Discovering Significant Persons, Locations and Organizations through Named Entity Ranking

In this paper, we propose a novel method based on the combination of Named Entity Recognition and Entity Rank algorithm for detecting key entities with significant influence and importance from huge sentiment data collected from Internet. Firstly, we extract entities from the target news websites and forums using a rule-based and CRF combined method. Secondly, we use the Entity Rank algorithm to calculate the hotness of entities extracted from the news and forums data. Finally, we validate the rationality of our algorithm by comparing our hot entities and current affairs. We believe this work will shed new lights on the online public sentiment supervision.