Dealing with Context Ambiguity in Context-Based Information Re-finding

Tangjian Deng, Liang Zhao, and Ling Feng

Abstract. Inspired by human memory that context can often play as important recall cues, access context under which information is previously accessed is being exploited to assist users in information re-finding, e.g., allow users to search their accessed web pages by context. Due to the nature that human memory will decay and could be impacted by interference which will lead to context misremembering, users may refer to unreliable contextual cues in context-based re-finding. In this paper, we focus on how to re-find users’ desired results under the circumstance of users’ misremembering of precise contextual retrieval cues. To this end, we first categorize three kinds of ambiguity in context-based information recall queries, which are context degradation, context confusion, and context error, and organize user’s access context in associated probabilistic context trees. We then propose an approximate matching approach to deal with users’ re-finding requests (formed as contextual keywords) carrying possible ambiguity by taking advantages of context associations, which are extracted guided by human memory’s decay and interference characteristics. Experimental results on both synthetic and real data confirm the effectiveness of our solution that can enable users to achieve a good performance in ambiguous context-based information re-finding.

Keywords: Ambiguous context, context association, information re-finding

LNCS 8185, p. 685 ff.

Full article in PDF | BibTeX

lncs@springer.com
© Springer-Verlag Berlin Heidelberg 2013