Work in progress
This page can be used to discuss work that is currently in progress, e.g. questions for interviews, experimental set-ups.
Planning Users Committee Meeting
Day: December 2nd 2009
Time: 16:00 - 20:00 incl diner
16:00: Aankomst met hapje en drankje
16:15: Introductie door Catholijn.
16:30: Presentatie Wietske: Automatisch redeneren over preferenties
16:50: Presentatie Alina: Meten van preferenties en acceptatie Pocket Negotiator.
17:10: Presentatie Joost: Emoties en onderhandelen
17:30: Presentatie Dmytro: Automatisch onderhandelen
19:00: Demo automatisch onderhandelen
19:45: Afsluiting, toekomst en koffie
Catholijn: Introductie en zo
Abstracts for the presentations (PLEASE FILL IN YOUR ABSTRACTS FOR YOUR PRESENTATIONS):
Wietske: argumentation + DUO models
In this presentation I will talk about two things: the conceptual model of negotiation that we built, and our argumentation-based approach to reasoning about preferences.
1. DUO model
A difficult issue in the development of negotiation support systems is how to extract qualitative data, e.g. from cases and interviews, and embed this knowledge in such a system. We present a meta-model for modelling the domain, user and opponent (DUO) in negotiation support systems. We focus on four main concepts in this meta-model - issues, preferences, interests and objective domain knowledge. We claim that (a) these concepts are essential in extracting data from unstructured sources, and (b) these concepts can be used as a formal basis for reasoning about user preferences and bids. To validate this claim, we first use the meta-model to structure data gathered with cases and interviews and analyse to what extent the meta-model helps us doing so, and second we show how parts of the meta-model can be formalized as a step towards a more computationally oriented model.
2. Reasoning about preferences
No intelligent decision support system functions even remotely without knowing the preferences of the user. A major problem is that the way average users think about and formulate their preferences does not match the utility-based quantitative frameworks currently used in decision support systems. For the average user qualitative models are a better fit. We present an argumentation-based framework for the modelling of and automated reasoning about multi-issue preferences of a qualitative nature. The framework presents preferences according to the lexicographic ordering that is well-understood by humans.
(Not part of the talk:) We also show how to reason about preferences when only incomplete information is available. An adequate strategy is proposed that allows reasoning with incomplete information and it is shown how to incorporate this strategy into the argumentation-based framework for modelling preferences.
Alina: social aspects for the PN + preference elicitation interfaces
The content of my presentation is two-fold:
1. Preference Elicitation
Through an extensive literature study we have gained important insides into the current state of the art in preference elicitation and we have learned that people rather construct preferences in the decision-making context than having them pre-defined in their minds. I designed and implemented 4 different preference elicitation interfaces combining existing methods with new ones. Each prototype is based on different concepts of user-system interaction and uses different input mechanisms. In an upcoming study we will investigate people's attitudes towards the different prototypes. We would like to see which concepts are generally more suited but also whether people with different characteristics prefer different methods and whether it is possible to form clusters of those.
2. Social Acceptance of the PN
Based on the results from several focus groups conducted at the beginning of the year we understood that social contexts play an important role in choosing whether to use the Pocket Negotiator and attitudes of people differ widely. To investigate this role we conducted an online survey over a period of 4 months with in total 120 respondents. The data collected in this survey showed (a) that subjective norm is an important factor influencing the intention to use the system and (b) that the acceptance of NSS depends on the use context. Therefore, we argue that the PN should be designed not merely as a tool being used in the actual negotiation but as social devices harnessing social networks to provide support in all negotiation phases.
Joost: emoties en onderhandelen
I will be talking about three things. First, I'll give a short overview of how emotions influence negotiation. Second I will be talking about a measurment tool we have developed to enable people to give feedback about their emotion. Third, I will explain a little on the next big studie we have planned, a study on a virtual negotiation training with artificial opponents able to explain thier negotiation behavior.
Dmytro: Designing Bidding Strategies
The central aim of this work is the design of generic and efficient automated strategies for two-party negotiations, in which negotiations parties do not reveal their preferences explicitly. Generic refers to the idea that the strategy needs no forehand knowledge about the opponent or the domain of negotiation. Efficiency refers to the strategy should be able to negotiate efficiently against another automated agent and a human negotiator. The design of the strategy is based on the analysis of the state-of-the-art negotiation strategies using an analytical method proposed in this work. The method significantly extends the existing negotiation benchmarks by analysing dynamic properties of a negotiation strategy. One of the main findings of the analysis is that the strategy should learn the opponent’s preferences in order to increase the negotiation efficiency. The strategy should be generic in a sense that it can be successfully applied to any negotiation domain and fine-tuned to the domain-specific features to produce even better results.
This work focuses on generic and efficient bidding strategies for single sessions bargaining between two negotiators. The bidding strategies can be used by software agents that negotiate on behalf of a user or to give bidding advice to a user during the negotiations. The stress on single session negotiations is motivated that various important negotiations in real life are of the single session type; e.g., buying a house or a car, negotiating for a job. From a technical point of view the restriction to single session bargaining implies that we cannot learn from previous experiences with the same opponent.
In this work we argue that the development of generic efficient bidding strategies requires an analytic framework for thorough evaluation of bidding strategies. For this purpose we developed the General Environment for Negotiation with Intelligent multi-purpose Usage Simulation (GENIUS) framework. We show that a proper analysis of negotiation strategies includes the dynamics of the negotiation instead of only studying the outcomes of negotiations as is typically done in the state of the art of automated negotiation. For this purpose we developed a range of dynamic properties that are proved their usefulness in our analyses and which are included in the analytical environment that is part of the GENIUS framework.
On http://jude.change-vision.com/jude-web/download/index.html you can register and download the JUDE software, or download the installer here http://www.joostbroekens.com/files/jude.exe . You will need it in order to edit the models we started to make for the DUO and the process. Everyone is asked to participate in the modeling process since this is a major part of the Pocket Negotiator system and important for all of us. The duo models are at the repository: https://email@example.com/svn/nego/Documents/DUOModels
Further, here is a list of Glossary on negotiation that we all should fill and edit in a wiki-way. It is very important we all do so.
Expert Interviews to obtain knowledge about DUO and their way of working:
Nederlandse versie van de vragen (versie 7 april 2009): Media:Vragen.pdf.