Tweet Segmentation and Classification for Rumor Identification using KNN Approach

Abstract

Twitter is a source of sharing and communicate recent information, ensuing into huge size of records produces every day. Even though, a various applications of Natural Language Processing and Information Retrieval go through rigorously from an erroneous and tiny nature of tweets. We thought to implement a framework in support of segmentation of tweet by collection form, called as HybridSeg. During tweet separating with trivial segments, surroundings information is preserved and simply takes out by the downstream application. HybridSeg glance for top segmentation of a tweet through increasing stickiness score of its candidate segment. The stickiness score is explanation the possibility of a segment is express in English (global context and local context). Finally we advise and assess two models to acquire with local context by concerning the term-dependency in a collection of tweets, in the same way. Testing on two tweet data sets give you an idea about tweet segmentation superiority is considerably enhanced by global and local contexts evaluate by use of global context simply. Assessment and relationship, we demonstrate that additional correctness is accomplished in Named Entity Recognition by part-of-speech (POS).