Dependence among terms in vector space model

• Published in: Proceedings. International Database Engineering and Applications Symposium, 2004. IDEAS '04 pp. 97 - 102
• Main Authors: Silva, I.R., Souza, J.N., Santos, K.S.
• Format: Conference Proceeding
• Language: English
• Published: Los Alamitos CA IEEE 2004; IEEE Computer Society
• Subjects: Applied sciences; Association rules; Computer science; control theory; systems; Data engineering; Data mining; Data processing. List processing. Character string processing; Exact sciences and technology; Information retrieval; Information systems. Data bases; Mathematical model; Memory organisation. Data processing; Proposals; Set theory; Software; Solid modeling; Thesauri; Data analysis; Correlation; Vector space; Database query; Hierarchical classification; Semantics; Association rule; Database; Information extraction; Data mining; Modeling
• DOI: 10.1109/IDEAS.2004.1319782
• ISBN: 9780769521688; 0769521681
• ISSN: 1098-8068

The vector space model is a mathematical-based model that represents terms, documents and queries by vectors and provides a ranking. In this model, the subspace of interest is formed by a set of pairwise orthogonal term vectors, indicating that terms are mutually independent. However, this is a simplification that doesn't correspond to the reality. Based on this scenery, we present, in this work, an extension to the vector space model to take into account the correlation between terms. In the proposed model, term vectors are rotated in space geometrically reflecting the dependence semantics among terms. We rotate terms based on a data mining technique called association rules. The retrieval effectiveness of the proposed model is evaluated and the results shows that our model improves in average precision, relative to the standard vector space model, for all collections evaluated, leading to a gain up to 31%.