Anaphora Annotation in Hindi Dependency TreeBank

Copyright 2012 by Praveen Dakwale, Himanshu Sharma, and Dipti M Sharma 26th Pacific Asia Conference on Language,Information and Computation pages 391–400 Anaphora Annotation in Hindi Dependency TreeBank Praveen Dakwale LTRC, IIIT-H, India dakwale.praveen@gmail.com Himanshu Sharma LTRC, IIIT-H, India himanshu s@students.iiit.ac.in Dipti M Sharma LTRC, IIIT-H, India diptims@gmail.com Abstract In this paper, we propose a scheme for anaphora annotation in Hindi Dependency Treebank. The goal is to identify and han- dle the challenges that arise in the annotation of reference relations in Hindi. We identify some of the issues related to anaphora anno- tation specific to Hindi such as distribution of markable span, sequential annotation, repre- sentation format, annotation of multiple refer- ents etc. The scheme hence incorporates some characteristics specific to these issues in or- der to achieve a consistent annotation. Most significant among these characteristics is the head-modifier separation in referent selection. The modifier-modified dependency relations inside a markable is utilized for this head- modifier distinction. A part of the Hindi De- pendency Treebank, of around 2500 sentences has been annotated with anaphoric relations and an inter-annotator study was carried out which shows a significant agreement over se- lection of the head referent using the proposed scheme as compared to MUC annotation for- mat. The current annotation is done for a lim- ited set of pronominal categories. 1 Introduction In this paper we present a scheme for annotating anaphoric relations in the Hindi Dependency Tree- Bank. Anaphora Resolution is one of the important problems in Natural Language Processing, and is used by various applications such as Text Summa- rization, Question answering etc. An anaphora an- notated corpus along with other features (like POS, morph, Parse structure etc.) is required in both sta- tistical as well as rule based anaphora resolution sys- tems. Various corpus based studies of anaphoric variation also make use of such a corpus. While a significant number of corpora with anaphora anno- tation for English and other languages like Spanish, Czech etc. are available, for Indian languages, such corpora are scarce. With a view of developing an Anaphora Resolu- tion system in Hindi, our project aims at extending the dependency annotated (Hindi Dependency Tree- Bank) corpus with anaphoric relations. Hence we propose an anaphora annotation scheme in accor- dance with the representation format (SSF)(Bharati et al., 2007) of the Treebank, that uses attribute- value pairs to represent linguistic information. In this scheme, we attempt to address some of the issues that are commonly faced while annotating anaphora and require efficient handling. Although the scheme is developed while keeping in view the structure of the Dependency Tree-Bank, it is con- vertible to other formats of annotation as well. In recent years, due to increasing interest in devel- opment of statistical systems for anaphora resolu- tion, there have been significant attempts for cre- ation of anaphora annotated corpora and annotation schemes. The most well known among these are MUC-7 annotation scheme (Hirschman and Chin- chor, 1997) and other MUC based schemes, which are used for co-reference annotation via …

Content truncated to 500 words