Design And Implementation Of Text Categorization System Based On Data Of Baimi-wifi

Baimi-WIFI is one kind of free WIFI which deployed in some public areas in many cities of our country,users can access the internet after certified through a landing page,Baimi-WIFI can meet the need that users cannot connect free and stable WIFI in some public area.Baimi-WIFI will collect user's mobile browsing desensitization datas and on this basis to build the user interest preferences model accurately targeted mobile advertising.In order to build the user interest model,the method used here is to acquire the web page collection which user browsed in recent days based on the user browsing history,classify the web pages into certain types which defined in advance,analysis the type of page which user browsed mostly,then match background advertisings under the appropriate category,and accurate delivery.Text classification system can establish the user interest model based on the analysis of user mobile browsing behavior data,providing the basis for mobile advertising accurate delivery.This system mainly includes three parts: source data preprocessing module,text feature vectors module,text classification and performance evaluation module.Data preprocessing module handle the collected original data through data filtering and data fields conversion,convert the raw data into the text of the initial vector which can be handled by subsequent module.Text feature vectors module is used for text feature word selection and feature weighting calculations,convert the initial vector of the text into text feature vectors which can be handled by the text classifier directly.The feature word selection algorithm used here is chi-square,and the feature word weight calculation used here is New-TF-IDF,which is improved through the traditional TF-IDF algorithm.Text classification and performance evaluation module used for text classification,and verification of the availability of text categorization system through the experiment.The classification algorithm used here is RBF neural network,training text classification model according to some predefined categories,and predict the category of the classification of text.In this thesis,we first introduced the research background and significance of the paper and research status of text classification at home and abroad,described each process of the text classification and the key techniques and algorithms used in text classification in detail.And then describes each function modules,network structure,processing flow of the system starting from the overall design of the system.Finally we tested the text categorization system through the experimental data.Validate the function integrity of the text classification system and the effectiveness of the classification effect.