Towards topics-based, semantics-assisted news search

Abstract

Identifying upcoming topics from a news stream is a challenging and time consuming task for editors since they have to recognize proper keywords, actively search with them, and need to browse the located media assets. To this end, our goal is to enhance an existing newsroom environment to automatically detect upcoming global and regional topics which are suggested for editors further work. To understand the impact of a topic, we provide its evolution over the time and the relations to other subjects as helpful indicators. To achieve our goals, we designed and prototypically implemented an automatic, semantics-based workow which heavily relies on non-ambiguous named entities extracted from the media assets. Further, we discuss the challenges encountered and point to proper solutions for building your own enterprise-scaled semantics-based application. Copyright © 2013 ACM.