Morphological Disambiguation of Hebrew: A Case Study in Classifier Combination

Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pp. 439–447, Prague, June 2007. c⃝2007 Association for Computational Linguistics Morphological Disambiguation of Hebrew: A Case Study in Classifier Combination Danny Shacham Department of Computer Science University of Haifa Haifa, Israel dannysh@gmail.com Shuly Wintner Department of Computer Science University of Haifa Haifa, Israel shuly@cs.haifa.ac.il Abstract Morphological analysis and disambiguation are crucial stages in a variety of natural language processing applications, especially when languages with complex morphology are concerned. We present a system which disambiguates the output of a morphologi- cal analyzer for Hebrew. It consists of sev- eral simple classifiers and a module which combines them under linguistically moti- vated constraints. We investigate a number of techniques for combining the predictions of the classifiers. Our best result, 91.44% ac- curacy, reflects a 25% reduction in error rate compared with the previous state of the art. 1 Introduction Morphological analysis and disambiguation are cru- cial pre-processing steps for a variety of natural lan- guage processing applications, from search and in- formation extraction to machine translation. For languages with complex morphology these are non- trivial processes. This paper presents a morphologi- cal disambiguation module for Hebrew which uses a sophisticated combination of classifiers to rank the analyses produced by a morphological analyzer. This work has a twofold contribution: first, our sys- tem achieves over 91% accuracy on the full disam- biguation task, reducing the error rate of the pre- vious state of the art by 25%. More generally, we explore several ways for combining the predictions of simple classifiers under constraints; the insight gained from these experiments will be useful for other applications of machine learning to complex (morphological and other) problems. In the remainder of this section we discuss the complexity of Hebrew morphology, the challenge of morphological disambiguation and related work. We describe our methodology in Section 2: we use basic, na¨ıve classifiers (Section 3) to predict some components of the analysis, and then combine them in several ways (Section 4) to predict a consistent re- sult. We analyze the errors of the system in Section 5 and conclude with suggestions for future work. 1.1 Linguistic background Hebrew morphology is rich and complex.1 The ma- jor word formation machinery is root-and-pattern, and inflectional morphology is highly productive and consists of prefixes, suffixes and circumfixes. Nouns, adjectives and numerals inflect for number (singular, plural and, in rare cases, also dual) and gender (masculine or feminine). In addition, all these three types of nominals have two phonologi- cally and morphologically distinct forms, known as the absolute and construct states. In the standard orthography approximately half of the nominals ap- pear to have identical forms in both states, a fact which substantially increases the ambiguity. In ad- dition, nominals take possessive pronominal suffixes which inflect for number, gender and person. Verbs inflect for number, gender and person (first, second and third) and also for a combination of tense and aspect/mood, referred to simply as ‘tense’ …

Content truncated to 500 words