A semantic approach for news event search

Abstract

News event search is the task of discovering important news that persists over a time tracking of a specific person the place related to an event and searching for a list of events within particular time duration The task can provide informative summaries of an event with the help of knowledge extracted from the news articles Existing news event search engines search for news events with a single event field and are not able to track events with multiple fields Besides they are yet to incorporate semantic knowledge for both document processing and search Moreover it is difficult to scale up to large data sets and cater to different domains or languages newlineOur approach uses Universal Networking Language UNL knowledge to represent the news documents semantically Therefore the event context has been extracted semantically and the semantic link of an event with its associated parameters such as Time Place and Person has also been preserved As we consider document level event context the semantic link between the concepts across the sentences has been retrieved and the whole picture of the event can be expressed in a single event context graph which has any number of nodes and relations newline newline