CPN-CORE: A Text Semantic Similarity System Infused with Opinion Knowledge

Second Joint Conference on Lexical and Computational Semantics (*SEM), Volume 1: Proceedings of the Main Conference and the Shared Task, pages 221–228, Atlanta, Georgia, June 13-14, 2013. c⃝2013 Association for Computational Linguistics CPN-CORE: A Text Semantic Similarity System Infused with Opinion Knowledge Carmen Banea♭∗, Yoonjung Choi‡, Lingjia Deng‡, Samer Hassan§, Michael Mohler♦ Bishan Yang √ , Claire Cardie √ , Rada Mihalcea♭†, Janyce Wiebe‡ ♭University of North Texas Denton, TX ‡University of Pittsburgh Pittsburgh, PA §Google Inc. Mountain View, CA ♦Language Computer Corp. Richardson, TX √ Cornell University Ithaca, NY Abstract This article provides a detailed overview of the CPN text-to-text similarity system that we par- ticipated with in the Semantic Textual Similar- ity task evaluations hosted at *SEM 2013. In addition to more traditional components, such as knowledge-based and corpus-based met- rics leveraged in a machine learning frame- work, we also use opinion analysis features to achieve a stronger semantic representation of textual units. While the evaluation datasets are not designed to test the similarity of opinions, as a component of textual similarity, nonethe- less, our system variations ranked number 38, 39 and 45 among the 88 participating systems. 1 Introduction Measures of text similarity have been used for a long time in applications in natural language processing and related areas. One of the earliest applications of text similarity is perhaps the vector-space model used in information retrieval, where the document most relevant to an input query is determined by ranking documents in a collection in reversed or- der of their angular distance with the given query (Salton and Lesk, 1971). Text similarity has also been used for relevance feedback and text classifi- cation (Rocchio, 1971), word sense disambiguation (Lesk, 1986; Schutze, 1998), and extractive summa- rization (Salton et al., 1997), in the automatic evalu- ation of machine translation (Papineni et al., 2002), ∗carmen.banea@gmail.com †rada@cs.unt.edu text summarization (Lin and Hovy, 2003), text co- herence (Lapata and Barzilay, 2005) and in plagia- rism detection (Nawab et al., 2011). Earlier work on this task has primarily focused on simple lexical matching methods, which produce a similarity score based on the number of lexical units that occur in both input segments. Improvements to this simple method have considered stemming, stopword removal, part-of-speech tagging, longest subsequence matching, as well as various weight- ing and normalization factors (Salton and Buckley, 1997). While successful to a certain degree, these lexical similarity methods cannot always identify the semantic similarity of texts. For instance, there is an obvious similarity between the text segments “she owns a dog” and “she has an animal,” yet these methods will mostly fail to identify it. More recently, researchers have started to con- sider the possibility of combining the large number of word-to-word semantic similarity measures (e.g., (Jiang and Conrath, 1997; Leacock and Chodorow, 1998; Lin, 1998; Resnik, 1995)) within a semantic similarity method that works for entire texts. The methods proposed to date in this direction mainly consist of either bipartite-graph matching strate- gies that aggregate word-to-word similarity into a text …

Content truncated to 500 words