Time-aware relation extraction for entities using news headlines

Abstract

Online news media is a popular and effective source of information about everyday events. News headlines contain all the primary entities of the news article and can be used to identify relations among them. Since the relation between two entities can vary with time, we need to have a time-aware relation extraction mechanism for the entities. In this paper, we identify entities and temporal relations between entity-pairs using the set of keyphrases and other entities present in news headlines. This can be useful for developing various news analytics applications like ranking of news article, knowledge base population with the temporal relation, etc.