Personalized Recommendation Algorithm of Literary Works Based on Annotated Corpus

Abstract

Literary work personalized recommendation service is a service that focuses on users’ needs, actively analyses users’ interests and hobbies, and intelligently and efficiently discovers users’ interesting information. Previous recommendation algorithms were unable to make effective and accurate recommendations in real time, resulting in poor recommendation outcomes. This study proposes a personalized recommendation algorithm for literary works based on the annotated corpus to address these issues. The dictionary is first used to mark the original text that was created by splitting words. The user’s reading behavior is then analyzed using a combination of individual personality characteristics and a set of factors that differ from the individual background, and personality characteristics and situations. Finally, we count the frequency of each modifier and word in the modifier vector in the corpus for each word, create the feature vector, and perform cluster analysis. The results show that this method’s MAE (mean absolute error) value is always lower than the traditional method, especially when the neighbor set size is 5, and that this method is clearly superior to the traditional method, with a maximum difference of 6.23%. Conclusions. The algorithm can produce satisfactory recommendation results and can be used to make personalized literary recommendations.