Research On Domain Name Recommendation Engine With Multifactor Fusion

With the rapid development and widespread use of the Internet,domain name as an essential content of the Internet,plays a more important role in promotion of personal websites and enterprise brands.On the one hand,if a domain name is simple,easy to remember,and has a good meaning,it can be used as a collection.On the other hand,the domain name in accordance with enterprise brand and the business value of brand complement each other.Due to the uniqueness of domain names and the scarcity of good domain names,construction of good domain names has attracted more attention.In this paper,through analysis of Alexa Topmillion domain names,the domain name dictionary was built and the domain name evaluation model was established.Then,based on the domain name dictionary and the domain name evaluation model,multi-factor domain name recommendation was realized.At the same time,in order to enrich the domain name dictionary and recommend more diversified domain names,this paper mined new words and hot words from MicroBlog and added them to the domain name dictionary for recommendation.First,the discovery of new words and hot words was studied.Based on the statistical method,new words were mined from MicroBlog,and hot words were mined from MicroBlog by the combination of Newton cooling and Bayesian averaging.In this way,the content of the domain name dictionary was expanded.Second,the method of domain name word segmentation was studied,and the domain name dictionary was built.Based on the combination of dictionary matching and maximum probability model,the method of domain name word segmentation was realized.At the same time,unlisted words were extracted from domain name word segmentation results based on the combination of information entropy and maximum public string.And finally through the statistics of word frequency and other information of domain names,the domain name dictionary was constructed.Third,the domain name evaluation model was established.The random forest classification algorithm was used to construct the domain name evaluation model.The classification algorithm adopted nine character characteristics including domain name length,the top level domain name,as well as five characteristics of traffic information including Baidu search index of the words in domain names,and WHOIS information.Finally,the domain name recommendation model was established.This paper realized domain name recommendation from two aspects,including generation domain names from the domain name dictionary automatically and generation domain names based on user selection.In addition,it could provide users with domain name evaluation results and other information.This paper has completed the study of multi-factor fusion domain name recommendation.Experiments showed that the model established in this paper could meet the demand of recommending domain names for users.