• Login
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Browse

    All of DSpaceCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

    My Account

    LoginRegister

    Statistics

    View Usage StatisticsView Google Analytics Statistics

    Cluster recommendation of scientific literature for research area analytics

    Thumbnail
    View/Open
    201611028 (1.128Mb)
    Date
    2018
    Author
    Anjana, Rita
    Metadata
    Show full item record
    Abstract
    Due to a fast-paced increase of the number of research articles published, it ishard for a researcher to find relevant articles. Research paper RecommendationSystems help in finding the relevant research article. Most Article RecommendationSystems are realized through clusteringIn this thesis, we mainly focus on clustering of documents which helps laterfor the recommendation of scholarly articles. When a user is searching for anypaper, he/she has a different kind of search purpose like "search for the paperwhich solves the same problem that a user wants to solve" or "the papers whichsolve the same problem with different approaches" etc. Here our aim is to groupall those papers that solve the same problem. Further in attempt to extract researchproblems from the article and attempt "problem-based" clustering. In thiswork, we explored various state-of-art clustering techniques for the said objective.We propose a hierarchical approach to cluster scientific papers which addressthe same problems. First we cluster research articles based on their abstracts.Within a topic, we further cluster them based on "problem" extracted from abstract.For first level clustering, we use topic modeling technique Latent Dirichletallocation.For second level clustering, we use Linguistic clues for extracting "problem" ofarticles. We try K-means algorithms with the different features. Also, merge twoclustering techniques K-means and K-nearest neighbors.We conducted experimental studies on DBLP. We show how the result of Kmeansalgorithm depends on features which are used to train algorithms. Comparethe results of different approaches.
    URI
    http://drsr.daiict.ac.in//handle/123456789/750
    Collections
    • M Tech Dissertations [820]

    Resource Centre copyright © 2006-2017 
    Contact Us | Send Feedback
    Theme by 
    Atmire NV
     

     


    Resource Centre copyright © 2006-2017 
    Contact Us | Send Feedback
    Theme by 
    Atmire NV