A front-page news-selection algorithm based on topic modelling using raw text

Date
2015
Authors
Toroman, C.
Can, F.
Advisor
Supervisor
Co-Advisor
Co-Supervisor
Instructor
Source Title
Journal of Information Science
Print ISSN
0165-5515
Electronic ISSN
1741-6485
Publisher
Sage Publications Ltd.
Volume
41
Issue
5
Pages
676 - 685
Language
English
Type
Article
Journal Title
Journal ISSN
Volume Title
Series
Abstract

Front-page news selection is the task of finding important news articles in news aggregators. In this study, we examine news selection for public front pages using raw text, without any meta-attributes such as click counts. A novel algorithm is introduced by jointly considering the importance and diversity of selected news articles and the length of front pages. We estimate the importance of news, based on topic modelling, to provide the required diversity. Then we select important documents from important topics using a priority-based method that helps in fitting news content into the length of the front page. A user study is subsequently conducted to measure effectiveness and diversity, using our newly-generated annotation program. Annotation results show that up to seven of 10 news articles are important and up to nine of them are from different topics. Challenges in selecting public front-page news are addressed with an emphasis on future research.

Course
Other identifiers
Book Title
Keywords
Diversity, Document importance, Front page, LDA, News selection, Priority scheduling, Topic importance, Topic modelling
Citation
Published Version (Please cite this version)