Sınırlı Alanlarda Konu Tespit ve Takibi Için Genişletilmiş Bir Mimari Yapı Önerisi

Abstract

This study investigates the effects of different retrieval functions and document representation techniques on performance in carrying out the tasks of story link detection and topic tracking. In this context, vector space and relevance models were used as retrieval functions. In addition, terms that scored the highest tf.idf values have been selected for document representation, performance tests have been repeated with these terms, and the most appropriate terms for each method have been identified. Moreover, the effects of choosing the appropriate threshold values for topic tracking on performance along with vector space, relevance model and k-means methods as retrieval functions have been examined.