Research And Implementation Of Recommendation System Based On Resold Housing Data

With the continuous growth of information on the Internet,humans have been submerged in the data ocean.Large amounts of data is overwhelming,and there is a lot of valuable information among them.The recommendation system,as an information service technology,helps users to filter out valuable information through recommendation algorithms.In this thesis,the related technology is introduced into the resold house network platform.In view of the complex problems of the traditional resold house network platform,a recommender system based on the resold house data is designed.Through the study of the traditional collaborative filtering model,this thesis designs a cooperative k-nearest neighbor algorithm.Based on the algorithm,this thesis designs a collaborative clustering and cooperative k-nearest neighbor recommendation framework.The algorithm framework firstly clusters the data from both the row and column dimensions through the preprocessed data.In this way,the data can be divided into several blocks collaboratively through the intersection of rows and columns.Then,the data block of the predicted data is determined,and the collaborative k-nearest neighbor recommendation algorithm is used to predict and recommend the data in the corresponding data blocks.As the overall calculation of the recommender framework is large,it is not suitable to put all calculations online,this thesis divides all the calculations into two parts: off-line processing and online processing.The offline part mainly uses the collaborative clustering algorithms to block the original data set,while the online methods predict and recommend the data through the collaborative k-nearest neighbor algorithm.Finally,based on the collaborative clustering and collaborative k-nearest neighbor recommendation framework,this thesis implements a recommendation system.In view of the rapid response of user behavior to the recommendation system and the continuous updating of the information of the house and users,this thesis builds a MongoDB storage cluster environment to meet the query analysis requirements of the recommender system under the big data environment.