Research On Library Bibliographic Recommendation Based On Social Labeling

Network technology is extremely developed today,personalized recommendation has been very common in a variety of e-commerce websites and social media applications.Personalized recommendation technology can be used to help users find their interested things through the book similarity calculation or analysis of users' characteristics.However,book recommendation is an important means for the library to meet the needs of the readers.Book recommendation can be used to improve the library's resource retrieval efficiency,enhance the users' experience and improve the utilization of book resources.With the emergence of Web 2.0 and the explosive growth of Web content generated by network users,various social networking sites emerge endlessly.At the same time,socialized web sites' development is also very fast.As a new resource organization site and social platform,the tag as a special "Keyword" has becomed the preferred medium for conceptual organization and generalization of information in social media sites,which can help users tag and search for content.Social tags help the library in a more efficient manner promote its collection of resources,and it can be a good data source for the user's personalized recommendation.Therefore,this article takes the social tag as a starting point to explore its effectiveness in the library's bibliography recommendation.This paper focuses on the following three works:Firstly,the paper summarizes the origin and characteristics of the social tag and tagging system,and analyzes the social tag in the typical tagging system website-the douban website.In addition,the paper studies the frequency distribution and the content distribution features of the tag.And then,the paper compares book tags and university library bibliographic records to find if the tag can reveal the subject of the books.Secondly,the topic model algorithm is used to excavate the hidden topic of the collected book tag set after a series of preprocessing.The biterm topic model is innovated on the basis of the LDA topic model.The Gibbs sampling algorithm is used to predict the subject of the tag,and the subject-tag model of the tag set is generated.By analyzing the result of the obtained label-subject model,it is explained that the BTM topic model is feasible to reveal the tag subject.Thirdly,based on the modeling of the social tag topic,this paper proposes a library bibliographic recommendation method based on the expansion of the subject feature of the book tag.The topic modeling of the tag is carried out by using the biterm model to excavate its hidden subject feature.The feature that reveals the subject feature can be used as the characteristic item which reveals the attribute of the book,and the feature of the library bibliography is supplemented.And then,the similarity between the book and the book is calculated by the similarity degree,and the recommendation is made according to the similarity.The finally,the experiment is carried out to prove the book's social tag as a bibliographic feature before and after the expansion of the recommended results.