상세정보
학술대회
Building a Topic Hierarchy from a Data Stream for Conceptualizing User Interest Profiles
- 발행일
- 201210
- 출처
- International Conference on Information and Communication Technology Convergence (ICTC) 2012, pp.253-254
- 협약과제
- 12PR4600, 빅데이터 활용을 위한 지식 자산(Knowledge Base) 구축 및 실시간 Linked Data 응용기술 개발, 조기성
- 초록
- Recent information technologies have produced a lot of user- participating community services, or Social Network Services (SNSs), and have received much attention as a pipeline of user behavioral information. This paper explains the conceptualization of a user interest profile (UIP) through a topic hierarchy, which learns topics and keywords from a domain data stream. The conceptualization enriches a UIP, consisting of user interests modeled as terms and term-weight, by providing contextual information of the UIP. For this, the topic hierarchy extracts topic-deterministic keywords and their semantic associations with domain topics, and thus the user keywords are converted into domain topics representing the context of user behaviors. This paper describes an empirical method to build a news topic hierarchy and its use for conceptualizing a series of keywords appearing in user SNS messages. © 2012 IEEE.
- KSP 제안 키워드
- Community Service, Contextual information, Data stream, Empirical Methods, Information Technology, Interest profile, Social Network Service, Topic hierarchy, User behavior, User interest, social network(SN)