HYBRIDSEG: CLUSTERING OF TWEETS AND IT’S SEGMENTATION

Abstract

Twitter is having lots of users to allocate and distribute a large amount of recent information, Various submission in Information Retrieval (IR) and Natural Language Processing (NLP) undergo harshly through the deafening and tiny kind of tweets. We recommend tweet segmentation framework in a group, called HybridSeg. By dividing tweets with signi_cant segments, the background information is conserved and simply extract with the downstream applications. HybridSeg search the best segmentation of a tweet by increasing the addition of the stickiness score. Two tweet data sets is a experiment it show that tweet segmentation quality is extensively increased by learning both global as well as local contexts compared by using global context alone. Additional accuracy is able to named entity recognition by putting segment-based part-ofspeech (POS) tagging.