1 / 1

Automated Agents for the Provision of Arguments Ariel Rosenfeld and Sarit Kraus (BIU)

Intel Collaborative Research Institute Computational Intelligence. Automated Agents for the Provision of Arguments Ariel Rosenfeld and Sarit Kraus (BIU). How to enhance a discussion by providing arguments to a deliberant

faxon
Download Presentation

Automated Agents for the Provision of Arguments Ariel Rosenfeld and Sarit Kraus (BIU)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Intel Collaborative Research Institute Computational Intelligence Automated Agents for the Provision of Arguments Ariel Rosenfeld and Sarit Kraus (BIU) How to enhance a discussion by providing arguments to a deliberant • Two people deliberate over a controversial issue (such as the death penalty). • The agent provides its user with contextual and relevant arguments that “best suit him”. • The agent’s success depends on the user’s use of the proposed arguments and his general satisfaction. Research Focus Integrating Argumentation Theory with Machine Learning and Recommendation Systems building on deliberations collected from many people. • Theoretical Modeling • General Argumentation Framework (GAF) is used to represent all arguments on a given topic as well as the relations among them, the support of and attack on them. • Deliberations are a series of arguments, representing the dialog. Every deliberation is in fact a sub-set of a known GAF. • Not all arguments are of the same strength. The strength of an argument is captured by its groundness, meaning the ability to support and defend the argument if needed. • We introduce the notion of relevance, capturing the temporal and ideological  nature of a series of arguments by edge-distance in the GAF. Prediction Model • History of used arguments (by both sides) • Groundness • Relevance • Psychological aspects • Known/approximated proneness • Predicting Future Arguments among 4 Possible Options • Participants choose one of 4 possible arguments to follow an existing conversation. • 76% prediction rate when 5 previous selections are available. • Indifferent to Cultural Gaps Between Americans and Israelis. • Prediction in Transcribed Dialogs • Using transcribed phone calls on “Capital Punishment” and “Trial by Jury”, we trained and tested our prediction model. • Recommendation Policies • Is “plain prediction” an efficient recommendation policy? • How to maintain a good hit-rate while offering novel arguments. • Experimenting with on-line chats.

More Related