Research On Vector Alignment Algorithms For Real Estate Data

Sichuan Province issued a 100 million yuan real estate data survey contract in 2018,collecting real estate data from Chengdu,Meishan and other four cities.The data collection of contract acceptance units is entirely realized under the artificial line.However,there are many data sources in the real estate market.The integration of housing data has many difficulties,such as large amount of data,heterogeneous data sources,lack of data,and so on.Data pre-survey has created obstacles.In order to improve the efficiency of data acquisition,entrusted by the receiver of the real estate data survey contract,this paper proposes a vector alignment algorithm for real estate data based on particle swarm optimization to match different intermediary second-hand housing sources.The main work is as follows:1.Write Scrapy crawler to obtain the initial second-hand housing source data.We will crawl the second-hand housing data for data preprocessing,including data incomplete data completion,normalization of two intermediary second-hand housing data.2.The normalized weighted vector model of real estate data is proposed.Firstly,according to the diversity of property types,three types of data,namely numerical value,text and picture,are numerically modeled as data vectors of [0,1].Then,combined with the different attributes of real estate data vectors,the impact of different attributes on the similarity judgment of housing sources is different,a weighted real estate data vector model is formed.3.A vector alignment algorithm for real estate data based on particle swarm optimization is proposed.The vector composed of different property attributes is regarded as a particle individual.The improved particle swarm optimization algorithm based on adaptive weight is used to optimize different property similarity weights.Finally,the thresholds of different property similarity weights and second-hand housing source similarity are obtained.The innovation of this paper is to construct the real estate data vector and its attribute model according to different data types according to the multi-source heterogeneity of the second-hand housing source data.The real estate data vector alignment based on particle swarm optimization is realized by entity alignment technology for the first time.Compared with several common classifiers and standard particle swarm optimization,the algorithm designed in this paper improves the alignment effect of real estate data vectors significantly.