The Design And Implementation Of A Personalized Advertisement Recommendation System Based On The Interests Of Weibo Users

With the advent of the Web 2.0 era,a large number of emerging social network platforms have arisen at the historic moment.Sina Weibo has quickly attracted a large number of users because of its features of strong communicability and convenience and it's among the most popular social software now.From the perspective of Weibo users,Weibo is a platform for them to acquire and share information,finding the information they are most interested in is their purpose.From the perspective of Weibo business,Weibo is a platform where they compete with each other and run advertising campaigns.Accurate delivery of advertisement to Weibo users which makes profit maximization is the purpose of Weibo business.However,at present,Weibo platform has presented the status of "information overload".Therefore,how to extract valuable information from massive data to accurately locate users' interests and recommend personalized advertisements based on positioning results is the research this article focuses on.Aiming at the problem of inaccurate delivery of advertisement to Weibo users,this thesis firstly proposes an algorithm for discovery of similar users based on the interactive links,which aims to find a collection of similar users and lay the foundation for personalized recommendations.Based on this algorithm,we propose an algorithm to mine weibo users'interests,which uses the collection of similar users to expand the target user's attention set.Based on the information of users in extended target user's attention set,the analysis results of Weibo users' interests are more accurate.The main work of this thesis includes the following aspects:(1)Aiming at the problems that the scope for finding similar users is too limited and the neglection of indirect interaction between users,an algorithm for the discovery of similar users based on interactive links is proposed in this thesis.This algorithm uses the related users in Weibo interactive links to expand the candidate sets for finding similar users.It calculates the basic information similarity between users based on the four basic attributes of users,and takes full account of the indirect interaction between Weibo users in the interactive links.By integrating user's basic information similarity and strength of interaction,the final user similarity is obtained.Experiments prove the effectiveness and feasibility of the proposed algorithm in the discovery of similar users.(2)Aiming at the problem of inaccurate mining of users' interests in Weibo platform,this thesis proposes an algorithm for mining interests of Weibo users.This algorithm expands the target user's attention set with the set of similar users obtained by the algorithm for discovering similar users.Based on the authentication information and tag information of users in extended user attention set,the long-term interest of the target user is extracted.By analyzing the target user's own microblogging text,short-term interest of target user is extracted.By combining long-term interests and short-term interests of the Weibo users,the final interest of the user is obtained.The experiment proves the effectiveness and feasibility of the proposed algorithm of analyzing users' interests.(3)Based on the above two algorithms,we design and implement a personalized advertisement recommendation system based on Weibo user's interest.The system includes data acquisition,data processing,mining of user interests,.discovery of similar users and personalized advertisement recommendation modules.Based on the two algorithms proposed in this thesis,we can effectively recommend personalized ads to target users.