Research On The Prediction Of Mlm Features Based On Multi-Directional Feature Sets

Network information has become an important factor in today's social environment and network environment.With the large coverage of network data traffic and the new generation of network technology,illegal network data is also constantly invading the network environment.In recent years,as a kind of illegal fraud,MLM will be used as the main channel of communication in the future.Through the open network environment such as social platform or recruitment website,the idea of disseminating ideas through illegal network will cause serious problems for netizens and even the network environment.Cybersecurity threats.Therefore,the research and control of network marketing data is of great significance.Based on the existing theoretical techniques,this paper uses the automatic extraction data and data feature analysis method to predict the MLM data,as follows:Firstly,a digital proximity feature set extraction algorithm is proposed,which can not only expand the existing feature set library,but also serve as one of the factors for predicting the MLM pre-judging algorithm.Experiments show that the digital proximity feature set algorithm has the same characterization effect as the existing feature set,and the characterization effect is better for some background data.Secondly,a pyramid scheme pre-judgment algorithm based on multi-directional feature set is proposed.The algorithm uses multi-directional feature set as the prejudgment basis set.Based on the improved algorithm of word similarity,each text data feature set is abstracted by vector,and similarity calculation is performed with the remaining text data feature set.The result is used as a predictor.The feature intersection rate of the feature set to be tested and the multi-directional feature set is used as an experiment to predict the suspected pyramid scheme.The two indicators are combined and analyzed to predict the MLM data.The feature set to be tested is taken as the experimental object by the data to be tested,and is processed into the feature set to be tested by the above method.Experiments show that the pre-judgment basis for constructing multi-directional feature sets is more reliable than the single feature set,and the pre-judgment basis is more credible.Third,combined with the Selenium framework based on the Python development environment,applied to the proposed multi-directional feature set of network data and feature set extraction method.The framework principle is based on the existing "antiMLM website" as the research data source.Based on the Selenium framework and browser driver,browser operation scripts,etc.,it tends to extract the HTML data source of the research website in an automated mode,and then transfer the source data.Perform preprocessing operations on text data,such as word segmentation,stop word filtering,feature extraction,etc.,to obtain multi-directional feature sets,including the digital proximity feature set proposed in this paper.The feature set is used for experimental data of multi-directional feature set pre-judging algorithm.In this paper,the effectiveness and reliability of the algorithm are fully verified by the research of data feature correlation algorithm,feature set based pyramid prediction algorithm and Selenium code framework running through the algorithm.In the research process,the research background of the actual data of the network is combined to prove that the research direction and research results of the thesis are of academic value and age significance.