Personalized academic literature recommendation method

Abstract

The invention discloses a personalized academic literature recommendation method. The method comprises the following steps: S1, data collection and cleaning, wherein paper data with papers and authorsas cores are collected, the paper data include paper titles, paper abstracts, author names, publication years, publication periodicals and reference literature, and data which have obvious format errors or miss data are cleared; S2, model establishment, wherein a process comprises the following steps: S2.1, constructing a training set, and S2.2, calculating features; S3, model training; and S4, academic literature recommendation, wherein a process includes the following steps: S4.1, establishing a candidate literature set, and requiring that publication time of cited literature selected at each step is earlier than publication time of a paper, and S4.2, carrying out prediction, and taking papers of k' highest probability values as lastly recommended reference literature. The method can more accurately and highly efficiently generate a reference literature list meeting user needs.

Description

technical field [0001] The invention relates to the fields of machine learning and data mining, and further provides a method for recommending references considering user preferences. Background technique [0002] Finding relevant and important references is an important way for researchers to understand the most cutting-edge research results in their field, and to obtain the latest research trends and development directions. [0003] Today, researchers still manually select papers that may be related to their current research field by using methods such as given topics and keywords in search engines such as Google Scholar or a specific database such as Web of Knowledge. The rapid growth of the number of published academic documents in recent years, coupled with the popularity of electronic publications and open databases, on the one hand, highlights the shortcomings of the current manual selection method, such as time-consuming, low accuracy, and mechanized operation. On t...

AI Technical Summary

Problems solved by technology

The rapid growth of the number of published academic documents in recent years, coupled with the popularity of electronic publications and open databases, on the one hand, highlights the shortcomings of the current manual selection method, such as time-consuming, low accuracy, and mechanized operation. On the one hand, the existence of a large amount of academic data also makes it possible to use various data-driven methods such as data mining to automatically generate reference lists

[0004] Existing literature retrieval and recommendation methods are often not fully functional and cannot produce satisfactory personalized recommendation effects. At the same time, there is also the problem of cold start, which cannot provide effective recommendations for users who lack sufficient information.

Images