Skip to main navigation Skip to search Skip to main content

Knowledge discovery across documents through concept chain queries

  • SUNY Buffalo

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

15 Scopus citations

Abstract

This paper focuses on detecting links between two concepts across text documents (e.g. two persons). We interpret such a query as finding the most meaningful evidence trail across documents that connect these two concepts. Here we propose a fast and efficient algorithm to perform this task. It is based on the idea of hypothesis generation originated by Swanson called "complementary structures in disjoint literatures" (CSD). We adapted the technique by (i) developing an alternate method of generating semantic profiles and (ii) extending the technique to generate concept chains. Counterterrorism corpus is used to evaluate the performance of this approach and demonstrates the effectiveness of our algorithm.

Original languageEnglish
Title of host publicationProceedings - ICDM Workshops 2006 - 6th IEEE International Conference on Data Mining - Workshops
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages448-452
Number of pages5
ISBN (Print)0769527027, 9780769527024
DOIs
StatePublished - 2006

Publication series

NameProceedings - IEEE International Conference on Data Mining, ICDM
ISSN (Print)1550-4786

Fingerprint

Dive into the research topics of 'Knowledge discovery across documents through concept chain queries'. Together they form a unique fingerprint.

Cite this