Optimal Sorting Of Information Search Based On Machine Learning

The growing material and cultural needs have promoted transformation of today's lives, and almost everyone has left some records or information more or less on the internet and has access to information from the internet. A variety of ways of obtaining information and forms of information result in a variety of user needs. This paper aims at meeting the user's actual needs. And amid information of matching users' requirement,we try to find the user's actual needs or help the users to get information, and increase the timeliness of accessing to information. For example,online shopping, users want to quickly find "Wumart, cheap, high quality" or"cost-effective" favorite goods, and hope to reduce the time consumption on the wait and see or comparison and contrast. This paper is to improve the users' experience of accessing to such a network of information.Search information of search engine does not take into account the actual needs of users, but the existing resource information achieves sorting with some of the more classic algorithm, and then presents it to the users; another way of popular information getting, recommended system make use of the user's behavior information, ignoring information of the user and the information itself. Although the recommendation is usually valid,but the validity is not high,and a lot of effective information has not been rationally used. To this end,new ways need to be explored to dig out the information as much as possible so as to increase the reliability of information. In addition, currently, it also improves the the blind use of information.In order to give full play to the advantages of search engine and recommendation system in this paper, we use the search engine and recommendation system combined with machine learning algorithm and data mining technology to improve the complexity of information acquisition, and obtain the following results: First, we present user state analysis which is combined with user nature attribute analysis and user behavior analysis of user characteristics. Unlike collaborative filtering,user state analysis does not require other people's information of other than the study object,and there is no need to worry about sparse data, and a feature dictionary is established to mitigate the mismatch of text mining so that it does not affect information matching.And then the information state analysis which is studying the nature attributes and give attributes,extracting the user status information corresponding to the keyword information. At the same time, we should also consider the cold start processing,which combines the static attribute data and collaborative filtering method, and then presents the processed information to the user. And then building the model system,do the overall analysis and sorting information. On the basis of the individual filling in the information and the behavior of all the information, it recommends a list of personal information,I@k.