A Semantic Kernel for Predicate Argument Classification

A Semantic Kernel for Predicate Argument Classification Alessandro Moschitti and Cosmin Adrian Bejan University of Texas at Dallas Human Language Technology Research Institute Richardson, TX 75083-0688, USA alessandro.moschitti@utdallas.edu ady@hlt.utdallas.edu Abstract Automatically deriving semantic structures from text is a challenging task for machine learning. The flat feature representations, usu- ally used in learning models, can only partially describe structured data. This makes difficult the processing of the semantic information that is embedded into parse-trees. In this paper a new kernel for automatic clas- sification of predicate arguments has been de- signed and experimented. It is based on sub- parse-trees annotated with predicate argument information from PropBank corpus. This ker- nel, exploiting the convolution properties of the parse-tree kernel, enables us to learn which syntactic structures can be associated with the arguments defined in PropBank. Support Vec- tor Machines (SVMs) using such a kernel clas- sify arguments with a better accuracy than SVMs based on linear kernel. 1 Introduction Several linguistic theories, e.g. (Jackendoff, 1990), claim that semantic information in natural language texts is connected to syntactic structures. Hence, to deal with nat- ural language semantics, the learning algorithm should be able to represent and process structured data. The classi- cal solution adopted for such tasks is to convert syntax structures in a flat feature representation, which is suit- able for a given learning model. The main drawback is structures may not be properly represented by flat fea- tures as: (1) these latter may not be able to capture the required properties or (2) the feature designer may not know what structure properties enable the processing of semantic information. In particular, these problems arise for semantic infor- mation represented via predicate argument structures de- fined on syntactic parse trees. For example, Figure 1 shows the parse tree of the sentence: "Paul gives a lecture in Rome" along with the annotation of predicate arguments. A predicate may be a verb or a noun or an adjective whereas generally Arg 0 stands for agent, Arg 1 for di- rect object or theme or patient and ArgM may indicate locations, as in our example. A standard for predicate ar- gument annotation is provided in the PropBank project (Kingsbury and Palmer, 2002). It has produced one million word corpus annotated with predicate-argument structures on top of the Penn Treebank 2 Wall Street Jour- nal texts. In this way, for a large number of the Penn TreeBank parse-trees, there are available predicate anno- tations in a style similar to that shown in Figure 1. Predicate Arg. 0 Arg. M S N NP D N VP V Paul in gives a lecture PP IN N Rome Arg. 1 Figure 1: Predicate arguments in a parse-tree representation. In PropBank only verbs are considered to be predicates whereas arguments are labeled sequentially from Arg 0 to Arg 91. In addition to these core arguments, adjunctive arguments are marked up. They include functional tags, e.g. ArgM-DIR indicates a directional, ArgM-LOC in- dicates a locative and ArgM-TMP stands for a temporal. An …

Content truncated to 500 words