Inproceedings,

Topic Identification: Framework and Application

, and .
Proc of International Conference on Knowledge Management (I-KNOW, (2004)

Abstract

Abstract: This paper is on topic identification, i. e., the construction of useful labels for sets of documents. Topic identification is essential in connection within categorizing search applications, where several sets of documents are delivered and an expressive description for each category must be constructed on the fly. The contributions of this paper are threefold. (1) It presents a framework to formally specify the topic identification problem along with its desired properties, (2) it introduces a classification scheme for topic identification algorithms and outlines the respective algorithm of the AIsearch meta search engine, (3) it proposes a hybrid approach to topic identification, which relies on classification knowledge of existing ontologies.

Tags

Users

  • @huiyangsfsu

Comments and Reviews