Mining And Identifying Key Scholars And Relations In Big Academic Networks

A network is a typical expressive form of representing complex systems in terms of vertices and links,in which the pattern of interactions amongst components of the network is intricate.The network can be static that does not change over time or dynamic that evolves through time.The complication of network analysis is different under the new circumstance of network size explosive increasing.This dissertation introduces a new network science concept called big networks.Big networks are generally in large-scale with a complicated and higher-order inner structure.This dissertation addresses several primary problems surrounding communities in a big network.The problems include identifying the most influential or key nodes in a network,discovering scientific teams or communities with leaders,and detecting key subnetworks in a big network.These problems generally revolve around some challenges of the area,such as accuracy,optimization,and scalability to big real-world networks.Implementation of these problems demands sophisticated solution designs to manage big networks with complicated inner structures properly.However,there are little to no solutions that could be applicable to big networks;rather.existing methods work well mostly in small-scale networks.While working on big networks,it is vital to unfold the hidden structure of those networks as it can provide us with interesting and significant information about the networks we are dealing with.Besides,it enables us to comprehend the complicated nature of large-scale and complex networks.Meanwhile,most scientific studies have involved collaborative,team-based,and coauthorship approaches that lead to knowledge production and high-impact research outcomes in academia.However,the previous studies are lacking in terms of identifying their real influential and productive scholars.Nevertheless,investigating the structure of scientific teams with their leaders is equally essential as investigating their community structure.Additionally,subnetwork identification plays a significant role in analyzing,managing,and comprehending the structure and functions of big networks.On top of that,ranking subnetworks and identifying one as optimal is as significant.Hence.we start by exploring various big network models and applications,studying state-of-the-art solutions that consider the complicated nature of networks.Subsequently,several new algorithms are designed by taking into account the mentioned problems.Firstly,an author ranking method is developed that captures how influential each co-author is in multiauthored publications.The method considers citation attributes of publications and similarity among the publications.The method computes the co-author's contribution in a given paper using fractional counting metrics.Next,it computes the contextual similarity between the given paper and its co-cited papers.Lastly,the method ranks each co-author using the mathematically defined metric,called KeyS core,and discovers the 'key' author among the coauthors of the given paper.The method is validated by identifying the publications of "Chinese outstanding youth" winning authors,regardless of the author's position in the author list.The method accurately identifies key authors in comparison with other existing baseline methods.Secondly,an algorithm,named CLeader,is proposed that starts by initializing candidate leaders of a given co-authorship network.Consequently,a mathematical model is designed to identify active and productive authors as real leaders,considering the publication year of their articles in a given period.Then,the method iteratively discovers subnetworks by grouping authors with their closest leaders and identifying key members using DHRank.The experimental results indicate that the proposed algorithms outperform existing algorithms,and they are applicable in large-scale networks.Finally,a topic-based optimal subnetwork identification approach(TOSNet)is designed.This dissertation addressed the following issues based on various fundamental characteristics:1)How to discover topic-based subnetworks with a vigorous collaboration intensity?2)How to rank the discovered subnetworks and single out one optimal subnetwork?The performance of the proposed method against baseline methods has been evaluated by adopting the modularity measure and assessing the accuracy based on the size of the identified subnetworks.The experimental findings indicate that our approach shows excellent performance in identifying contextual subnetworks that maintain intensive collaboration amongst researchers for a particular research topic.The dissertation designs novel algorithms and mathematical metrics for the discussed research problems and presents the experimental results of the designed algorithms using big real-world academic datasets,which outperforms substantially on the accuracy,effectiveness,and performance of the baselines in several cases.