{"references": ["E.Gabrilovich andS.Markovitch, \"Overcoming the brittleness bottleneck\nusing Wikipedia: enhancing text categorization with encyclopedic\nknowledge,\"inProceedings of the 21st National Conference on Artificial\nIntelligence, Boston,2006, pp. 787-788.", "E.Gabrilovich andS.Markovitch, \"Computing semantic relatedness using\nWikipedia-based explicit semantic analysis,\"inProceedings of the 20th\nInternational Joint Conference on Artificial Intelligence, Hyderabad,\n2007, pp. 1606-1611.", "P.Wang andC.Domeniconi, \"Building semantic kernels for text classification\nusing Wikipedia,\"inProceeding of the 14th ACM SIGKDD International\nConference on Knowledge Discovery and Data Mining, Las\nVegas, 2008, pp. 713-721.", "S.Banerjee, K.Ramanathanand A.Gupta, \"Clustering short texts using\nWikipedia,\"inProceedings of the 30th Annual International ACM SIGIR\nConference on Research and Development in Information Retrieval,\nAmsterdam, 2007, pp. 787-788.", "J.Hu, L.Fang,Y.Cao, et al.,\"Enhancing text clustering by leveraging\nWikipedia semantics,\"inProceedings of the 31st Annual International\nACM SIGIR Conference on Research and Development in Information\nRetrieval, Singapore, 2008, pp. 179-186.", "X.Hu, X.Zhang, C.Lu,E. K. Park and X. Zhou,\"Exploiting Wikipedia as\nexternal knowledge for document clustering,\"inProceedings of the 15th\nACM SIGKDD Conference on Knowledge Discovery and Data Mining,\nParis, 2009, pp. 389-396.", "Y.Miao andC.Li, \"Enhancing query-oriented summarization based on\nsentence wikification,\"inWorkshop of the 33rd Annual International\nACM SIGIR Conference on Research and Development in Information\nRetrieval, 2010.", "Y.Li,W.P.R.Luk,K.S.E.Ho and F.L.K. Chung,\"Improving weak ad-hoc\nqueries using Wikipedia as external corpus,\"inProceedings of the 30th\nAnnual International ACM SIGIR Conference on Research and Development\nin Information Retrieval, Amsterdam, 2007, pp. 797-798.", "Y.Miao andC.Li, \"Mining Wikipedia and Yahoo! Answers for question\nexpansion in opinion QA,\"inAdvances in Knowledge Discovery and Data\nMining, vol. 6118/2010, pp. 367-374. Springer, 2010.\n[10] G.Jeh andJ.Widom, \"SimRank: Ameasure of structural-context similarity,\"\nin: Proceedings of the 8th ACM SIGKDD International Conference\non Knowledge Discovery and Data Mining, Edmonton,2002, pp.\n538-543.\n[11] I.Antonellis, H.Garcia-Molina andC.-C.Chang, \"Simrank++: Query\nrewriting through link analysis of the click graph,\"in Proceedings of the\nVery Large Databases,vol.1, iss.1, pp. 408-421, 2008.\n[12] D.Lizorkin, P.Velikhov, M.Grinev andD.Turdakov, \"Accuracy estimate\nand optimization techniques for Simrankcomputation,\"inProceedings of\nthe Very Large Databases,vol.1, iss.1, pp.422-433, 2008.\n[13] S.Zhong andJ.Ghosh, \"Generative model-based document clustering: A\ncomparative study,\"inKnowledge and Information Systems, vol.8, no.3,\npp.374-384, Springer, 2005.\n[14] H.Small, \"Co-citation in the scientific literature: A newmeasure of the\nrelationship between two documents,\"Journal of American Society for\nInformation Science,vol.24,iss.4,pp. 265-269, 1973.\n[15] A.Hotho, S.Staab andG.Stumme, \"Wordnet improves text document\nclustering,\"inSemantic Web Workshop of the 26th Annual International\nACM SIGIR Conference on Research and Development in Information\nRetrieval, 2003.\n[16] I.Yoo, X.Hu and I.-Y.Song, \"Integration of semantic-based bipartite\ngraph representation and mutual refinement strategy for biomedical literature\nclustering,\"inProceedings of the 12th ACM SIGKDD Conference\non Knowledge Discovery and Data Mining,Philadelphia, 2006, pp.\n791-796.\n[17] L'aszl'o andLov'asz,\"Random walks on graphs: A survey,\"Bolyai\nSociety Mathematical Studies, vol.2, pp.1-46, 1993."]}

Text similarity measurement is a fundamental issue in many textual applications such as document clustering, classification, summarization and question answering. However, prevailing approaches based on Vector Space Model (VSM) more or less suffer from the limitation of Bag of Words (BOW), which ignores the semantic relationship among words. Enriching document representation with background knowledge from Wikipedia is proven to be an effective way to solve this problem, but most existing methods still cannot avoid similar flaws of BOW in a new vector space. In this paper, we propose a novel text similarity measurement which goes beyond VSM and can find semantic affinity between documents. Specifically, it is a unified graph model that exploits Wikipedia as background knowledge and synthesizes both document representation and similarity computation. The experimental results on two different datasets show that our approach significantly improves VSM-based methods in both text clustering and classification.