George Karypis has research interests that span the areas of data mining, bioinformatics, cheminformatics, high performance computing, information retrieval, collaborative filtering, and scientific computing. His research has resulted in the development of software libraries for serial and parallel graph partitioning (METIS and ParMETIS), hypergraph partitioning (hMETIS), for parallel Cholesky factorization (PSPASES), for collaborative filtering-based recommendation algorithms (SUGGEST), clustering high dimensional datasets (CLUTO), finding frequent patterns in diverse datasets (PAFI), and for protein secondary structure prediction (YASSPP). He has coauthored more than 210 papers on these topics and two books: Introduction to Protein Structure Prediction: Methods and Algorithms (Wiley, 2010) and Introduction to Parallel Computing (Publ. Addison Wesley, 2003, 2nd edition). In addition, he serves on the program committees of many conferences and workshops on these topics, and on the editorial boards of the IEEE Transactions on Knowledge and Data Engineering, Social Network Analysis and Data Mining Journal, International Journal of Data Mining and Bioinformatics, Journal on Current Proteomics, Advances in Bioinformatics, andBiomedicine and Biotechnology.
In our research group, we are currently working on two different areas related to social media. In the area of pharmaceuticals and social media, we are using social media data to study adverse drug reactions (ADRs) using post-marketing surveillance (also known as pharmacovigilance). ADRs for drugs are usually discovered either through clinical trials or pharmacovigilance. Pharmacovigilance has mainly relied on voluntary reports by health-care professionals, such as the Adverse Event Reporting System (AERS) used by the US Food and Drug Administration (FDA). However, the reporting rate of such systems is low, which delays the detection of ADRs and increases the possible number of deaths due to ADRs. In addition, patient reports have been shown to be of equal or more importance than the reports of health professionals, especially with the increasing use of social media and social networks (such as Twitter) in which users freely share their personal experiences. By mining the relationships between drugs and ADRs from data reported by online users on health-related issues, we can speed up the process of detecting and confirming ADRs. A related project is to mine social media data for off-label uses of medications. By mining associations between drugs and medical conditions in social media, and comparing the reported medical conditions with the ones officially approved for the drugs, we can discover new uses for these drugs, if such new associations were reported as being positive (i.e., the drug helped to cure the medical condition). These newly discovered associations can then be confirmed by clinical trials for market approval.
A second research project considers the use of social/trust networks to improve recommender systems. With increased usage of social networks like Facebook, Twitter, Epinions etc., users’ behavior is strongly influenced by the people with whom he is connected and interacts frequently. We are looking at how a user’s interaction in these networks can improve the quality of recommendations provided by the recommender system. Specifically, we are looking at incorporating the homophilic characteristics (tendency of individuals to associate with similar others) present in the social networks, to better model the users’ behavior and thus improve prediction accuracy. A related project explores recommendation models that incorporate more product review content to augment item purchase data.
“A Segment-based Approach to Clustering Multi-Topic Documents”. Andrea Tagarelli and George Karypis, Knowledge and Information Systems, Vol. 34, pp. 563—595, 2013.
“Algorithms for Mining the Evolution of Conserved Relational States in Dynamic Networks”. Ahmed Rezwan and George Karypis. Knowledge and Information Systems, Vol 33, No. 3, pp. 603—630, 2012.
“SLIM: Sparse Linear Methods for Top-N Recommender Systems”. Xia Ning and George Karypis. 11th IEEE International Conference on Data Mining (ICDM), 497—506, 2011.
“Automatic Detection of Vaccine Adverse Reactions by Incorporating Historical Medical Conditions”. Zhonghua Jiang and George Karypis, ACM Conference on Bioinformatics, Computational Biology and Biomedicine. Chicago, August, 2011.
“Item-based Top-N Recommendation Algorithms”. Mukund Deshpande and George Karypis. ACM Transactions on Information Systems. Volume 22, Issue 1, pp. 143—177, January 2004.