Research On User Portraits Construction Method For User Interests Of Campus Forum

With the development of the Internet,people began to conduct transactions and social activities online.More and more people know and understand people through network data.Therefore,how to understand users through network data has begun to attract attention.User portraits are just one of the ways to solve this problem.The user portrait contains many dimensions,of which the interest dimension is widely used in the fields of personalization and social interaction.In addition,the closeness and hierarchy of the college forum itself can help reduce the scope of the portrait and reduce the complexity of the process of constructing the portrait.Therefore,this paper studies the process of constructing the user portrait in the forum of interest in college.This article builds the user's interest dimension portrait from two aspects:the user's keyword interest and the weighted concept interest,the specific research work is as follows:In order to obtain a more accurate user keyword interest,this paper uses the emotional information in the text to reflect the characteristics of the user interest to a certain extent,adds the emotion analysis to the existing interest extraction method,and proposes a key based on the emotion word pair Word interest extraction method.This method first uses the theme of the forum to select the text features,and then uses the part-of-speech path template matching method to extract the emotional word pairs in the text.Finally,the keyword interest is obtained according to the emotion word pairs to obtain the final user interest.The experimental results show that compared with the interest obtained by using the content-based interest extraction method,the interest obtained by using the method proposed in this paper improves the average precision and obviously improves the classification accuracy.In the concept of user-weighted concepts,this paper studies two aspects.First of all,this paper adopts the method of hierarchical concept screening based on string matching to solve the problem of too large conceptual level.In this paper,we use the hierarchy of the forum and use the method of judging the literal similarity to screen the hierarchy concepts constructed by the Wikipedia category diagrams,so as to remove the redundant concepts and obtain the hierarchy concepts applicable to the university forums.After that,this paper uses the hierarchical classification to achieve the text-to-level conceptual matching.Based on the concept of text and the interest weight of text keywords,this paper proposes a method to calculate the user's conceptual interest weight.Compared with the existing methods,the concept of interest obtained by the method proposed in this paper has improved in average precision,and also has certain advantages in the classification accuracy.Finally,this paper merges the weighted concept interest and the weighted keyword interest into the user interest dimension portrait,which together with the user attribute portrait make up the final user portrait.This portrait contains two basic dimensions of object and interest,which can reflect more accurately the attributes and interests of the users it represents.