Development of a Persian Syntactic Dependency Treebank

Development of a Persian Syntactic Dependency Treebank Mohammad Sadegh Rasooli Department of Computer Science Columbia University New York, NY rasooli@cs.columbia.edu Manouchehr Kouhestani Department of Linguistics Tarbiat Modares University Tehran, Iran m.kouhestani@modares.ac.ir Amirsaeid Moloodi Department of Linguistics University of Tehran Tehran, Iran a.moloodi@ut.ac.ir Abstract This paper describes the annotation process and linguistic properties of the Persian syn- tactic dependency treebank. The treebank consists of approximately 30,000 sentences annotated with syntactic roles in addition to morpho-syntactic features. One of the unique features of this treebank is that there are al- most 4800 distinct verb lemmas in its sen- tences making it a valuable resource for ed- ucational goals. The treebank is constructed with a bootstrapping approach by means of available tagging and parsing tools and man- ually correcting the annotations. The data is splitted into standard train, development and test set in the CoNLL dependency format and is freely available to researchers. 1 Introduction1 The process of manually annotating linguistic data from a huge amount of naturally occuring texts is a very expensive and time consuming task. Due to the recent success of machine learning methods and the rapid growth of available electronic texts, language processing tasks have been facilitated greatly. Con- sidering the value of annotated data, a great deal of budget has been allotted to creating such data. Among all linguistic datasets, treebanks play an important role in the natural language processing tasks especially in parsing because of its applica- 1This research is done while working in Dadegan Research Group, Supreme Council of Information and Communications Technology (SCICT), Tehran, Iran. The project is fully funded by SCICT. tions in tasks such as machine translation. Depen- dency treebanks are collections of sentences with their corresponding dependency trees. In the last decade, many dependency treebanks have been de- veloped for a large number of languages. There are at least 29 languages for which at least one depen- dency treebank is available (Zeman et al., 2012). Dependency trees are much more similar to the hu- man understanding of language and can easily rep- resent the free word-order nature of syntactic roles in sentences (K¨ubler et al., 2009). Persian is a language with about 110 million speakers all over the world (Windfuhr, 2009), yet in terms of the availability of teaching materials and annotated data for text processing, it is undoubt- edly a low-resourced language. The need for more language teaching materials together with an ever- increasing need for Persian-language data process- ing has been the incentive for the inception of our project which has defined the development of the syntactic treebank of Persian as its ultimate aim. In this paper, we review the process of creating the Per- sian syntactic treebank based on dependency gram- mar. In this treebank, approximately 30,000 sen- tences from contemporary Persian-language texts are manually tokenized and annotated at morpholog- ical and syntactic levels. One valuable aspect of the treebank is its containment of near 5000 distinct verb lemmas in its sentences making it a good resource for educational …

Content truncated to 500 words