Personalization of the collaborator recommendation system in multi‐layer scientific social networks: A case study of ResearchGate

Abstract

The development of knowledge sharing platforms, like scientific social networks encourages researchers to establish international collaboration in scientific projects.After reviewing the previous methods for collaborator detection in social networks, gaps in the earlier models are investigated, and the present study aims at filling the gaps by introducing a new scientific collaborator recommendation system. Accordingly, in the present paper, an integrated model is presented based on multilayer networks that can personalize the proposing scientific collaborators. Our proposed model involves various types of collaboration features based on researchers' needs.In our method a scientific social network is modelled as a multi-relational network (MRN), in which collaborators are determined by a community detection algorithm.It provides us with an approach to integrate the personalized features into the collaborator detection model that is due to the essence of semi-supervisory learning of our community detection algorithm. This MRN helps us prevent information loss in the network. We considered two techniques for examining the models. The first one was General Collaborator Recommendation, and the second one was a Collaborator with Personalization Ability. The proposed method is applied to the data of the ResearchGate (RG) social network and is evaluated by criteria, such as F index and NMI.

Design and Implementation of Attention-Based CR System in the Context of Big Data

“…The model can serve the purpose of preventing information loss in the network. Recommendation experiments were conducted using research data and the experimental results showed that the model has high accuracy in recommending research collaborators [9].…”

“…0,1 y  represents the click labels (0 means not clicked, 1 means clicked), j w is the j th vector, j n is the number of occurrences of the j th feature in the whole sample, and  is the canonicalised assignment weight coefficient. The judgement function in the improved regularization method is shown in equation (9).…”

Research on MOOC Curriculum Recommendation Model of Higher Vocational English based on Improved Intensive Learning Network

“…Therefore, the research team designed a recommendation model based on improved collaborative filtering algorithm. The test results show that the model can effectively improve the accuracy of content recommendation on the knowledge platform [8]. Zeng et al found that the way of extracting information from the user's history is widely used to define the user's fine-grained preference to build an interpretable recommendation system.…”

Developing smart city services using intent‐aware recommendation systems: A survey

“…Tu et al 177 propose a CF parallel algorithm to respond to the data sparsity, lack of scalability, and lower efficiency challenges in big data, using Apache Spark. An improved technique of normalized processing in ALS-WR (Alternating Intent-aware behavior extraction 168,[192][193][194][195][196] Intent-aware results optimization 31,[197][198][199][200][201][202][203][204] least squares with weighted regularization) is proposed to solve the problem of data filtering on large-scale clusters. 177 ALS-WR algorithm is incorporated into the Apache Spark parallel computing framework to enhance efficiency in terms of computation speed.…”

“…Bleize et al 223 propose a personalized recommender based on intentions extracted from social media and other information resources. Roozbahani et al 195 propose to personalize the scientific collaborators using multilayer networks. They model a scientific research social network to a multi‐relational network and integrate personalized features into the collaborator detection mode and recommend personalized research collaborators.…”