Research On Book Recommendation Algorithm Based On Word Vector

With the rapid development of data science and the continuous improvement of software and hardware technology,the level of intelligence in all walks of life is getting higher and higher.The basic book retrieval and hot recommendation provided by traditional digital libraries can no longer meet the needs of readers.Personalized book recommendation services are becoming more and more popular,and their most significant significance is to enhance the reader experience and improve the utilization of book resources.The recommendation algorithm is the core of the book recommendation system.The traditional collaborative filtering algorithm faces the problem of sparse data and cold start.The content-based recommendation algorithm faces the problem of missing and unaccurate content of the book.For these problems,the paper proposes an improved strategy for word vectors.Firstly,based on the traditional data-based collaborative filtering algorithm,this paper proposes an improved algorithm based on word vector for the sparse problem of traditional algorithms in libraries with only borrowing records.The algorithm obtains the book vector through the word2 vec training borrowing record,calculates the book similarity based on the book word vector to realize the book recommendation,and verifies the validity of the word vector method in improving the coverage.In view of the cold start problem,the paper designs a cold start strategy,and the experiment proved that the accuracy of the recommendation results was improved under the condition of guarantee coverage.Secondly,in view of the lack of library books with only borrowing records,the paper proposes to model the obtained Douban book tags through the BTM theme model,and then integrates them with Douban book content to enrich the book information.Based on the traditional content-based recommendation algorithm,the word vector representation of the extracted keywords is carried out by using the word vector.The validity of the label to improve the recommendation accuracy rate and the advantage of doc2 vec in expressing the semantics of the book content are verified by experiments.Finally,based on the actual situation of Yanshan University Library,the paper applies the recommendation algorithm of word vector optimization to the actual recommendation system,and makes the basic recommendation system framework design.