Event identification in social media through latent dirichlet allocation and named entity recognition

Abstract

Nowadays people use social networks such as Facebook, Twitter, Orkut for sharing personal information and also significant events that occurs all over the world. Online news broadcast the most momentous events among users. Then the users discuss those events and post their reviews in micro blogs or in blogosphere. Since the web seems to be too huge and also the information available on the web is constantly updating, the task of identifying such events that are accessible on the web and the comments that are discussed related to those events in social networks is much daunting. Usually an event is temporal in nature, they change in time. Due to the temporal nature of events, the identification of event becomes even more difficult. Existing visual text analysis system extracts temporal themes from the large document stream and with that information; it provides temporal views of changes that occur on the event. Consequently, events that cause the changes can't be identified. In this paper, an Event Identification System is proposed to identify the most important events that occur and to identify the user discussion and also to rate their reviews. The proposed methodology make use of topic clustering, named entity recognition, latent dirichlet allocation, topic modeling to identify the significant events that are available on the web.

Mapping of topics in DESIDOC Journal of Library and Information Technology, India: a study

“…Topic-Modeling-Toolkit (Google code archive 2011a; MALLET 2018) is a graphical interface tool for LDA topic modeling which is powered by Java. "It is a simple GUIbased application for topic modeling that uses the popular MALLET toolkit for the back-end" (Abinaya and Winster 2014). "Topic models provide a simple way to analyze large volumes of unlabeled text.…”

A hybrid approach for generative process model with topic modelling towards efficient and dynamic document clustering

“…It was meant for discovering useful recommendations with optimized clustering process based on the given historical data. Abinaya and Winster [23] explored the importance of a technique known as "named entity recognition" and combined it with LDA. Their methodology is meant for discovering events from the social media text corpora.…”