Topic summarisation on tweets

Due to the sheer volume of tweets generated by a micro blog site like Twitter, it is often difficult to summarize the required content of the user or the data analyst to evaluate the stream of Twitter data from tweets in million amount which contain enormous redundancy and amount of noise is large....

Full description

Saved in:
Bibliographic Details
Published in:2015 Seventh International Conference on Advanced Computing (ICoAC) pp. 1 - 6
Main Authors: Divyasree, T., Sujatha, P. Kola
Format: Conference Proceeding
Language:English
Published: IEEE 01.12.2015
Subjects:
Algorithm design and analysis
Big Data
Blogs
Clustering
Clustering algorithms
Decision support systems
Frequency measurement
Hadoop
Information technology
Summarisation
Twitter
Twitter Data
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Due to the sheer volume of tweets generated by a micro blog site like Twitter, it is often difficult to summarize the required content of the user or the data analyst to evaluate the stream of Twitter data from tweets in million amount which contain enormous redundancy and amount of noise is large. In an attempt to efficiently summarize the Twitter data and achieve better retrieval of the required topic, this research work focuses on the topic summarization on tweet streams, which produce better performance enhancement when processed in the distributed system. This work enhances the Weighted PageRank algorithm which considers both the Inlink and the Outlink value of the Tweets and Summarises the tweets efficiently and compare it with existing concept of similarity which was concerned with the mathematical cosine formulation. It also presents a timeline generation of tweets which enhance the summarization to analyses the tweet content according to the time. In order to enhance a special feature of Twitter posts, this work includes an effective analysis of performance results, that gives more improvement than the existing system. The experimental results on frequency measure when comparing it with the Weighted PageRank algorithm shows more efficiency than the existing cosine similarity computation.
DOI:10.1109/ICoAC.2015.7562805