Temporal Analysis on Pairs of Classified Index Terms of Literature Databases

Trend analysis and anomaly detection is gaining more and more interest since more than a decade. It gains focus in the new era of web2.0. As part of the project Knowledge Discovery in Scientific Literature, we developed new promising perspectives to detect trend and other interesting temporal patterns of index terms in the literature databases of educational domain. More specifically, we assign categories to index terms and investigate the index term pairs of special interest. We designed several measures to capture the characteristics of the evolution of individual index terms and pairs of index terms. Result shows our methodology is effective to find interesting temporal patterns, e.g. dependency relationship between index terms and helpful to detect trend