PISA: Pooling Information from Several Agents.

PISA : Research Abstract!!!!

the PISA (Pooling Information from Several Agents) Framework, is intended to allow any number of participants (presented by software agents) to engage in “Arguing from Experience” dialogues. This is particularly beneficial when there are more than two possible “views” (classifications, opinions, etc), since each possible “view” can then have its own champion. Here an application to enable the PISA process between any number of players (agents) is provided to download. For more details about PADUA please refer to the following papers: ArgMAS09, DAWAK09 and AI08.

PISA concerns dialogues where there is a range of options for classification, and each of the participants is the advocate of one of these options. Where there are more agents than opinions, the agents will act in groups, one for each opinion. Additionally, there will be one agent, the chairperson, who will not be the advocate of any position, but rather manage and facilitate communication between the clashing advocates. This style of dialogue thus determines the roles of its parties: a chairperson, and, for every option, at least one player acting as its advocate. Each group will act as one entity in the ongoing dialogue, thus masking the internal decision making process that is taking place between different players sharing the same objective. The dialogues are open, in a sense that participants may enter or leave when they wish. For turn taking, a structure with rounds is adopted, rather than a linear structure where a given agent is selected as the next speaker. In each round, any participant who can make a legal move may do so. The chairperson then updates a central argument structure, which will be termed the “Argumentation Tree”, and another round occurs. In every round there is a number of speakers (participants contributing in that round) and a number of addressees (participants which positions are under attack). The rest of the parties (which did not participate and are not attacked in the given round) need to be aware of the developments in the dialogue and are thus assumed to be passive listeners.

This page provides a downloadable implementation of the PISA Framework, along with an associated GUI interface. The implementation combines the various components described previously into a functional application, using the Java programming language. The objective of this implementation was to provide a tool to produce multiparty “Arguing from Experience” dialogues. The accompanying GUI enables the user to test and assess the resulting dialogues, and can also be used to examine the various components of PISA, such as the Argumentation Tree and the History Log. The implementation described here was used to evaluate a variety of test scenarios and examples discussed in the remainder of this chapter, and the next chapter. Some further specialised components for this application will be included in the next chapter. A more detailed description of the implemented PISA Application, and the accompanying design documentation can be found here.

PISA Application Executable JAR File is Avaliable Here.

PISA Java Source Files are Avaliable Here. (Requires NetBeans 6.5)

PISA Test Data Sets are Avaliable Here.

NEW PISA Application Executable JAR File is Avaliable Here. And its DATA SETS.

How to Run the PISA Application

By simply downloading the JAR file and run it on your machine (this requires jdk1.6.0_11). Alternately you may download the source files (the java classes for this application are available in one zip file downloadable from the above list) and import them using NetBeans 6.5 (the name of the project should be PISAInterface).

A number of datasets are available to test the application with, a zip files with these sets is downloadable from the above list. This file includes a number of folders with the datasets, such that ds_Px.num is player Px's share of the dataset ds, some of the included datasets were splitted into groups/players hirarchy. Each folder also contains a ds.gmd, which is the game dictionary file for the given dataset (ds). The user should upload this file and the players dataset using the PISA Application and run the game according to the explanation given in this web page (a simple user manual can be found here).