A Method for Automatic Content Curation of News Events

Abstract

The read habit of readers have changed more and more readers use the computers or mobile devices to browse news and the news industry is also digitized Various news broadcaster published online news to pass information A large number of online news bring readers a variety of information but at the same time readers also need to spend more time digesting them When readers quering a news event it often returns a large number of search result which leads readers to spend extra effort to sort out In order to solve the problem some platforms apply the concept of Content Curation which aggregates the news articles based on event and then organize and present to readers At present most of Content Curation is manually organized Different from the way of the past platform this study proposes an automated method of news curation We first extract the topics from the dataset and use the word sequence to find out the topic sequence through the Hidden Markov Model Then calculate the strength and the variation to detect important time points during the development of the event Finally generate a concise summary to every time points We combine chronology and summary to design the curation and look forward to help readers to quickly grasp the context of the news event Experiments has found that the method has a good performance in each modules The curation result have good practicality for the readers But in terms of coherence there is slightly insufficient to improve

Date of Award	2019
Original language	English
Supervisor	Hei-Chia Wang (Supervisor)