Design And Implementation Of Search Recommendation System Based On Semantic Relevance And Users' Click Preference

In recent years,artificial intellligence technology has been continuously developed,and search recommendation systems have also been more widely used.In the era of information explosio,information in various fields is huge.Finding effective information in a short period of time has become a key issue.Although the search recommendation system in the general fields is becoming more and more perfect,the search recommendation system in the partial vertical fields needs to be improved.The data selected in this paper belongs to partial vertical domain content.This paper proposes a design scheme of search recommendation system based on semantic correlation and users' click preference.Then we implement and test for the system.The main content of the design scheme consists of three parts:(1)The first part is to propose a semantic correlation calculation method based on word vector weighted average and support vector regression for search text and content text.The search text entered by users is the prefix of the content text they are looking for.The search text and content text which to be recommended are semantically related.We construct text vectors by using word vector weighted average technique.Support vector regression algorithm is used to measure the semantic correlation between the search text and content text which to be recommended.(2)The second part is to propose a weight modified method based on transfer learning and users' click preference.In order to achieve the purpose of balancing semantic relevance and users' click preference,the semantic relevance model of the first part is modified by the users' click preference.Based on the user's click probability on different texts under the same search prefix,the click probability is used to represent the users' click preference,and the semantic relevance model is further modified based on the transfer learning and support vector regression.(3)Based on the proposed method,a search recommendation system based on semantic correlation and users' click preference is designed and implemented,and the system is tested.The results of system function test and system performance test show that the search recommendation model designed and developed in this paper has high accuracy and the performance test of the search recommendation system is good.