Scientific research cooperation relationship prediction method based on heterogeneous information network

Abstract

The invention discloses a scientific research cooperation relationship prediction method based on a heterogeneous information network. The method comprises the following steps: 1, converting a scientific and technological literature heterogeneous network into an author-author isomorphic network; 2, performing vector representation of author nodes in the isomorphic network; and 3, calculating the similarity between authors. The method is applied to predict the possibility of future cooperation of any two authors who do not cooperate yet, so that the problems of information loss and simplification in an author-author isomorphic network are effectively solved, the accuracy of cooperation prediction is improved, and scholars are helped to carry out scientific research more efficiently.

Description

technical field [0001] The invention belongs to the technical field of information retrieval, and in particular relates to a method for predicting scientific research cooperation relationships based on heterogeneous information networks. Background technique [0002] In recent years, with the diversification and complexity of research problems, multidisciplinary integration has become more and more common. At the same time, changes in scholars' research directions have also led to an increase in the number of author collaborations in various fields. Finding the most valuable collaborators from the vast academic digital library is a great challenge. This makes the prediction of scientific research partnerships more and more important. [0003] At present, the existing scientific research cooperation relationship prediction mainly adopts similarity-based methods and machine learning-based methods. The similarity-based methods mainly focus on text similarity and structural sim...

AI Technical Summary

Problems solved by technology

Among them, the literature [The link-prediction problem for social network, 2007] judges the similarity between authors by comparing the node topology similarity index in several networks in the co-authored network, including common neighbors, Jaccard, SimRank and Katz, etc. Through experimental analysis, the Katz index is superior to several other similarity calculations. The method based on machine learning mainly uses the classification model, especially the idea of ​​binary classification, to predict the cooperation between authors. The literature [Predicting co-author relationship in medical co-authorship network, 2014 ] Treat scientific research cooperation prediction as a binary classification problem, use the structural similarity index as a feature in the co-authorship network, and then train the model, and use logistic regression and SVM as the prediction model to predict the links between author nodes, but the above methods are not It is based on a homogeneous network. Although the calculation is simple, it loses rich semantic information, which may lead to errors in prediction results.

[0004] In addition, there are also a small number of studies based on heterogeneous networks. Among them, the literature [Co-author relationship prediction in heterogeneous bibliographic network, 2011] took the lead in applying meta-path-based topological features to heterogeneous networks and using logistic regression models to predict cooperative relationships. , but the computational complexity of several topological features used in this method is relatively high

Images