A simple approach for Monolingual Event Tracking system in Bengali

Real-world applications have to take into consideration both topics and sentiments for precise opinion measurement. Topic and sentiment alignment is crucial for opinion retrieval, extraction, categorization, and aggregation on various issues. In this paper, we have reported a Monolingual Event (or, topic) tracking system for Bengali. The system has been developed based on a newspaper corpus developed from the Web archive of a leading Bengali newspaper. The goal of the system is to determine whether two news documents within a range of dates describe the same event. An event is a vector consisting of person, location, organization, title and date. A particular news document is described as a collection of such event vectors. A particular threshold value has been considered to check whether the number of event vectors of two separate news documents match at least by this threshold. Any particular news document of a date has been selected as the initial story. All the news documents within the preceding 15 and following 15 days have been considered as the target stories (or, documents). Evaluation results have demonstrated the Recall and Precision of 58.93% and 84.62%, respectively. The future works will look for interactions between topics and associated sentiments.