Early Detection And Analysis Of Rumors In Social Network

Social network plays an important role in our daily life. With the rapid growth of the speed of network and the accessibility of devices that can be connected to the Internet, information on the Internet is becoming more and flexible. The problem that it brings out is the possibility of the rapid propagation of rumors while users on the Internet are not able to detect rumors by themselves and the platform itself can not detect rumors automatically and efficiently. So we designed an automatic detection framework to detect rumors in social network as early as possible and applied it to Sina platform.The main focus of our work is to analyze and detect rumors as early as possible when the keyword is given. The experiment consists of two parts, one for the detection of sub-topics which has the probability to be a rumor while the other for the analysis of the credibility of the sub-topics detected. The detailed solution is that once a keyword is given, we crawl data from Sina at regular intervals. After that we process the data using nlp tools for word segmentation and word vector training. We can get sub-topics by clustering method. During the next section, we can analyze the sub-topics to check whether they have the probability to be a rumor.In a word, the main contribution of our work is to detect rumors as soon as possible. We conducted experiments by data crawled from Sina Weibo and get excellent results from that.