Research On Rumor Information Source Discovery Technology In Social Networks Based On Representation Learning

In the time of big data,massive amounts of information were widely disseminated on social media,in which false inf-ormation was rampant.Previous studies have focused on the detection and dissemination of rumour information,but with the rapid development of the scale and number of social media,the data on social media is growing exponentially.In the face of the complexity and diversity of information content and the rapid growth of the number of information,the detection of a single piece of information is sometimes difficult to solve the source problem.Fast and efficient discovery of rumor information sources has become an effective means to solve this problem.In this paper,we mainly use the method of network representation learning to help us solve the discovery of social media rumor information sources.Firstly,a multi-layer attribute network representation learning method is proposed to solve the problem of discovering high quality rumor information sources on a single platform.This method takes into account the network topology and text information of the information sources,which is able to learn representation vectors with more information.Then,aiming at the problem of cross-platform rumor information source discovery,a representation learning method based on attention mechanism is proposed,which can take advantage of the synergy of similar information on different platforms.It can effectively improve the effect of cross-platforrm rumor source discovery.Finally,we propose a basic framework for simil ar high-quality rumor source discovery tasks,which can select the corresponding representation learning methods according to different tasks,and then input the learned representation vectors into the appropriate discriminant model to discover sources of similar rumors.Experiments on real data sets verify the effectiveness of the method and have important application value.