in a more holistic way can be 'best'. Bear in mind that - as always in EM - developing an environment to enable exploration is more appropriate than delivering results in the first instance. I think you have chosen good references from one perspective - but perhaps they suggest too narrowly abstract and scientific an approach. The restaurant management / warehouse models developed by Chris Roe and Chen may be worth consulting. Not "effect" but "affect" the boarding time.
SBR - who proposed this theme, comments: Well-motivated but very brief on what you would do. As I have said to you - danger of producing simulations as in conventional programming without playing to the 'from construal to algorithm' theme of EM. Will definitely help to get credit if you can represent the different construals of the boarding task from (one or more) passenger points of view and from airline staff point of view. Consider the information flow from checkin (passenger finds seat no. on boarding card and given a time to be at gate) and the procedure at the gate and in 'waiting area' before boarding. Use details either from your own experience, or others, or make up what seems reasonable. Could be useful to describe these construals in English - but best with some LSD (cf my bus journey outline) - to complement an animation / visualisation of the boarding and some problems arising. I'd suggest aiming first at small plane of (say) 10 rows of 2+2 seats.
0927324 Clare Donnelly Modelling agreement between self-interested agents
Your writing is interesting, and reflects some independent thought, which is always welcome. There is a danger that you make use of EM vocabulary in ways that aren't entirely consistent with the 'standard' way. For instance, you refer to an 'observable situation', where observable is an adjective, and this is quite confusing when the term 'observable' might be more appropriate. It is true that EM is highly relevant to decision support, and you will find explicit references to this at the url: http://www2.warwick.ac.uk/fac/sci/dcs/research/em/applications/humancomputing/ Your references could include EM papers devoted to decision-support cited there. EM may indeed lead to automated tools for decision support, but its primary role is in providing a playground for the agency that underlies decision-making. The most interesting contributions that EM can make here may be to semi-automated activities that may be best classified as Human Computing in the sense introduced by SBR. It's certainly ambitious to think of the product of EM being sw for "negotiating amongst the agents ... to provide the 'least-unpreferable' decision" - this would only be plausible after a great deal of semi-automated decision-making had been validated empirically. In general, there seems to be a danger that you are thinking of taking the use of definitive scripts far beyond the preliminary role of mediating between manual and automated agency. This is what I infer from your observation that "The project will explore the difficulties of creating a peer to peer decision system i.e. one in which the agents decide amongst themselves. This may be challenging in a definitive language." It would be useful to know what platform and what weighting you have in mind. 'duke box' should be 'juke box' and 'less eloquently' 'less elegantly'?
SBR - who has supervised EM research relating to this theme, comments: This is coming along well I think. It's good to develop the 'smart' concept along with decision support - although there is a bit of tension in what you have written between 'automated' and decision support (when with the latter it is not fully automated but requires negotiation with user - hence the word 'support'). For example in your music example, genres will not have sharp boundaries so you might want to interrogate a participant for clarification in certain cases. Sometimes people's construal of a genre may be different from most people ('wrong'). I see the point of comparison with 'collisions' - where there is rapid tacit agreement (usually) on the 'least-unpreferable' route. (I am sure there is a better word for that!) You might have a look at the decision support article from a PhD student of mine Suwanna Rasmequan about 2002 - EM paper #061. You might browse some issues of the journal in which that was published for anything like your application. Certainly interesting also if you can link at all to 'agent technology' work in AI.
0908362 David Getley A Model for Interactively Examining Normal Form Two Player Games
You write: "I will start the project with very little knowledge of game theory, and use the experience of modelling as a way of learning about the fundamental features of games discussed above." This is an interesting premise, and it's important for the reader to understand how it applies to you. For instance, have you been following actual modules on this theme already, what prerequisites are essential etc? I like the narrative idea - but bear in mind that a conference paper would be an account of / discussion of a narrative that would be an independent piece of empirical study, rather than the report that spells out the narrative itself. There is no problem in principle in including such a report in the documentation of your modelling study (which can be quite independent of the paper) but the scale of the work needed is an issue when you consider that you might need a lot of effort to build the learning model. Hopefully you can make use of existing web resources if necessary to speed the development (or maybe use packages like Sylvester within the emile variant of JS-EDEN if that would give underpinning support). The planimeter (see Charlie Care's examples in the EM project archive) might be an interesting precedent for "learning through construction". Note carefully that there is literature (I believe) that contends that building doesn't necessarily promote learning - we would argue that EM is distinctive and potentially different - ideally your paper could critique / illuminate this proposal. Generally my concern about the references you cite is that it's not just any references relating to game theory you need - you have to consider how these may be developed / critiqued from an EM perspective.
0824599 Robin Janssens A study into potential uses of Empirical Modelling in improving the Efficiency of the Road System
This is a good theme - accident scenarios are particularly appropriate, and definitely suited to the idea of using artefacts for communication. The scope of your project is quite broad though, and it will be important to start with some small and realistic examples, and develop from that. Modelling blindspots is one of several issues that looks like a good starting point. LSD analysis would be topical in that connection, and generally helpful in relation to your grander objectives (cf. the railway accident scenarios). It is interesting to reflect on what observation is needed of internal car state - as in putting on foot brake / handbrake / steering / speedometer reading etc. It's always challenging to know how to address the limitations of unrealistic modes of input. I see that you are proposing to use JS-EDEN. Performance may be one issue to consider for example, when contemplating the animation of scenarios involved movement of vehicles etc. A good feature of your 'blindspot' idea is that it admits a relatively simple 2d animation interpretation and exploits the 'integration of design and use agendas' that EM affords. (Would be good if you could consider something rudimentary about the design of the vehicle / characteristics of the driver as influencing the blindspot). Matt Cranham's JS-EDEN car steering model from last year's WEB-EM may be a good starting point (this is cited in several variats of JS-EDEN, but only seems to run at http://cranman.uwcs.co.uk/js-eden/). Other points: No references - should cite the 'roundabouts' model etc (from previous WEB-EM proceedings) explicitly - perhaps also worth taking a look at EM archive models such as roadsysGardner1999 and roadtrafficStein2005. I'm not sure that "... the Efficiency of the Road System" - is the right title for your study. Weighting perhaps high towards the paper - justified if you do a good enough analysis, but you ideally need to do quite a significant amout of modelling. "The paper will [not shall] consider existing methods for teaching learner drivers" ... "are expected on the paper"
0830934 Ivan Steven Leong Finding the optimum strategy in a game of darts
This is a most interesting topic for which there seem to be excellent existing web resources. There are two rather different ways in which you can orient your submission. You may be interested in whether in principle applying EM principles can enhance existing models that have presumably been built without exploiting general-purpose dependency-maintenance. (For instance, making an EM model should enable on-the-fly redesign of the dartboard - even changing the board while the dart is in flight, which would suit my skill level, though moving the dartboard would sometimes also be necessary.) You might also focus more simply on the technical challenges of exploiting existing models on the web by incorporating them into a JS-EDEN environment. (I'm not sure how technically feasible this is in the case of the resources you cite.) Knowing where to put the effort in the modelling e.g. at what level of abstraction to focus the modelling is important. To what extent do you propose to model the dart board? to model the throwing of the dart? to develop new visualisation etc? There is a precedent for making a simulation of dart throwing in EDEN (this is in WEB-EM-7, but is yet to be made public), but I don't think this is where your effort is best focused. The key point may not be to simulate the physical act of throwing, though you may need to have some kind of interface for empirical assessment (e.g. a strategy for assessing a player's profile by monitoring some actual dart throwing). Ideally, I think the most interesting direction to explore is exploiting JS-EDEN as a new platform with the potential for re-using external artefacts etc that this affords (this would be from my perspective the most interesting modelling exercise). There is a delicate matter here of exploiting without plagiarising. Overall, I think you probably have over-ambitious / inappropriate expectations, as in: "Lastly the model will be able to host a game of darts where a human can play against the computer using the optimum strategies learnt previously." There is a danger here of falling into a common trap of setting out with a clear functional objective and trying to produce an "EM product". Having too specific a functional objective will limit the scope for EM somewhat. The more narrowly specified the agency you consider, the less interesting your EM exercise is likely to be. This is why I am uneasy when you write: "The agent will be the player which throws the dart to the board".
0814909 Sarah Marshall Empirical Modelling as an Experimental Problem Solving Application
I had some initial difficulty interpreting the way you refer to EM in your title and the first line of the abstract. Is EM "an application"? EM "as a potential medium for accessible problem solving and the study of problem solving methods." seems nearer to the mark. But is EM a medium? Your abstract is thought-provoking - I'm not sure how to maintain the EM focus when addressing problem-solving in general. From that point of view, relating EM principles to a few specific instances of problem-solving / illustrating general problem-solving principles with wide applicability seems a good idea. I suspect that we have to have understood how to solve a class of problems before we know how to observe and interact. For instance, "chess-problem solving" and the related "chess problem-solving" are contexts where a particular mode of observation may be significant in guiding the search for a solution. I'm intrigued by what you write in your modelling study: "All models will be accompanied with presentation environment slides to suggest routes of exploration, explanation of model features and our own stream of thought. We also hope to use the JS-EDEN specific input features, such as slider bars, to produce a more flexible way of reasoning within a given puzzle or problem than is possible in other environments." This seems to me to cast your EM construal not so much in the role of a Problem Solving Application but that of an instrument for a mentor to use in teaching problem-solving. There are definite advantages in being able to take actions that take you off the right track for instance, and this is the sort of option that EM can - uniquely? - support. Though it may be outside the scope of your concern, I'm also interested in your emphasis on problem-solving in Mathematics, Programming and Computer Science, which - on the face of it - seems to be limiting the scope for EM. (I'm thinking of the potentially inhibiting influence on modelling of a right/wrong answer. I don't know how to compare this with say crossword puzzle solving, where relevant observation can be exceptionally broad in character.) But perhaps by focusing on problem-solving itself, you are recognising that - however precisely / narrowly specified the final answer may be - our states of mind whilst solving a puzzle can be subject to just as complex and subtle states of confusion. Perhaps of some interest as potential case studies are Ian Jones's Sum Puzzles http://homepages.lboro.ac.uk/~maij/ and http://www.sumpuzzles.org/ - which we have looked at from an EM perspective in the past, and of course the various Sudoku construals. Perhaps your critique could be as important here as your practical contribution.
1266015 Judy Palimonka The Operation of Web Tracking in Targeted Advertising
This proposal looks really interesting. You have identified an application that lends itself well to conceptualisation in terms of ODA, and represents a well-motivated practical application. I presume that you have a JS-EDEN implementation in mind, and trust that JS-EDEN will be mature enough to support this. It strikes me that your topic is well-suited for taking advantage of the potential for new EM applications opened up by JS-EDEN, and is interesting and timely for that reason, whether or not it is possible to achieve all you would like. Revisiting your own previous project on this theme is an excellent idea so long as you ensure that the novelty in your modelling study is clear, and that your paper brings out the relevance of an EM perspective. I presume that your existing application doesn't exploit a general-purpose dependency maintainer in the way that JS-EDEN can. This potentially gives scope for a comparison of implementation approaches. I think the weighting you propose looks appropriate, and you have good fallback positions if the model-building proves unexpectedly problematic. A good study of feasibility and an account of problems encountered could still attract a high mark. There are also some interesting aspects of your application that would potentially fit well with the EM theme. The LSD notation may be quite relevant here, for instance, since it is concerned with what agents can see and manipulate.
0921741 Stefania Papaconstantinou Migration of the 'Hunt The Wumpus' game model into JS-EDEN
The use of the term "migration" in your title is interesting cf. porting - though I'm not sure if it's significant!? It would be rather nice if you could transform a model from EDEN to JS-EDEN in an incremental way, but there isn't yet a close enough integration of the two environments for this to make sense. "JS-EDEN ... enables the user to implement and use the model in a familiar object-oriented way" - is probably not the best way to characterise what you can do in JS-EDEN. There are some aspects of prototype-based OO that may be relevant, but obviously the key issue is whether you will be modelling with reference to the observable, dependency and agent concepts and this is not consistent with thinking primarily in OO terms. (It is more appropriate to think of objects as emerging in EM, as clusters of observables are recognised as being associated with each other in the context of interest as it is established.) The phrase "literature of communicating construals" is slightly odd - do you have a concept of 'communicating construals' where 'communicating' is an adjective qualifying 'contruals', or is 'communicating' a verb here? Your meaning is a little obscure either way. There is definite value in revisiting Yob's original motivation, and checking out what Cole had in mind in creating his model. The significance of the EM construal relates primarily to the light it throws on how reasoning relies on assumptions about context - assumptions that can be readily subverted in the EM version, and may be interpreted as highlighting the limitations of a logicist outlook on AI. On that basis, I don't know to what extent it's appropriate to think of Cole's construal - as opposed to Yob's original game (if I'm guessing correctly) - as "demonstrating this AI concept". (Actually Cole's construal is problematising an AI concept.) So important here to be reflecting on the grounds for comparisons you make with other implementations of Hunt the Wumpus on the web, and the extent to which these can deliver the same insight into the relation between reasoning and the context for action. Plus, of course, the purely technical issues that might inform such a comparison.
0902814 Adam Riley Empirical Modelling Proposal
Your title - if it is a title - is very uninformative! It's always a good strategy in choosing a title for your paper to refer to keywords - such as 'emotion' and 'EM' that give the reader some idea of what to expect. Modelling human emotion is certainly an ambitious theme. You write: "I propose to use the EM modelling environment to define a state based Artificial Intelligence capable of simulating human emotion." This seems to put the primary focus on an automation task ... not sure that this kind of task is most appropriate for EM - though it may depend on how you propose to tackle it. There are lots of reference to agents in your abstract - but few to "observables", and just one to "dependency": "representing each emotions [emotions's] overriding effect on others as a dependency network, with stronger emotions having more of an effect when altering weaker ones." This is quite vaguely expressed, and I don't know whether it reflects vagueness on your part re the actual mechanism. Perhaps interesting (even essential?) to ground in a particular scenario, as for instance when modelling emotions in a scene from a play? (cf. David Pechey's 3rd year project last year which involved creating a 3d environment in which animated characters could 'express' emotions) You otherwise have a rather abstract topic and will find it hard to apply the notion of relating a model under construction to an independent experience. Regarding your references, Aaron Sloman is someone who has written quite a lot on the theme of emotion in a computing context that may be worth looking into. observable measur[e]ments from a human subject "that|each" "PROPOSESED"
0910210 Daniel Robertson Wireless Sensor Network
Note that this is not really a title! - though it does designate the subject area. WSN does look like a promising area of application for EM thinking. I'm assuming that the choice of topic is suggested by a 4th year group project where you are working alongside Joseph Yarnall, and that it would be natural for you to liaise to some degree in preparng your WEB-EM submissions. This is potentially interesting, as it is possible to take account of how an EM approach might help you in to work collaboratively (both in practice and in principle). It's obviously important to ensure that you are doing work that is sufficiently independent - an issue to which our external examiner drew specific attention last year. It's perhaps a good idea to propose a collaboration, but to find a clear differentiation between your projects so that you are working with a different primary emphases in mind. For instance, you could focus on orthogonal concerns (such as model-building to assist in minimising energy consumption, exploring a design approach based on modelling using dependency, visualisation of a WSN via an EM construal etc). Alternatively, you might consider comparing the use of EDEN and JS-EDEN in model-building. Of course, it would be unrealistic to imagine that you can do more than demonstrate proof-of-concept in addressing some aspect(s) of the application domain. A useful precedent for a collaborative project (which took the form of an MSc dissertation project in 1997-8) is the Virtual Electronic Laboratory which can be found in the EM archive as velShethDOrnellas1998. In this project, which illustrated the benefit for collaboration of modelling with definitive scripts, the electronics was modelled by D'Ornellas and the visualisation was created by Sheth. (There may be other references to this project in subsequent EM doctoral theses.) Another source of inspiration might be the EDEN model of wiimotes that was developed by Antony Harfield. In your abstract, you give a good account of wireless sensor networks, but bear in mind that you will earn credit in your actual paper for writing about EM-related issues. (A less than satisfactory WEB-EM paper sometimes resembles a summary of an application field, with little relevance to the theme of the module!). You will also need a establish a more concrete and specific focus for the model-building - in this respect, consulting Joseph's submission and my feedback on this may be helpful. "the (one) I will"
0914857 Bernard Sexton SVGJSEden: extending the observable capacity of a drawable
This is an excellent abstract and modelling study. I think the content and conception are good, and my main concerns are about notation and vocabulary: I think it's more appropriate to refer to "the concepts of observables, dependency and agency" rather than "the principles". "jsEDEN"? - perhaps we should converge on JS-EDEN! SVGJSEden should somehow be consistent with whatever notation you choose: perhaps JS-SVG-EDEN or JS-EDEN-SVG reflects the semantics better. Also, the phrase 'observable capacity of a drawable' in the title seems a little odd. Certainly 'capability' is closer to the right term that 'capacity', but perhaps it's the observer who has the 'capability', and the drawable that offers an 'affordance'. If you can find a way to express what you have in mind in simpler words that might be better ('drawable' is just a word that Matt Cranham coined when drawing up the left-hand panels of the emile interface as far as I know). The phrase "extending the range of JS-EDEN drawables" might be an improvement, for instance, though I like having the reference to "observables" in the title. So even "extending the range of JS-EDEN drawables to support richer observables" seems better. Elsewhere, you could consider being more economical with words (my problem too!). Do you need "for WEB-EM9"? Is it possible to simplify phrases such as: "consider an SVG implementation of the JS-EDEN environment in pursuit of enhancing the notion of ..." and "The motivation for doing this is grounded in improving how models are construed ...". You have lots of very good things to say, and you will need to write concisely in order to do full justice to your work in a conference-style paper format. I particularly like the way in which you have linked your modelling study to themes such as 'efficiency of browser state update' that you highlight in your abstract. This will enable you to make a clear connection between your paper and your model that is one of the strongest indications of an effective WEB-EM contribution. it's [should be: its] state, 'effect' should be 'affect'
0904778 Amit Shah Modelling the positioning of players in a five-a-side Football game
You are proposing an interesting approach to simplification: focusing on where players should be, rather than simulating their actual movement. I interpret your proposal as modelling the current state of play in terms of the ideal configurations in which players should be when the ball, or a key player is transferred from one place to another. (You make an interesting analogy with American Football which reminds me of Craig Rice's WEB-EM-06 model of American Football which I will make available for you.) Such a mode of observation is restrictive in some respects, but has merits for EM I think (cf. the kind of simplified observation in the racingcarsGardner1999 project). There is scope for an interesting discussion here. For instance: modelling based around relocation of players in an ideal configuration rather than on actual movement restricts your scope for modelling obstruction etc. You need to be clear about how dependency is to be invoked. In some ways this is related to the issue of "level of abstraction". When modelling the movement of players "realistically", there are many intermediate states en route to setting up a configuration, but your vision for dependency is based I think on how the configuration of players in an situation should ideally change 'immediately' in response (e.g.) to a movement of the ball (which may again be modelled by a relocation rather than a trajectory in time). I'm not sure that introducing buttons is the most appropriate way to model this in the first instance, as this suggests a very tightly circumscribed repertoire of reconfiguration patterns. In principle, the future locations of players should be influenced strongly by the specific configurations in which they are currently, and not simply specified by a button press that imposes preconceived relocation pattern. It could be interesting to consider the relationship between an idealised relocation of players (which is "what a button press might specify") with an actually feasible relocation (taking account of factors governing capability for movement). In modelling of this kind, relevant observables might include time, fatigue levels and the speediness of players. In that connection, it's not clear to what extent you are appreciating the potential scope of observables involved in making a good model! - cf. "The observables in the model will be the players themselves." your references are thin - and do not show much awareness of what has been introduced in the module by way of LSD etc. For instance, footballTurner used LSD and ADM. From your references, I presume that you intend to use JS-EDEN, but your reference is definitely to last year's version! Performance may well be an issue if you attempt to model actual movements. Note though that you would not necessarily need to visualise player movements in order to determine whether they are likely to be able to reach an ideal location, or whether their path would be obstructed. Lots of little typos: dependEnt, a constRual (more than one instance!), taCtics
0903457 David Robert Walton Empirical Modelling and Guitar Notation
Relating music notations to interpretation (here in the limited sense of 'what are the actual notes?' and basic musical characteristics - harmonic and melodic) is an excellent idea. Ideally you might also like to consider observables like current key etc (cf. the solfa notation where key is abstracted away), and fingering. The kind of observation that you might consider is in fact so rich that it would be easy for your modelling study to get out of hand, but it might still make good sense to contemplate a broader range of observables in principle than you can study in practice. Your observation that "All forms of notation are different ways of converting the writer's [composer's?] informal, hermeneutic interpretation of a piece of music into a formal artefact" applies particularly to Chopin's piano music, where it's well-recognised that Chopin had difficulty notating what he actually played. There is a danger that the technical challenge of delivering on the modelling front will prove too great - there's a lot involved in delivering to the agenda: "This model should include visual representations of each of the different forms of notation, and allow the user to view a number of examples of chords and short pieces of music in each of the different notations. It should also be possible to play the written music, and at the same time as parts of the piece are played, corresponding elements of notation should be highlighted." A significant question is whether you propose to use EDEN (perhaps building on the musicWai2000 project in your references) or whether you venture to use JS-EDEN. Whichever choice you make, you need to be careful in selecting the focus for your study so that you don't hit unforeseen technical problems. Perhaps it would be good in principle to identify musical notation in online JScript examples that you can then re-use. You have a good selection of references, and lots of ideas for model-building ingredients. (As another reference that may help you gauge what is realistic, you might want to look at Henry Franks's music-related submission from WEB-EM-05.) The key concern will be to find things you can model in practice that can clarify the connection between the many themes you can identify as topical in principle.
0905247 Joseph Yarnall Empirical Modelling of Wireless Sensor Networks
[Many of the comments I have made about Daniel Robertson's abstract submission apply to your submission also - and I shall not repeat them here.] Your abstract has a more appropriate title than his, and gives more indication of where EM thinking may be relevant. Your reference [4] is particularly interesting in this respect, and perhaps suggests one possible theme that may be topical: using EM to support concurrent systems modelling that is based around state. You have spelt out a good focus for a modelling study with a concrete proposal for the scale of your model, and clear indications of the key observables and dependencies with which you are concerned. Though you can surely pursue this as an individual project, it may be to your mutual advantage to find a way of collaborating with Daniel, as I have outlined in my comments on his abstract submission. Your references should be more precisely specified, to include dates and page numbers where appropriate.