LARGE

Repast: An Agent Simulation Toolkit, March 17, 2010

2010-03-29T14:53:00.014+02:00

In this meeting Milan gave an introductory presentation about Repast Simphony (Repast S) agent-based modeling toolkit. During the meeting, Milan showcased the official Repast S Predator Prey tutorial and provided links for software download, documentation and tutorials. A brief introduction to agent-based modeling (ABM) paradigm and potential applications domains have also been given.

The advantages of a specialized toolkit for ABM, instead of using a general (usually object-oriented) programming language, are firstly in the collection of libraries for agent-based design, which covers features such as scheduling, communication mechanisms, interaction topologies (networks, grids, GIS), facilities for storing and displaying agent-states etc. Another advantage of an agent-based toolkit such as Repast S is the runtime environment, which allows us to set up simulation parameters, to execute and visualize simulations, to probe agent states, to generate graphical displays of the outputs etc. Repast S has particularly rich support for linking the results to various external programs, as well as libraries for advanced computational techniques (genetic algorithms, neural networks, regression models, Monte Carlo method etc.).

One of the most interesting features of Repast S, however, is the graphical development environment which allows us to create agents by drawing (drag-and-drop) flowcharts on top of which groovy code is then automatically generated. Also, by setting properties of various model elements and by using wizards it is possible to make further adjustments to the model without or with little actual coding, which seams an interesting features for those who would like to develop agent-based models but still possess limited programming skills. Of course, an experienced Java programmer can also write Java code from scratch, without using this graphical user interface.

Modeling Competitive Bidding: a Critical Essay, March 3rd 2010

2010-03-03T17:49:00.003+01:00

Today I (Meditya Wasesa) presented a paper review in LARGE group. The paper is from Management Science (1994) entitled “Modeling Competitive Bidding: a Critical Essay” written by Michel Rothkopf (RIP) and Ron Harstad. These 2 big economists in auction area write an essay which explains that there is big discrepancy between the existing (though its 1994 paper, I think it’s still relevant to the current research context) auction theory and the real practice, so that the developed theory is having limited application to the real live condition. My main interpretation of the whole paper is bundled by the figure shown below:

From this paper, we learn that the theories which try to model the competitive bidding behavior in auctions (both decision theory and game theoretical model) are mostly written in a single and isolated auction context. Moreover the assumptions which frame the theories (e.g. single isolated auctions, fixed number of bidders, symmetric Nash equilibrium, risk neutral bidders, etc) also limit the real world applicability. They suggest the researchers to adjust and enrich the elegant but simplified existing models so that the bidding model could be utilized in the real world situations.

There were a was discussion on how should we develop an analytic model which will fit to our research interest, which is the Dutch Flower Auction. Our research context is unique, since we are dealing with multi unit, multi bidders, multi suppliers, multi attributes, and highly interdependent auctions. We agreed that this paper would be considered as our main reference since it brings a lot of insights on how to make a realistic an applicable model, not just an elegant model which is good for theoretical sake but could not be applied in reality.

Why and When Preferences Convex? Threshold Effects and Uncertain Quality: February 3rd, 2010

2010-02-04T19:58:00.005+01:00

In the LARGE meeting, Yixin presented the paper by Trenton G. Smith and Attila Tasnadi.

The authors discuss the circumstances under which convexity of preferences are beneficial. Particularly, they investigate a setting in which goods possess some hidden quality with known distribution, and the consumer chooses a bundle of goods that maximizes the probability that he receives some threshold level of this quality. It is shown that if the threshold is small relative to consumption levels, preferences will tend to be convex; whereas the opposite holds if the threshold is large.

The proposed theory helps explain a broad spectrum of economic behavior (including, in particular, certain common commercial advertising strategies), to some extent. However, some of the assumptions used in developing this theory are often violated in real-world problems. Further, the cases discussed in this paper are extremely simple: the consumer only need to make a decision about his/her consumption of two products with respect to a single attribute quality measure. Therefore, the generalizability of the theory to more complicated real-world cases is not clear.

Nevertheless, this paper does add some insights to the existing work on preferences and suggests that we need to rethink when taking the convexity assumption of preferences for granted.

Representing Eliciting and Reasoning with Preferences: Nov. 11 and Nov. 18, 2009

2009-11-18T22:15:00.010+01:00

In the LARGE meeting, Yixin presented the tutorial paper about preference handeling by Brafman&Domshlak.

In the first session (Nov. 11, 2009), we focused on models and languages for representing preferences. Starting with the quantitative languages, we discussed about two important classes of value functions: Additively Independent (AI) and Generalized Additively Independent (GAI) functions, as well as some nice properties of the corresponding representation structure. Despite of the computational efficiency for preference comparison or ordering, those quantitative languages put too much cognitive burden on the users' slide and they are usually difficult for users to reflect upon. Hence we need somewhat more "easy'' languages which, hopefully, carry with them desirable properties from both cognition and computation aspects. Given these considerations, a natural choice would be taking generalizing preference expressions. When equipped with CP-nets, such quanlitative language can handel different kinds of queries very efficiently.

In the second session (Nov. 18, 2009), we started with some discussion about preference compilation techniques, namely, structure-based and structure-free. Both aim to combine the quantitative and qualitative models in such a way that we can map diverse statements into a single, well-understood representation. Then, we had a quick look at uncertainty and utility functions. During the rest part of the meeting, we focused on preference specification and elicitation, discuessed about different methods which include "prior-based'' ones such as maximum likelihood and Bayesian reasoning, as well as ''prior-free" minmax regret method. For elicitation, since it usually works in a sequential way, we need to come up with either nice heuristics to determine the optimal sequence of queries. Or if the computational price is not a big concern, we can use the elegant partially observable Markov decision process (POMDP) model.

The paper provides a very nice framework of preference handeling. In the nex few months, we will study the existing methods for preference elicitation and representation and ideally be able to construct a more concrete framework to model preferences in energy market.

Dutch Flower Auction Recommender Agent

2009-11-03T22:22:00.002+01:00

Last LARGE meeting, I (Meditya Wasesa) presented the status of the Recommender Agent development in the Dutch Flower Auction domain.

I reported that we have finalized the first step of data exploration, that is meant to elicit all determining factors which significantly influence the revenue of a good in Dutch Flower Auction domain. Our exploratory research infers that there are several auction design parameters which the auctioneer can adjust to control the level of revenue.

Knowing these auction design parameters, we have developed an analytical that we will drive the logic of the future recommender agent. At future we will implement our analytical model to a recommendation agent and will test it with simulation, lab experiments, and field experiments.

Decision coordination in a supply-chain agent

2009-10-20T21:47:00.003+02:00

Why is decision coordination an important problem? Decision support is especially critical when human decision makers are faced with combinatorial problems, uncertain and partially-visible data, and our built-in cognitive biases. When the results of decisions in different domains interact with each other, we have an additional level of complexity that is easy to ignore.

We use the TAC SCM simulation to model complex, interacting decisions, and the fully-autonomous MinneTAC trading agent to test our ideas for decision support. The agent must trade in two competitive markets simultaneously, while managing its own inventory and production facility. A successful agent must buy parts it needs to manufacture and sell finished products, and it must sell its finished products at a profit. Inventory is relatively costly, competition is stiff, and agents that cannot effectively coordinate their decisions are easily defeated.

We reviewed the coordination strategies employed by a number of successful agents. These include a "sales pull" method, inventory-centric approaches, a production-centric approach that fills its future production schedule with the products that are expected to give the highest marginal profit, an approach that projects a "target demand" into the future that is expected to satisfy profit targets, and a multi-layer system of internal "markets" in which projected customer demand bids on products, which in turn bid on parts and production capacity, etc.

We wrapped up by looking in some detail at the behaviors of the two top agents in the 2009 competition, DeepMaize from the University of Michigan and TacTex from the University of Texas. They were very nearly tied, although TacTex bought and sold considerably more volume and carried much larger inventories. We looked at an example where TacTex built up a large finished-goods inventory during a period of low customer demand, when parts are inexpensive, and used that inventory to keep prices depressed during a later period of high demand, when other agents were competing for parts and were consequently squeezed for profits. This is clearly a very risky strategy, but we assume that the TacTex team has used its machine-learning expertise to recognize the market signal patterns that indicate a reasonable probability of such a situation occuring. The regime model used by MinneTAC could presumably predict such a situation also, but so far it's not being used to drive strategic decisions.

Meeting Minutes - 27 May 2009

2009-05-29T17:40:00.003+02:00

At 27th May 2009, we have 2 guest researchers, Xiaoyu Mao and Ducco Ferro, from Almende, a research company that focusses it research in multi-agent system. Both were discussing their papers that they will present at AAMAS.

Xiaoyu Mau presented an application of agent technology in the airport operation entitled "Heterogeneous MAS Scheduling for Airport Ground Handling", while Ducco presented his paper entitled "The Windmill Method for Setting up Support for Resolving Sparse Incidents in Communication Networks". That was a great meeting with great crowd also.

Meeting Minutes - 6 May 2009

2009-05-07T11:03:00.004+02:00

At 6th of May, we have a guest speaket from TU-Delft. His name is Mathijs de Weerdt. He presented his ACM paper regarding the Qualitative Vickrey Auction concept. A concept at which bidders and auctioneers (centers by his definition) put the agreement of the bid not based on the monetary value (highest or lowest price) but based on the offer demand attributes suitability rank.

He explained this concept in the reversed auction environment example. A concept at which the auctioneer has already defined the rank of offers preferences (the offer contains of many attributes). And the bidder that submits the offer of highest rank wins. The winner bidder then can choose to deliver the offer ranges from his bid point offer to the second highest bidder point offer (Vickrey Auction).

Nice concept, based on the attributes rank preferences. For more explanation you are welcomed to browse our site to download the paper directly (http://large.rsm.nl/meetings.xml).

Meeting Minutes - 22 April 2009

2009-04-22T19:11:00.004+02:00

Today I (Meditya) presented a Herbert A. Simon's Paper entitled "Machine as Mind" appeared in "Android Epistemology" by C.Glymour, K.Ford, and P.Hayes.

Defining that the primitives of mind consist of, Symbols, Complex Structures of Symbols, and Processes that Operate on Symbols (Newell & Simon, 1976), the central thesis of his writings states, "Conventional computers can be, and have been programmed to represent symbol structures and carry out processes on those structures in a manner that parallels, step by step, the way human brain does it.".

In his writings he (Simon) basically argued that every angle of human mind is translable into definable representations, and the definition can be translated and embedded to a thinking machine. In order to defend his thesis, he explained several important points of disscussion such as:

- The Concept of Decomposable System
- Two Approaches of Artificial Intelligence (humanoid and non humanoid)
- The Concept of Mind from Psychological Perspective
- Selective Heuristic Search
- Recognition: The Indexed Memory
- Seriality: The Limits of Attention (short term memory)
- The Architecture of Expert System

In addition he also responded and defended his thesis to the disputes that people usually have regarding the angle of thinking that machine could not copy such as:

- Semantics
- Intention
- "Ill Structured" Tasks
- Language Processing
- Intuition
- Insight
- Creativity

At the end of this meeting we have a nice discussion about the state of the art of this writing (1995) and the condition that is happening now, and the extent of the applicability of the Simon's vision in the present and the future.

Meeting Minutes - 15 April 2009

2009-04-15T18:08:00.002+02:00

Today I presented the paper “Computational Intelligence in Economic Games and Policy Design” by Herbert Dawid, Han La Poutré, and Xin Yao (IEEE Computational Intelligence Magazine, 3(4), 22–26, 2008). The paper provides an overview of applications of computational intelligence techniques in economics. Both strong and weak points of the use of computational intelligence techniques are discussed. According to the authors, the two most important weak point are the issue of empirical validation and the issue of robustness.

I also discussed my own view on the relation between mainstream economics on the one hand and agent-based computational economics on the other hand. Due to the increasing popularity of experimental economics and evolutionary game theory, mainstream economics focuses more and more on bounded rationality and dynamic (rather than static) analysis. From this perspective, the difference between mainstream economics and agent-based computational economics is smaller than is sometimes thought. I argued that the main difference is between following a mathematical approach (as mainstream economics does) and following a simulation approach (as agent-based computational economics does). There is much to be gained by combining these two approaches. Today’s meeting ended with a discussion of the difference between mathematical analysis and computer simulation, and how this difference relates to the difference between deduction and induction in science. We also discussed the issue of implicit assumptions that are hidden in technical details in agent-based computational economics research.

Meeting Minutes - 25 March 2009

2009-04-09T13:19:00.003+02:00

In the presentation, Romke and Otto presented the outcomes from the master seminar in computational economics. The research was focused on improving the price prediction mechanism of the MinneTAC agent. The MinneTAC agent is an agent-based computer model that is developed by the university of Minnesota in cooperation with the Erasmus University.

The MinneTAC agent competes with other Agent based computer models in the Trading Agent Competition for Supply Chain Management (TAC SCM). The TAC SCM game was designed to come to the best solution for an Agent based computer model that is capable of dealing with the problems of a dynamic supply chain.

In the MinneTAC agent, there is an ensemble, consisting of multiple price predictors used to predict the future market prices. The function of the model selection mechanism is to determine the most accurate price based on the predictions from all the individual predictors making up the ensemble. The advantage of using multiple predictors is the ability to capture more features in the data then a single predictor. The disadvantage of using multiple predictors is that different features are captured that causes different predictions. A second disadvantage is that not every price predictor is performing optimal for every time horizon and quantity of training data. To overcome these disadvantages, there is a dynamic weighting mechanism with adaptive weights developed for the MinneTAC agent. This weighting mechanism has to find to the optimal weights for every price predictor for every time horizon. The weights are learned during the game, while the agent is competing with its competitors for customer orders. When the agent starts, every price predictor has an equal weight. During the game, the MinneTAC agent starts using the optimal weights. This means that the price prediction mechanism is not working with the optimal weights during the first phase of the game.

In our seminar research, we found the optimal weights for every price predictor during the game. This data is used to bootstrap the agent to increase the performance in the first phase of the game.

R.J. Romke de Vries
O.B. ter Haar

Meeting Minutes - 11 March 2009

2009-03-12T18:43:00.004+01:00

At this meeting Milan presented his paper "Overconfident Investors in the LLS Agent-Based Artificial Financial Market," as a preparatory talk for the upcoming IEEE SSCI CIFEr 2009 conference.

This paper is a part of the relatively new research stream where agent-based models of financial markets are used to study various topics of behavioral finance. It has been recognized in the literature that such models could be very suitable to make a link between the behavioral biases of individual investors and the aggregate market phenomena, such as the dynamics of the market prices.

In this paper we focus on overconfident investors, and model overconfidence as miscalibration, in such a way that investors are too certain in their predictions of future returns of a risky asset. In the methodological sense, this paper follows an incremental approach, where an existing model (Levy, Levy, Solomon, 2000) is modified to study the consequences of introducing a variation in investor behavior, namely the overconfidence bias.

We find that overconfident investors create less frequent, but more extreme bubbles and crashes in the market, compared to the original model. Furthermore, more overconfident investor introduce more excess volatility of the market price (over the volatility of the fundamental price), and also reduce the trading volume (as they are highly invested in stock during bubbles). Since this is a rising market, overconfident investors tend to take a larger share in the total wealth of all the market participants.

Furthermore, we study the emergence of overconfidence through biased self-attribution. Investors, who attribute successful predictions to their own skill and unsuccessful predictions to bad luck, learn quickly to be overconfident, and remain at such a high level of overconfidence. For unbiased self-attribution, the level of overconfidence varies greatly depending on the success of predictions.

During the meeting we had a fruitful discussion about the implications of these results, and there were also very interesting suggestions for the future research. In the light of the recent developments of the real-world financial markets, it would be particularly interesting to study the consequences of investor overconfidence in a declining market.

Meeting Minutes - 25 February 2009

2009-02-26T15:10:00.002+01:00

Today I presented my research done in the context of my master's thesis. I presented a novel product pricing approach for the TAC SCM game, where products are sold through reverse auctions with sealed bids (i.e., traders bid on customer requests for quotes and cannot observe their competitors' pricing behavior).

The new approach is based on price distribution estimations, where the relation between on-line available data and distribution parameters is dynamically modeled using economic regimes (characterizing market conditions) and error terms (accounting for customer feedback). Given the parametric approximations of price distributions, acceptance probabilities are estimated using a closed-form mathematical expression. These probabilities that a customer accepts a price offered by a trader can be used to determine the price yielding a desired quota. The approach has been implemented in the MinneTAC agent and tested against a price-following product pricing method in the TAC SCM game. The novel approach significantly improves performance; more orders are obtained against higher prices. Profits more than double.

During the presentation, (adaptations to) the game specifications of the TAC SCM game were discussed in the group. We briefly discussed the possible impact of new entrants during a game, more competitors, and a randomized game length. Apparently, according to game theory, end-of-game effects would be reduced when the game end is randomized.

Alexander

Meeting Minutes - 18 February 2009

2009-02-19T16:09:00.004+01:00

Today I discussed the main results of my master's thesis, in which I try to incorporate procurement information into an economic regime model based on sales information. This model is used in the MinneTAC agent, which is an artifical trading agent that competes in the TAC SCM game.

First, I gave a brief description of the TAC SCM game, after which I introduced the regime model as it is currently used in the MinneTAC agent. Then, I introduced a new procurement variable, i.e., offer prices, after which I elaborated on both regime identification and regime prediction. Finally, I discussed some experimental results.

During the last part of the presentation where I introduced some experimental results, we had a discussion on the causes of these results. As it seems, implementing procurement information into the regime model does not lead to better performance of the MinneTAC agent in TAC SCM games. In fact, the agent gets more orders from customers, but generates lower profits. Other competitors seem to take advantage of the situation, since their profits increase when MinneTAC uses the new regime model.

A suggested cause is the fact that there could be a delay regarding the procurement information. Procurement information might be a leading indicator for regimes, so perhaps creating a regime model based on the sales price of yesterday and the procurement offer price of for instance three days earlier could improve the performance of the agent. However, this is dependent on the cost allocation of the agent. Furthermore, there is a lack of adaptivity in the regime model and therefore, the model cannot adjust properly during a game. Finally, regime information is not used for price setting in the experiments discussed today. Establishing this connection between regimes and price setting could improve the overall results as well.

Frederik

Meeting Minutes – 11 February 2009

2009-02-12T08:17:00.002+01:00

Today I presented some preliminary results of a research project that is intended to be part of my PhD thesis. I discussed an economic model of price competition between spatially organized firms. I discussed the Nash equilibrium solution of the model, but most of the time I focused on the outcomes achieved by so-called evolutionary dynamics. These outcomes were analyzed mathematically (partly analytically, partly numerically) but also through computer simulations. Interestingly, on average firms turned out to cooperate (or collude) with each other under the evolutionary dynamics.

We had a very fruitful discussion about the conditions necessary/sufficient to achieve the cooperative outcome, focusing in particular on the plausibility of various assumptions. We also discussed on a general level the methodology typically used in economic/game-theoretic research.

Ludo

Meeting Minutes - 21 January 2009

2009-01-26T17:16:00.002+01:00

The 21st of January Peter Berends presented "The Impact of Clock Speeds on Bidders’ Arousal in Dutch Auctions". A paper by M. Adam, J. Krämer, C. Weinhardt (2008).

In this study an experiment was conducted to show the impact of clock speed on "competitive arousal" and seller revenue. This paper was of interest to us, since in order to create intelligent agents that support human actors in the Dutch Flower Network we need to investigate the influence of speed in Dutch auctions.

When considering Adam's et al. study, the results in regard to seller revenue in this experiment are somewhat contradictory to prior experiments. It was found that seller revenue in a slower auction was not significantly different from fast auctions. Furthermore, differences in the variance of seller revenue between the slow and fast auctions were observed. Faster auctions have greater variance. This led to authors to believe that competitive arousal is not a function of elapsed time alone. This was previously asserted by Katok and Kwasnica (2008). The results clearly prove that, higher levels of skin conductivity are found in fast auctions compared to low. This means that arousal is higher for fast auctions.

In the discussion between the participants at LARGE we discussed whether the variance between the slow and the fast auction could be explained by the fact that subjects might become bored when participating in the auction experiment. They might become bored, because they play against only one other participant and in the slow auction the maximum length is theoretically 16:40min.

Meeting Minutes - 14 January 2009

2009-01-15T10:06:00.002+01:00

Today I presented an overview of the priorly discussed articles on Dutch auctions. The following articles were discussed:

Van den Berg and Van der Klauw, 2008. A Structural Empirical Analysis of Dutch Flower Auctions
Katok and Kwasnica, 2008. Time is money: The effect of clock speed on seller’s revenue in Dutch auctions
Carare and Rothkopf, 2005. Slow Dutch Auctions

There was a discussion on what we can learn from the theoretic models as modeled in these articles when building a theoretic model for super fast auctions as in the DFA.

It was noted that the DFA has sequential auctions, which means that the cost of returning to the auction exists but not at the level of an individual auction.
It was noted that 'Competitive Arousal' as discussed in Katok and Kwasnica exists in a different form in super fast auctions, since it is unlikely that within an individual auction levels of arousal differ.
It was agreed that 'competitive arousal' in fast auctions could very well be a function of time, possibly not linear per se but this could be a good first start for building a model. Futhermore, the issue of arousal for experience was discussed because bidders who have been trading for years might not be influence very much by emotions.

Peter

Meeting Minutes - 7 January 2009

2009-01-07T18:07:00.003+01:00

Today I discussed the paper entitled "Slow Dutch Auctions" by Octavian Carare and Michael Rothkopf (Management Science, 51(3), 365-373, 2005). These are a few of the issues that were raised during the discussion:

Is it realistic to think that the results of Lucking-Reiley (1999) may be due to transaction costs? Aren't there other more plausible explanations (e.g., some form of bounded rationality)? Aren't transaction costs too low to have a significant effect?
Why do the authors assume randomly arriving bidders in their game-theoretic analysis? What are the implications of this assumption?
How would the analysis in the paper change if we take into account that there can be many auctions (simultaneously or sequentially) of similar objects? This may create competition among auctioneers. What effect can be expected from this?
Should we regard the results in the paper as surprising or not? Do they make sense intuitively? Can the analysis be generalized?
In what way can agent-based research contribute to our understanding of auction mechanisms? What could be the added value of agent-based research over game-theoretic and experimental research?

Ludo

Meeting Minutes - 18 December 2008

2008-12-18T16:11:00.003+01:00

Today, I (Meditya Wasesa)have presented a working paper entitled "Time is money: The Effect of Clock Speed on Seller's Revenue in Dutch Auction" written by Elena Katok and Anthony M. Kwasnica of Department of Economics of Smeal College of Business, Penn State University.

The paper presents an experiment and a formula that show the correlation between the clock speed and the sellers revenue in dutch auction. In brief, slower clock speed brings lower revenue, and faster clock speed brigs higher revenue. They have built a nice simple formula that explains this phenomenon by the use of the monitoring cost and the non monetary enjoyment, as the time affected parameters as important factors that affect the end revenue.

In the discussion, the group discussed troughly about the experiment settings that this paper utilized. We tried to analized the pros and the cons about the settings and the formulation which we want to extend at our proposed dutch flower auction experiment.

Meeting Minutes - 3 December 2008

2008-12-03T20:47:00.009+01:00

Today, I (Meditya Wasesa) have presented a working paper entitled “A Structural Empirical Analysis of Dutch Flower Auction” written by Gerard van der Berg and Bas van der Klaauw of Department of Economics of Free University Amsterdam.

The paper goals are to define the bidders valuation, determine the optimal seller's reserve value, and see the effect of reserve value adjustment (changing the value of current reserve value to the calculated reserve value) to the corresponding revenue. It is presumed that by adjusting the existing reserve value to an optimized value, a higher revenue could be gained.

The one that is interesting, the authors did not only observed the winning bids record, but also the losing bids (in an interval up to 1 second). Similar previous papers usually only consider the winning bids in their model. They believe that this extra observation could improve their prediction on the distribution of private values of the bidders. The other thing that the writers did is that they use a markov chain monte-carlo and gibbs sampling in the projection of the private values of the bidder.

In general, the flow of the research can be presented as the scheme above. First they observe the recorded data (winning and losing bids). Second they try to mimic the historical data to a valuation distribution of bidders by using the a markov chain monte-carlo and gibbs sampling (they do this step in 4 scenarios). Third then they calculate the optimal reserve bid and also the corresponding delta of revenue. By their finding the increment of the reserve price will not bring big change to the increase of the revenue.

However the conclusion of this paper is not final yet, the team planned to have another discussion about this paper in some other time to really grasp the essence of the paper.

Meeting Minutes – 26 November 2008

2008-11-28T17:59:00.003+01:00

Today, Paul R. Schrater a guest researcher from University Minnesota, has made a presentation about active preference learning topics. He explained a topic of probability model in which an agent can derive a continous valuation formula which is a result from learning algorithm from a set of discrete data. He tried to explained the algorithm which can decide what approximate formula can be presented to an individual in order to find the projected result that they value highly in as few trial as possible, without making an accurate model the entire valuation surface.

About the presenter,

Paul R. Schrater hold a joint faculty position at the University of Minnesota, in the departments of Psychology and Computer Science. His current research interests generally involve using probabilistic methods to study issues in perception and motor control. He received his Ph.D. from the Department of Neuroscience of the University of Pennsylvania, under David Knill, then of the Department of Psychology and Eero Simoncelli in the GRASP Laboratory who at that time had a primary appointment in the Department of Computer Science at Penn. His dissertation involved a psychophysical and ideal observer analysis of local motion processing.

Meeting Minutes - 19 November 2008

2008-11-20T11:33:00.001+01:00

Today, the LARGE session has had a very nice presentation from Carsten Block, a ready to finish doctoral candidate of Information & Market Engineering department, TU Karlsruhe Germany. At our session he presented the topic of "Market-Based Control and Agent-Based Trading in Combined and Power Grids." which is pretty much the topic which has been the area of his doctoral specialization.

The background of the research is the emerging trends of the energy supply which is no longer in hierarchical setup of up down (e.g. from the source to household) distribution, but also in the reversal direction. So that not only the source can sell their energy to the downstream but also the downstream line if they produce energy, the downstream costumer can also sell their energy to the upstream.

This issue of developing two ways of distribution is explained pretty much from different perspectives. But especially, since the energy transaction mechanism between the upstream and the downstream mechanism is built in the agent based point of view and also the mechanism of transaction is involving auction mechanism, so a lot of things what Carsten have applied in his research pretty much coherent with the LARGE field. This session off course has brought big lesson fur the group.

The discussion of this reversible energy transaction is pretty much done in multiple perspectives, since in this session a lot of people come from different area of specialization (e.g. economics, business network, information and decision sciences, etc). Carsten did an excellent explanation also to the multi-background viewer’s questions.

Since Carsten have a strong computer science background, he also shared some of his experiences (e.g. tool for development) in the development of his project. He introduced us a new tools for faster software development like grailsTM and make some demo also about what the platform advances. This session was very successful and everybody was enthusiastic.

Meeting Minutes - 12 November 2008

2008-11-12T19:32:00.003+01:00

Today Uzay Kaymak gave a presentation on the Ming Hua and Nicholas Jenning's paper of "Designing a Successful Trading Agent: A Fuzzy Set Aproach".

The paper tells us about the writers experience in implementing fuzzy logics in their SouthamptonTAC agent.

SouthhamptonTAC agent participated successfully in the first and second (TAC) Trading Agent Competition, a competition which facilitates the competition of the participant's "travel" agents in fulfilling their customer demand of travel package (flight ticket, hotel, and extra entertainment ticket). Those participating agents should compete with each other in different (flight ticket, hotel, and extra entertainment ticket) auctions provided by the TAC platform to statisfy the demand of the agent's costumer.

Uzay briefly explains how the SouthamptonTAC implements the fuzzy logics in the hotel, ticket, and entertaiment bidding actions. He (Uzay) found it quiet surprising how the performance of the fuzzy rules implementation can work well on the competition.

The presentation, is wrapped up by a small discussion about the competition conditions. Wolf gives the audience some description about the condition of the TAC competition (The rounds, the finalists, etc.).

Meeting Minutes - 15 October 2008

2008-10-20T21:41:00.002+02:00

Katalin led a discussion on the agent definitions and agent properties. She presented first a number of definitions from the literature mainly based on the article by Franklin and Gaesser. We compared and analyzed these definitions, especially w.r.t. to their properties and also discussed how these relate to the broad definition by Norvig and Russel.

It seems that there are three types of agents defined: “agents”, “autonomous agents” and “intelligent agents”. It is not always clear, however whether authors mean in fact agents with different properties when using these notions. Autonomy is a basic property that seemingly all agents should possess. However, we couldn’t really agree on what autonomy means. Explanations we gave varied from the notion of being “independent” to being “goal-oriented”, “pro-active”. The conclusion is that there are different degrees of autonomy.

With respect to the agents properties it also seems that learning and adaptation are sometimes interchanged.

In the second part of the meeting the CAS (Complex Adaptive System) definition has been focused, and agent-based modeling as an approach. Here the focus is on agents as interactive individual entities, and the emergent system properties. Agents are classified in three classes based on the rules they use and their level of adaptation: simple agents, complex-rule agents and advanced-rule agents.

Katalin concludes that we shouldn’t really bother about the many definitions; there are agents with different properties and different complexities. What should be important, however, is that authors give a clear description of what they call an “agent” in their paper, which properties the agents have, with definition of these properties.

Meeting Minutes - 21 July, 2008

2008-07-23T18:41:00.002+02:00

Today Peter Berends held a presentation discussing the case “The Application of Intelligent Personalised Agents for Buyer Decision Empowerment at the Dutch Flower Auction (DFA)”.

The DFA consists of six individual auctions located throughout the Netherlands which each host a couple of clock auctions on which goods are auctioned. In total there are 39 clocks on the DFA.

When buying through a clock auction, buyers aim to buy at the lowest price. Thus, they try to show their interest at the very latest possible moment. However, caution needs to be exercised because reacting too late means forgoing the ability to buy the auctioned product because other buyers might have jumped on the opportunity. As a consequence, buying through clock auctions is not an easy job. There is also a remote application through which buyers can buy at the DFA from any location. Many buyers procure for their customers who are located at different locations, and there are six different auctions of the DFA in the Netherlands. This means that when remotely buying, buyers could optimise the transportation cost and transportation time by making sure they buy products from an auction that is in close proximity to the location where they need to ship the goods.

This means that there are four main decision parameters to be considered with every buy: (1) price; (2) quality measures; (3) transportation costs; and (4) transportation time. There was a discussion on how intelligent agent-based systems could be utilized to empower buyers in their decision making.

After the presentation a discussion was held about where best to apply the agents, their benefits, and the information the agents need to base its decision on.

Meeting Minutes - 23 June, 2008

2008-06-24T09:45:00.004+02:00

Today Wolf gave a presentation on the ICEC '08 paper "A Semantic Web Architecture for Advocate Agents to Determine Preferences and Facilitate Decision Making". The paper deals with the architecture of a new kind of autonomous agents.

The main motivation of applying these personalized agents is that they can complement the cognitive limitations of the human mind, and therefore facilitate the decision making process to, reduce information overload (bounded rationality), increase work efficiency (i.e. speed up real-time managerial decisions), increase productivity (cost savings and ROI), increase solution (product or service) quality. Besides these tangible benefits, there are also intangible benefits, e.g. greater customer and employee satisfaction. In order to do this, these agents need to work effectively and efficiently with the human user. Meaning that the agent must learn the human user's interests, habits and preferences (as well as those of their communities). In an online retail example, recommendations can be given as to what to buy (product-brokering) and from whom to buy (merchant-brokering), based on customer criteria.

Agents and the human work in a bi-directional way through the interface called: Economic Dashboard.
"You cannot manage what you do not measure"
"What gets watched, gets done."

These statement demonstrate what the Economic Dashboard is, an "Organizational Magnifying Glass" – to focus the work of employees so everyone is going in the same direction! It business people: (1) Monitor, (2) Analyze, (3) Manage, (4) and Communicate and give feedback to the agent.

In order to work with the Economic Dashboard at all of the different organizational levels, these Economic dashboard has three types that relate to Business Intelligence:
Strategic BI: Achieve long-term organizational goals
Tactical BI: Conduct short-term analysis to achieve strategic goals
Operational BI: Provide a decision-making environment that reduces the latency between the time a significant business event happens and the business' ability to react to it.

In order to bring these personalized results, and work with the personalized results in the Economic Dashboard preferences are elicitated. Preference elicitations is the central concept of decision making and is fundamental for the analysis of human choice behavior, since people have different preferences for different roles. There are four methods or preference elicitation: (1) Questionnaire, which define roles, areas, objectives, and tasks; (2) Implicit feedback through user observation through browser extension (Piggy Bank, etc.), (3) Explicit user feedback through economic dashboard, and none intrusive sidebar in browser window, and (4) Business and Social Networks (Professional (intra company e.g. IBM, Linkedin, Plaxo, etc.) Personal (Facebook, Hi5, Hyves, etc.).

These preferences are saved in RDF stores, which allows the best abilities to apply Semantic Web agents.

In conclusion, this paper demonstrates the feasibility of Advocate Agents by presenting an architecture that integrates current technologies, such as Enterprise Service bus, XML, RDF, and machine learning techniques into a unique system and demonstrating that all the components of Advocate Agents can be built from already existing methods and elements.

After the presentation a discussion was held.

Next LARGE meeting is scheduled for 21 July.

Meeting Minutes - 16 June, 2008

2008-06-21T11:40:00.003+02:00

Elfriede Krauth presented a paper (together with Wolf Ketter and Jacqueline Bloemhof ) on Business Intelligence: Academic vs. Industry Perspective. Dashboards are distinguishing features between the disciplines.

Meeting Minutes - 9 June 2008

2008-06-09T19:28:00.002+02:00

Today, Ludo presented his research under the title ‘An Algorithm for Calculating the Long-Run Behavior of Genetic Algorithms in Economic Modeling’. Ludo first gave a brief overview of the research topic with which he is concerned. This is the topic of economic modeling using genetic algorithms. Ludo then discussed, at an informal level, the theoretical results that he has obtained. From these results, an algorithm can be derived that makes it possible to calculate the long-run behavior of genetic algorithms in economic modeling. Finally, Ludo discussed the application of this algorithm to a frequently cited study by Robert Axelrod (1987) on genetic algorithm modeling in iterated prisoner’s dilemmas.

Meeting Minutes - 2 June 2008

2008-06-02T17:31:00.005+02:00

Today, 2 June, Jordan presented a summary of three presentations given at the OMPL seminar on Agent-based control of Production and Logistics: theory applications and prospects. This seminar was originally hosted on 27 March at the university of Twente. Jordan provided an overview of the presentations given by Prof. Heragu and Prof. La Poutre as well as a summary of the MAS/"OR heuristics" comparison performed by Martijn Mes.

Meeting Minutes - May 26, 2008

2008-05-27T11:31:00.001+02:00

Today, Ludo presented the paper Robust Evolutionary Algorithm Design for Socio-Economic Simulation by Floortje Alkemade, Han La Poutré, and Hans Amman (Computational Economics, 28(4), 355–370, 2006). After a brief overview of the paper, the discussion focused on the main reasoning followed by the authors. Some objections against this reasoning were raised. We also discussed an error in the paper, focusing on what we can learn from the error and how such errors can be avoided as much as possible.

Meeting Minutes - 14 April, 2008

2008-04-17T20:18:00.003+02:00

During this meeting Katalin has summarized a working paper on a taxonomy proposed for artificial stock markets (ASMs). The taxonomy is aimed to give a guideline for analyzing, comparing, replicating and designing ASMs. The main factors proposed are extracted from the market microstructure literature, and representation aspects are added based on the literature on ASMs. Two main groups of factors are distinguished that are used to describe real and artificial stock markets: organizational aspects and behavioral aspects. The first group is related to market organization and it contains static, well-defined aspects, such as: the type of stocks, orders, participants, trading sessions, and exchange mechanism. The second group is related to the hardly observable and varying aspects of markets such as price formation and the behavior of the various market participants. An overview of 15 various ASMs has been given using the taxonomy. The variety of various trading strategies implemented within ASMs has been illustrated. An interesting discussion point is whether and to what degree are the represented traders in these ASMs "agents", and to what degree are they "autonomous".

Meeting Minutes - March 6, 2008

2008-03-08T11:07:00.001+01:00

During this meeting Ludo presented the paper Evolutionary Games and Spatial Chaos by Martin Nowak and Robert May (Nature, 359, 826-829, 1992). The authors introduce a simple evolutionary model for the emergence of cooperative behavior in prisoner’s dilemmas. Unlike previous models, the model introduced by the authors does not rely on reciprocity to explain cooperative behavior. Instead, the model assumes that agents are located on a two-dimensional grid and interact only with their neighbors. Evolution also takes place locally. It turns out that under these assumptions cooperative behavior can emerge for certain values of the prisoner’s dilemma payoffs. The contribution of the paper is that it provides the important insight that apart from reciprocity the spatial distribution of agents can help to sustain collusion. At the time the paper was published this was a new insight. The results of Nowak and May have been confirmed and generalized in a large number of studies. In addition to biologists, some economists have also been inspired by Nowak and May’s paper and have studied the implications in economic contexts.

Meeting Minutes - February 21, 2008

2008-02-26T21:35:00.002+01:00

At this meeting Milan made a presentation about his research, of which the basic idea is that the agent-based methodology (artificial financial markets), due to its flexibility and bottom-up-approach, are a suitable tool to test the predictions of behavioral finance (a sub discipline that combines financial theories with the insights from psychology and neuroscience in order to give a better description of individual investor behavior).

Milan gave a theoretical background of his research, discussing the components of the conceptual model of individual investor, such as risk attitude, time preference, motivation and goals, strategies, heuristics and biases, emotions, personality, demographics, and dual-process theories of human cognition.

The presentation finished with open questions regarding the actual implementation of the conceptual model.

Meeting Minutes - February 14, 2008

2008-02-26T20:13:00.005+01:00

During this meeting Milan presented a paper by Jong-Hwan Kim and Chi-Ho Lee: "Multi-objective evolutionary generation process for specific personalities of artificial creature," IEEE Computational Intelligence Magazine (2008).

The summary of the paper is the following:

Goal: creating a believable artificial creature based on a genome that contains its personality.
Personalities (the big 5 dimensions) are user-defined by assigning preference values for internal states and behaviors.
MOEGPP – multi-objective evolutionary algorithm for maintaining the population of genomes.
Result: the algorithm can create specific (user-defined) as well as diverse (in-between) personalities.
Verification: the algorithm is verified by checking that the personality characteristics are maintained on a different perception scenario than what is used for optimization.

Some conclusions/questions about the paper are:

The topic of creating believable artificial creatures was interesting, and seemed like a digression from agent-based papers discussed until now.
The paper is not self-contained, i.e. it is missing many implementation details (e.g. what is a mask, what are the activation values αk, what is the perception scenario for testing?), some of which are to be found in other cited papers.
The paper needs more introduction/explanation on why this particular algorithm has been used. Can we do it some other way?
Jordan wondered whether this algorithm could be used for other OR applications that require multi-objective optimization.
Uzay asked why the learning module has a direct input from the external environment, rather than through the perception module.
How can learning be added to these artificial creatures, and how would it be combined with their personalities?

I also found some information about Rity and one video here:
http://rit.kaist.ac.kr/home/Artificial_Creatures_%22Rity%22_and_%22Humanoid%22
http://rit.kaist.ac.kr/~ritlab/research/Artificial_Creatures/rity.wmv

MAS March Madness

2008-02-25T10:35:00.002+01:00

Good Morning Everybody,
I just received an invitation to attend a full day workshop on MAS for Production and logistics hosted at the University of Twente. The one catch is that this workshop is on 27 March - the same day Mark Hoogendoorn was supposed to talk to LARGE. So, here are my questions to the group:
1) Who would like to attend the event at the University of Twente?
2) Should we go as a group?
3) When would be a good day to host Mark if we all go to Enschede on the 27th?

Here is the workshop announcement to help you in answering these questions.

OMPL seminar:

Agent-based control of Production and Logistics: theory, applications, and prospects

March 27, 2008: 10:00-18:00
University of Twente, Enschede

With invited talks by:
Jos van Hillegersberg, University of Twente (NL)
Sunderesh Heragu, University of Louisville (USA)
Han La Poutré, CWI and Technische Universiteit Eindhoven (NL)

In recent years there has been a growing interest in distributed control within the production and logistics domain due to the necessity of greater adaptability and flexibility to changes in the market demand. Agent technology is considered an approach that holds high promises for developing such systems. Multi-agent systems (MAS) are believed to be particularly suited for decentralized systems in real-time and dynamic environments. Because problems are solved locally, these systems should (1) be able to deal with a high level of complexity, (2) require less information exchange than central control methods, (3) respond fast to unexpected events, and (4) reduce system nervousness compared to global optimization. In this seminar we provide a short introduction into agent-theory and present applications of agent-based control of production and logistics.

In the morning we have three invited talks; by Jos van Hillegersberg, Sunderesh Heragu, and Han La Poutré. Peter Schuur will facilitate the discussions. In the afternoon, Martijn Mes will defend his thesis on this topic.

Below we have included the program information. Full abstracts and biographies will be available at our website www.mb.utwente.nl/ompl/agent-seminar. Please register before March, 18th by sending an e-mail to m.r.k.mes@utwente.nl.

Program:
Date: March 27th, 2008, 10:00-18:00
Location: Building Logica (nr 65), University of Twente, Enschede

10:00 – 10:15 Welcome ( Peter Schuur)
10:15 – 10:45 Prof. Dr. Jos van Hillegersberg (University of Twente)
Multi-agent systems: introduction and applications
10:45 – 11:30 Prof. S.S. Heragu (University of Louisville)
Multi-agent systems: warehousing
11:30 – 12:00 Coffee break
12:00 – 12:45 Prof. dr. ir. Han La Poutré (CWI, Technische Universiteit Eindhoven)
Multi-agent systems: transportation
13:00 Lunch
14:45 Introductory talk in Room SP2 (Martijn Mes)
15:00 Public defense in Room SP2 (Martijn Mes)
Sequential Auctions for Full Truckload Allocation
16:30 – 18:00 Drinks and closing

The department OMPL looks forward to seeing you March 27th!

Martijn Mes

Meeting Minutes - February 7, 2008

2008-02-12T16:34:00.000+01:00

During this meeting Ludo presented Thomas Schelling's spatial proximity model from the book Micromotives and macrobehavior (Norton & Company, 1978), and demonstrated his own applet.
These are some of the questions raised by the participants of the discussion:

- Can we construct a set of individual rules which would lead to a wanted distribution?
- How do we construct complementary rules that lead from segregation to mixing?
- In the model there are only two groups of different preferences. What would happen if some heterogeneity is introduced within the group, e.g. some people are more tolerant of (or even demanding for) people of different kind around?
- What would change if an asynchronous simulation is used? Which implementation choices should be made (e.g. modeling race condition in one thread vs. multiple threads)? Which individual rules would then be used? Would it affect the final result of segregation?
- Are the main findings robust with respect to various implementation choices? For example, there are many ways for choosing dissatisfied people to move (e.g. randomly). According to Schelling these choices do not affect the segregation, but it should be checked in a more thorough way.

This is an interesting page where one can find a link for the implementation of the model in NetLogo (with source code). There is also a link to the extension of the Schelling model by sociologists Burch and Mare, which says (according to abstract) that Schelling’s results are not supported when other (more plausible) preferences are introduced:

http://hsd.soc.cornell.edu/Segregation.htm

This paper by Vinkovic and Kirman examines the shape of the segregated areas using the concept of surface tension force from the physics of liquid:
http://www.pnas.org/cgi/reprint/103/51/19261

New meeting time!

2008-02-02T17:29:00.000+01:00

Starting from this week the LARGE group has started to meet weekly on Thursday from 16:00 to 17:00. Please mark you calenders! This generates more opportunities for people to present their research, and the breaks are not that long if there is a holiday in between.

Looking forward to an exciting year with many great LARGE discussions!

- Wolf Ketter

Minutes, January 31, 2008

2008-02-02T17:23:00.000+01:00

Today Ruud presented the SugarScape simulation model of Epstein and Axtell.
Ruud briefly discussed the underlying philosophical motivation to the work of Epstein and Axtell. The notion of explanation turns out to be especially important. Epstein and Axtell call their approach generative rather than inductive or deductive, and they regard the generation of a given macrostructure from agent-interaction rules at the micro-level as a necessary condition for explanation of the macrostructure. We did not discuss these philosophical issues in much detail. Instead, we discussed the notion of heterogeneity of agents, which according to Epstein is a characteristic feature of agent-based computational models. This seemed to contradict Wolf's claim from two weeks ago that the heterogeneity of agents in his model is quite unique. We agreed that heterogeneity can exist at two
levels: heterogeneity of an agent's parameter values and heterogeneity of an agent's structure (i.e., the equations/algorithms inside an agent). Although these two kinds of heterogeneity are perhaps not fundamentally different, they may provide an explanation for the apparent contradiction between Wolf's claim and Epstein's paper. We further discussed Epstein's statement that any microspecification that generates a macrostructure of interest is a candidate explanation of that macrostructure. According to Epstein the choice between competing candidate explanations of the same macrostructure should be made by comparing the microspecification of each candidate explanation with the results from laboratory experiments. We doubted whether this is always possible. Uzay and Ruud suggested that a preference for simplicity may be another way to choose between competing candidate explanations. Ruud then showed us some simulations of the SugarScape model.
It was not always immediately clear why certain patterns emerged in the simulations, but at least the patterns usually looked quite intriguing. We realized that for each pattern there may be many different parameters on which the formation of the pattern is crucially dependent. Ruud also showed us simulation examples of something that may be regarded as emergent behavior. Ruud further mentioned that apart from economic research agent-based simulations are also useful in sociological research and perhaps in psychological research. At the end of the meeting the discussion moved from the SugarScape model to the spatial proximity model of Thomas Schelling. However, we decided to devote a separate meeting to the discussion of this model.

-- Ludo Waltman

Discussion Topics

2007-11-29T12:01:00.000+01:00

I'm interesting to continue the discussion of preference modeling, and maybe extending into the area of complexity economics, but I'm also open to new topics and suggestions. Please tell me what are you suggestions? Thanks, Wolf

CfP - TADA 2008

2007-11-27T14:51:00.000+01:00

Call for Papers for the 2008 Trading Agent Design and Analysis workshop. Note that the 14 July date is not yet completely certain - it may be one day earlier on 13 July, but we hope to nail that down very soon. Watch the website for updates: http://tac.cs.umn.edu/tada08

We hope to see you in Chicago next year!

Wolf Ketter, RSM Erasmus University

Alex Rogers, Southampton University

Norman Sadeh, Carnegie Mellon University

William Walsh, CombineNet

MAS course + article library

2007-11-23T14:23:00.000+01:00

The multi-agent systems research workshop at RSM will be held on January 21-23 and on March 10-11. A tentative schedule is available now. This can also serve as a start for categorized agent library for LARGE which we update each time we discuss new articles. It can also serve as a basis for discussion in the light of our next meeting on Dec 6th where Uzay will finish presenting the last paper and then we'll think about the future topics. Enjoy the articles! If you have questions please contact me. Wolf

Collection of PCA sites

2007-11-21T22:50:00.001+01:00

Production Differentitation PCA:
http://pricescan.com/ -- often biased to one brand in one particular category

Product Evaluation PCA:
http://www.bizrate.com/

Consumer Preference Identification PCA:
http://www.epinions.com/

Derivative PCA:

www.bestbuyfinder.com
http://dealtime.com/
http://www.shopping.com -- (rebranded from dealtime.com) close-coupled agent design that integrates different PCAs into its architecture -- while you start searching it gives you product suggestions, an advocate agent would give you products related to your preferences since it got to know you over time

An aggregate of the 3 main PCAs with social tagging and customer reviews
http://www.pronto.com/

Refine as you type comparison shopping agents:

http://www.smarter.com/
http://www.become.com/

Meeting minutes November 8, 2007

2007-11-12T09:51:00.000+01:00

We had a discussion on the paper "E-Commerce Product Recommendation Agents: Use, Characteristics, and Impact" by Bo Xiao and Izak Benbasat that appeared in MIS Quarterly, Vol. 31, No. 1, pp.137-209 (2007). Here are some of the discussion questions and remarks.

Can we use the findings of this paper to apply RAs in B2B environments, in particular in supply chains, or in another domain fit to our interests?
Can we use the findings of this paper to come up with general design principles of such RAs?
Do we appreciate the methodology of this paper, do we understand it? Do we need to?
Do we want to study this paper and (some of) its references in more detail, and what will be our objectives?
How would (virtual) social networks moderate the impact of and trust in RAs on consumer decision making; should we introduce the concept of multi-RA systems? What are requirements on the (virtual) social network here?
Important characteristic of an RA is the use of a model of user behavior (similarity). In this sense, descriptive theory on e.g. behavior of actors in financial markets may be a first step towards the design of RAs in that context.

Meeting Minutes - 25 October 2007

2007-10-26T15:54:00.000+02:00

In the LARGE meeting on 25 October, Jordan gave a dry run of her presentation for the upcoming INFORMS meeting. The presentation was entitle How "Good" are Multi-Agent Systems for the Dynamic Vehicle Routing Problem? Jordan began the presentation by defining the terms in the title, then motivated the topic by a review of the literature, described the methodology of her comparative analysis, and concluded with the results of the analysis.

Overall, the results indicate that multi-agent systems have the capacity to perform competitively in highly dynamic settings. The presentation as a whole could benefit by spending more time on the problem specification, definition of terms, and details of the methodology. Jordan acknowledged these comments and is presently working to add more graphics to the presentation to facilitate defining all the technical terms.

Following Jordan's presentation a brief discussion of the paper, Designing a Better Shopbot, by Montgomery, Hasanagar, Krishnan, and Clay, was held. In this discussion it was generally felt that:
1) The theory was interesting and the detailed description of the shopbot algorithm was computationally intriguing. However, the approached missed out on the psychology of on-line shopping. Thus, the usefulness of such a shopbot was called into question.
2) The product under analysis in this paper is books. As a result, the real power of the algorithm was never truly challenged. That is, the power of the algorithm lies in an ability to track and forecast prices in order to make 'intelligent' suggestions of books to purchase. Unfortunately, book prices do not fluctuate significantly; it would therefore be particularly interesting to apply this model to products with moe price variation - i.e. travel.
3) It was further noted that books do not have too many attributes. So again products, like travel or cell phones, with many many attributes would be a more interesting test of the agent framework.

The meeting concluded with a particularly brief mention of the big paper for next meeting's discussion - E-Commerce Product Recommendation Agents: Use, Characteristics, and Impact by Xiao and Benbasat. Ruud mentioned that he liked what he saw in this article because it seemed that the newer generation of shopping agents described therein incorporated more elements of behavioural decision making theory.

Finally, Wolf announced that the next meeting, 8 November 2007, would focus on the Xiao and Benbasat paper including a presentation by Rob Zuidwijk.

Welcome to the LARGE Blog

2007-10-12T15:59:00.000+02:00

The LARGE (Learning Agent Research Group at Erasmus) blog is a place where our team of researches can discuss all aspects of multi-agent systems and machine learning applied to business and economic environments.

Our homepage at the Rotterdam School of Management (RSM) at Erasmus University:

http://large.rsm.nl

If you want to participate in LARGE please email Wolf Ketter (wketter@rsm.nl).

- Wolf