The Community for Technology Leaders
Green Image
<p>Effective information customization systems must adjust their behavior to the user's task context. WordSieve, a text analysis algorithm, generates representations of users' topics of interest based on their browsing patterns. By finding terms associated with sequences of related documents, WordSieve learns topic-relevant keywords in real time with no predetermined corpus. You can use these keywords to form search engine queries to suggest relevant documents to the user. This article sketches the project goals, the WordSieve algorithm, and encouraging experimental results comparing WordSieve to TFIDF (term frequency inverse document frequency) and LSI (latent semantic indexing) at a precision-recall task.</p>
interest model capture and representation, task and situation modeling, interest modeling over multiple domains, information customization agents
David Leake, Travis Bauer, "Using Document Access Sequences to Recommend Customized Information", IEEE Intelligent Systems, vol. 17, no. , pp. 27-33, November/December 2002, doi:10.1109/MIS.2002.1134359
363 ms
(Ver 3.3 (11022016))