An Approach to Argumentation Context Mining from Dialogue History in an E-Market Scenario

Islam, K.S.

Argumentation allows agents to exchange additional information to argue about their beliefs and other mental attitudes during the negotiation process. Utterances and subsequent observations may differ during argumentation due to the gap in internal and external information with other agent. Contextual information is one reason of deviation between utterance and subsequent observations. Historic dialogues are a key source for extracting contextual information regarding illocutions, ontological category or semantically similar category. How historical dialogues contribute to contextual information during argument generation, selection and evaluation process is crucial to modeling the commonsense that human being apply in managing dialogues. Identifying, managing and augmenting contextual information and use that information in agent dialogue requires attention to several dimensions, e.g., illocution, interaction protocol, ontology, context, contract etc. which is an important problem in electronic market research area. This paper presents an approach for extraction of argumentation context from historical dialogues between intelligent agents in e-market. We are developing an argumentation system to extract context from historical dialogue and exploit context for dialogue moves between agents. An agent architecture using context monitor, context network, context miner is presented for argumentation context mining.

Cite as: Islam, K.S. (2007). An Approach to Argumentation Context Mining from Dialogue History in an E-Market Scenario. In Proc. 2nd International Workshop on Integrating Artificial Intelligence and Data Mining (AIDM 2007), Gold Coast, Queensland, Australia. CRPIT, 84. Ong, K.-L., Li, W. and Gao, J., Eds. ACS. 73-83.

(from crpit.com) (local if available)