A concept driven graph based approach for estimating the focus time of a document

Abstract

Many text documents are temporal in nature, i.e., the contents of the document can be mapped to a specific time period. For example, a news article about the Kargil War can be mapped to the year 1999. Identifying this time period associated with the document can be useful for various downstream applications such as document reasoning, temporal information retrieval, etc. In this work, we propose a graph based approach for estimating the focus time of a document. The idea is to treat documents and years as nodes which are connected by intermediate Wikipedia concepts related to them. The focus year of a document can then be identified as the year which has the maximum influence over the document computed using the flow between the year node and the document node through all intermediate Wikipedia concept nodes. We evaluate our approach on two different datasets which were curated as a part of this work and show that our approach outperforms a state of the art method for estimating document focus time.