| As the mobile Internet began to lead the whole world to the way to access information, weibo and wechat as the representative of the mobile Internet software began to subvert the traditional way people read the news, such as television, newspapers, radio and other traditional media. The many advantages of mobile Internet(high portability, convenience, orientation, etc.) and the rapid development of mobile terminal makes mobile news reading for most people’s habits, but the demands of people who want to quickly and accurately obtain interest from the Internet and massive Internet news and information lead to contradictions. That is to say, people are faced with a serious problem of information overload, especially for education news interested users, such as educators, parents, students, etc., and they often have only fragmentation of time so they could not find news of interest from a large number of news reports. However, the existing information platform will rarely take education as the focus of the news.This paper optimized the design and implementation of the kapok news platform to solve people’s reading preferences for mobile client software but faced with massive information overload.Firstly, this article made optimized design for education news crawler and parser module. In view of the news source simplification of kapok education news platform, new and important data sources for news research- WeChat public articles were added, and Scrapy framework based on Python was used to crawl and parse WeChat public articles, in order to meet the needs of people in the field of education to read news by the mobile client software. On the other hand, in order to provide users with personalized news reading service, this paper make efficient and accurate crawling, parsing for news’ comments of the news portal based on Java language HttpClient client network library, which is the basic user behavior data for personalized recommendation. By using predefined json format as the entire system of communication between modules, this article realize the integration of crawling and parsing module by using different programming languages and solve the compatible problem while using Java and Python as crawler and parser solution, which improved system flexibility and scalability.Then, for overload massive news, the paper designs and implements a news recommendation modules, which integrated a hybrid recommend policy: LDA topic model and user collaborative filtering based on news topic, and the strategy take the following factors: multi-topic feature of news, the entity feature of news, the effect of user synergy, and the timeliness of news into consideration. Users and news were respectively modeled: the LDA constructed topic features, and mining core entity by TextRank algorithm to build entity features, and mining similar users based on the user’s similarity of topic interests, in order to understand the user’s actual information needs and interest in reading. By providing education news recommendation service, the system can improve information overload difficulties and reading experience of users who faced in the field of education news. This article also carried out experimental verification of recommended method based on user behavior(news comments).The deployment, operating results and performance of the optimized education news platform illustrate the rationality of optimize design and the integrity of system implementation. |
PDF Full Text Request