Đang chuẩn bị liên kết để tải về tài liệu:
Báo cáo khoa học: "Interactive grammar development with WCDG"

The manual design of grammars for accurate natural language analysis is an iterative process; while modelling decisions usually determine parser behaviour, evidence from analysing more or different input can suggest unforeseen regularities, which leads to a reformulation of rules, or even to a different model of previously analysed phenomena. We describe an implementation of Weighted Constraint Dependency Grammar that supports the grammar writer by providing display, automatic analysis, and diagnosis of dependency analyses and allows the direct exploration of alternative analyses and their status under the current grammar. . | Interactive grammar development with WCDG Kilian A. Foth Michael Daum Wolfgang Menzel Natural Language Systems Group Hamburg University D-22527 Hamburgh Germany foth micha menzel @nats.informatik.uni-hamburg.de Abstract The manual design of grammars for accurate natural language analysis is an iterative process while modelling decisions usually determine parser behaviour evidence from analysing more or different input can suggest unforeseen regularities which leads to a reformulation of rules or even to a different model of previously analysed phenomena. We describe an implementation of Weighted Constraint Dependency Grammar that supports the grammar writer by providing display automatic analysis and diagnosis of dependency analyses and allows the direct exploration of alternative analyses and their status under the current grammar. 1 Introduction For parsing real-life natural language reliably a grammar is required that covers most syntactic structures but can also process input even if it contains phenomena that the grammar writer has not foreseen. Two fundamentally different ways of reaching this goal have been employed various times. One is to induce a probability model of the target language from a corpus of existing analyses and then compute the most probable structure for new input i.e. the one that under some judiciously chosen measure is most similar to the previously seen structures. The other way is to gather linguistically motivated general rules and write a parsing system that can only create structures adhering to these rules. Where an automatically induced grammar requires large amounts of training material and the development focuses on global changes to the probability model a handwritten grammar could in principle be developed without any corpus at all but considerable effort is needed to find and formulate the individual rules. If the formalism allows the ranking of grammar rules their relative importance must also be determined. This work is .