Cold-Start User-Based Weighted Collaborative Filtering for an Implicit Recommender System for Research Facilities

The developed recommender system recommends research facilities at UNC-CH to researchers that are not necessarily at UNC and identified solely by their name and email-ID. The recommendations are ranked based on an analysis of the researcher's publication records as captured by Google Scholar. The researcher publications record is then compared to the publication records of the researcher at UNC-CH and their usage of the UNC core facilities. The proposed system is composed of two modules: (1) Publication Profile Construction (PPC) and (2) Recommendation Generation (RG). In the PPC, we perform the necessary web crawling and construction of the new researcher profile. We also identify the group of UNC researchers with most similar research publications using a variety of similarity measures. In RG, we incorporate the research facility usage data (implicit data) to compute the similarity between the facilities and the cross investigator list of researchers derived from the PPC. Based on these similarity scores, recommendations are suggested. Therefore, this is an example solution of the cold start problem of figuring out how to suggest recommendations for a new researcher, who has not yet used any research facilities. We use leave-one-out cross-validation to test our method. Essentially, we remove the UNC researcher from our database and treat data as test dataset and train the model on the remaining dataset. We verify whether the recommended facilities match their actual usage.

[1]  Jonathan L. Herlocker,et al.  Clustering items for collaborative filtering , 1999 .

[2]  Sergei Vassilvitskii,et al.  Finding the Jaccard median , 2010, SODA '10.

[3]  Scott Sanner,et al.  Social collaborative filtering for cold-start recommendations , 2014, RecSys '14.

[4]  Marwan Bikdash,et al.  Modeling the effect of atmospheric attenuation on a propagation outdoor noise , 2016, SoutheastCon 2016.

[5]  Mohamed Nadif,et al.  A dynamic collaborative filtering system via a weighted clustering approach , 2016, Neurocomputing.

[6]  Chun Chen,et al.  An exploration of improving collaborative recommender systems via user-item subgroups , 2012, WWW.

[7]  Enrique Herrera-Viedma,et al.  A multi-disciplinar recommender system to advice research resources in University Digital Libraries , 2009, Expert Syst. Appl..

[8]  Adam Prügel-Bennett,et al.  Novel centroid selection approaches for KMeans-clustering based recommender systems , 2015, Inf. Sci..

[9]  John Riedl,et al.  Recommender Systems for Large-scale E-Commerce : Scalable Neighborhood Formation Using Clustering , 2002 .

[10]  Arnd Kohrs,et al.  Clustering for collaborative filtering applications , 1999 .

[11]  Nikolaos Polatidis,et al.  A dynamic multi-level collaborative filtering method for improved recommendations , 2017, Comput. Stand. Interfaces.

[12]  Wei Jiang,et al.  Secure Multiset Intersection Cardinality and its Application to Jaccard Coefficient , 2016, IEEE Transactions on Dependable and Secure Computing.

[13]  Gediminas Adomavicius,et al.  Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions , 2005, IEEE Transactions on Knowledge and Data Engineering.

[14]  Marwan Bikdash,et al.  Simulating the effect of soft undulating surface on outdoor noise propagation using equivalent sources , 2015, SoutheastCon 2015.

[15]  Shie-Jue Lee,et al.  A clustering based approach to improving the efficiency of collaborative filtering recommendation , 2016, Electron. Commer. Res. Appl..

[16]  Marwan Bikdash,et al.  An Extensible Recommendation System for Health Research , 2017, ACM Southeast Regional Conference.

[17]  Enrique Herrera-Viedma,et al.  A recommender system for research resources based on fuzzy linguistic modeling , 2009, Expert Syst. Appl..

[18]  Yuhua Qian,et al.  Multi-level preference regression for cold-start recommendations , 2018, Int. J. Mach. Learn. Cybern..

[19]  Tengke Xiong,et al.  Combining Collaborative Filtering and Clustering for Implicit Recommender System , 2013, 2013 IEEE 27th International Conference on Advanced Information Networking and Applications (AINA).

[20]  Marwan Bikdash,et al.  Line-graph based modeling for assessing the vulnerability of transmission lines , 2017, SoutheastCon 2017.

[21]  John G. Breslin,et al.  Computing the Semantic Similarity of Resources in DBpedia for Recommendation Purposes , 2015, JIST.

[22]  John Riedl,et al.  Item-based collaborative filtering recommendation algorithms , 2001, WWW '01.

[23]  Thomas Werner,et al.  LitMiner and WikiGene: identifying problem-related key players of gene regulation using publication abstracts , 2005, Nucleic Acids Res..

[24]  Christopher Meek,et al.  A unified approach to building hybrid recommender systems , 2009, RecSys '09.

[25]  Matthias Jarke,et al.  A Clustering Approach for Collaborative Filtering Recommendation Using Social Network Analysis , 2011, J. Univers. Comput. Sci..