Title :
Improving Human-Agent Conversations by Accessing Contextual Knowledge from Wikipedia
Author_Institution :
Center of Excellence Cognitive Interaction Technol., Bielefeld Univ., Bielefeld, Germany
fDate :
Aug. 31 2010-Sept. 3 2010
Abstract :
In order to talk to each other meaningfully, conversational partners utilize different types of conversational knowledge. Due to the fact that speakers often use grammatically incomplete and incorrect sentences in spontaneous language, knowledge about conversational and terminological context turns out to be as much important in language understanding as traditional linguistic analysis. In the context of the KnowCIT project we want to improve human-agent conversations by connecting the agent to an adequate representation of such contextual knowledge drawn from the online encyclopedia Wikipedia. Thereby we make use of additional components provided by Wikipedia which goes beyond encyclopedical information to identify the current dialog topic and to implement human like look-up abilities.
Keywords :
Web sites; cooperative systems; interactive systems; linguistics; natural language processing; KnowCIT project; Wikipedia; contextual knowledge; conversational context; human-agent conversations; language understanding; linguistic analysis; terminological context; Context; Data mining; Electronic publishing; Encyclopedias; Humans; Internet; Contextual Knowledge; Dialog Topic; Embodied Conversational Agents; Human-Agent Conversation; Wikipedia;
Conference_Titel :
Web Intelligence and Intelligent Agent Technology (WI-IAT), 2010 IEEE/WIC/ACM International Conference on
Conference_Location :
Toronto, ON
Print_ISBN :
978-1-4244-8482-9
Electronic_ISBN :
978-0-7695-4191-4
DOI :
10.1109/WI-IAT.2010.231