Which Are the Best Features for Automatic Verb Classification

Proceedings of ACL-08: HLT, pages 434–442, Columbus, Ohio, USA, June 2008. c⃝2008 Association for Computational Linguistics Which Are the Best Features for Automatic Verb Classification Jianguo Li Department of Linguistics The Ohio State University Columbus Ohio, USA jianguo@ling.ohio-state.edu Chris Brew Department of Linguistics The Ohio State University Columbus Ohio, USA cbrew@ling.ohio-state.edu Abstract In this work, we develop and evaluate a wide range of feature spaces for deriving Levin- style verb classifications (Levin, 1993). We perform the classification experiments using Bayesian Multinomial Regression (an effi- cient log-linear modeling framework which we found to outperform SVMs for this task) with the proposed feature spaces. Our exper- iments suggest that subcategorization frames are not the most effective features for auto- matic verb classification. A mixture of syntac- tic information and lexical information works best for this task. 1 Introduction Much research in lexical acquisition of verbs has concentrated on the relation between verbs and their argument frames. Many scholars hypothesize that the behavior of a verb, particularly with respect to the expression of arguments and the assignment of semantic roles is to a large extent driven by deep semantic regularities (Dowty, 1991; Green, 1974; Goldberg, 1995; Levin, 1993). Thus measurements of verb frame patterns can perhaps be used to probe for linguistically relevant aspects of verb meanings. The correspondence between meaning regularities and syntax has been extensively studied in Levin (1993) (hereafter Levin). Levin’s verb classes are based on the ability of a verb to occur or not occur in pairs of syntactic frames that are in some sense meaning preserving (diathesis alternation). The fo- cus is on verbs for which distribution of syntactic frames is a useful indicator of class membership, and, correspondingly, on classes which are relevant for such verbs. By using Levin’s classification, we obtain a window on some (but not all) of the poten- tially useful semantic properties of verbs. Levin’s verb classification, like others, helps re- duce redundancy in verb descriptions and enables generalizations across semantically similar verbs with respect to their usage. When the information about a verb type is not available or sufficient for us to draw firm conclusions about its usage, the infor- mation about the class to which the verb type be- longs can compensate for it, addressing the perva- sive problem of data sparsity in a wide range of NLP tasks, such as automatic extraction of subcategoriza- tion frames (Korhonen, 2002), semantic role label- ing (Swier and Stevenson, 2004; Gildea and Juraf- sky, 2002), natural language generation for machine translation (Habash et al., 2003), and deriving pre- dominant verb senses from unlabeled data (Lapata and Brew, 2004). Although there exist several manually-created verb lexicons or ontologies, including Levin’s verb taxonomy, VerbNet, and FrameNet, automatic verb classification (AVC) is still necessary for extend- ing existing lexicons (Korhonen and Briscoe, 2004), building and tuning lexical information specific to different domains (Korhonen et al., 2006), and boot- strapping verb lexicons for new languages (Tsang et al., 2002). AVC helps avoid the expensive hand-coding of such information, …

Content truncated to 500 words