Microblog rumor recognizing method and system based on propagation structures

Abstract

The invention provides a microblog rumor recognizing method and system based on propagation structures. The microblog rumor recognizing method includes the steps that according to the machine learning method of a support vector machine, feature data of information features, user features and propagation features of microblogs serves as data, a graph kernel function is calculated with a random walking method, and a support vector machine model is accordingly built in cooperation with a classical radial basis kernel function and used for recognizing rumor; each microblog is represented with a propagation tree and a feature vector, wherein the propagation trees are used for representing the propagation structures in the propagation features, and the dimensions of the feature vectors are the information features, the user features and the propagation features of non-propagation structures; when the microblogs are recognized, the to-be-recognized microblogs are classified through the support vector model built with a model building module, and whether the to-be-recognized microblogs are rumor or not is judged according to the classification result. According to the microblog rumor recognizing method and system, due to comparison of the propagation structures of microblog information, the microblog rumor identifying accuracy and the microblog rumor recalling rate can be greatly increased.

Description

technical field [0001] The invention relates to database technology, in particular to a microblog rumor identification method and system based on propagation structure. Background technique [0002] After searching the prior art, the following relevant retrieval results are found. [0003] Related Search Results 1: [0004] Application (Patent) No.: 201210586904.8, Name: A method and device for identifying Internet rumors [0005] This patent document is applicable to the field of Internet communication, and provides a method for identifying network rumors. The method includes: analyzing network information in the database and extracting features; modeling with machine learning to generate a scoring function; using the scoring function Authenticating network information. This patent document analyzes the network information in the database and extracts features. By analyzing the types (topics) of network information, publishers of network information, disseminators of net...

AI Technical Summary

Problems solved by technology

The microblog features considered in this patent are relatively simple and simple, and manually setting the weights of each feature may easily lead to inaccurate identification results, low accuracy and recall rates

Images