Extracting Appraisal Expressions

Proceedings of NAACL HLT 2007, pages 308–315, Rochester, NY, April 2007. c⃝2007 Association for Computational Linguistics Extracting Appraisal Expressions Kenneth Bloom and Navendu Garg and Shlomo Argamon Computer Science Department Illinois Institute of Technology 10 W. 31st St. Chicago, IL 60616 {kbloom1,gargnav,argamon}@iit.edu Abstract Sentiment analysis seeks to characterize opinionated or evaluative aspects of nat- ural language text. We suggest here that appraisal expression extraction should be viewed as a fundamental task in sentiment analysis. An appraisal expression is a tex- tual unit expressing an evaluative stance towards some target. The task is to find and characterize the evaluative attributes of such elements. This paper describes a system for effectively extracting and dis- ambiguating adjectival appraisal expres- sions in English outputting a generic rep- resentation in terms of their evaluative function in the text. Data mining on ap- praisal expressions gives meaningful and non-obvious insights. 1 Introduction Sentiment analysis, which seeks to analyze opin- ion in natural language text, has grown in interest in recent years. Sentiment analysis includes a vari- ety of different problems, including: sentiment clas- sification techniques to classify reviews as positive or negative, based on bag of words (Pang et al., 2002) or positive and negative words (Turney, 2002; Mullen and Collier, 2004); classifying sentences in a document as either subjective or objective (Riloff and Wiebe, 2003; Pang and Lee, 2004); identifying or classifying appraisal targets (Nigam and Hurst, 2004); identifying the source of an opinion in a text (Choi et al., 2005), whether the author is expressing the opinion, or whether he is attributing the opinion to someone else; and developing interactive and vi- sual opinion mining methods (Gamon et al., 2005; Popescu and Etzioni, 2005). Much of this work has utilized the fundamental concept of ‘semantic orien- tation’, (Turney, 2002); however, sentiment analysis still lacks a ‘unified field theory’. We propose in this paper that a fundamental task underlying many of these formulations is the extrac- tion and analysis of appraisal expressions, defined as those structured textual units which express an evaluation of some object. An appraisal expression has three main components: an attitude (which takes an evaluative stance about an object), a target (the object of the stance), and a source (the person tak- ing the stance) which may be implied. The idea of appraisal extraction is a generaliza- tion of problem formulations developed in earlier works. Mullen and Collier’s (2004) notion of classi- fying appraisal terms using a multidimensional set of attributes is closely tied to the definition of an appraisal expression, which is classified along sev- eral dimensions. In previous work (Whitelaw et al., 2005), we presented a related technique of find- ing opinion phrases, using a multidimensional set of attributes and modeling the semantics of mod- ifiers in these phrases. The use of multiple text classifiers by Wiebe and colleagues (Wilson et al., 2005; Wiebe et al., 2004) for various kinds of senti- ment classification can also be viewed as a sentence- level technique for analyzing appraisal expressions. Nigam and Hurst’s …

Content truncated to 500 words