The NWO Priority Programme Language and Speech Technology is a research programme aiming at the development of spoken language information systems. Its immediate goal is to develop a demonstrator of a public transport information system, which operates over ordinary telephone lines. This demonstrator is called OVIS, Openbaar Vervoer Informatie Systeem ( Public Transport Information System). The language of the system is Dutch.
At present, a prototype is in operation, which is a version of a German system developed by Philips Dialogue Systems in Aachen , adapted to Dutch. This German system processes spoken input using `` concept spotting'', which means that the smallest information-carrying units in the input are extracted, such as locative phrases (mostly names of train stations) and temporal expressions, and these are translated more or less individually into updates of the internal database representing the dialogue state. The words between the concepts thus perceived are ignored.
The use of concept spotting is common in spoken-language information systems [47,24,6,2]. Arguments in favour of this kind of shallow parsing are that it is relatively easy to develop the NLP component, since larger sentence constructs do not have to be taken into account, and that the robustness of the parser is enhanced, since sources of ungrammaticality occurring between concepts are skipped and therefore do not hinder the translation of the utterance to updates.
The prototype presently under construction (OVIS2) is based on a grammar which describes grammatical sentences, i.e. complete and well-formed user utterances, and thus differs radically from a concept spotting approach. This article presents a detailed account of a computational grammar for Dutch, and a robust parsing algorithm which incorporates this grammatical knowledge as well as other knowledge sources, such as acoustic evidence and Ngram statistics. We argue that robust parsing can be based on sophisticated grammatical analysis. In particular, the grammar describes full sentences, but in doing so, also describes the grammar of temporal expressions and locative phrases which are crucial for concept spotting. Robustness is achieved by taking these phrases into consideration, if a full parse of an utterance is not available. We show that our approach is feasible in terms of both accuracy and computational resources, and thus is a viable alternative to pure concept spotting.
Whereas some (e.g. Moore, Pereira, and Murveit ) argue that grammatical analysis may improve recognition accuracy, our current experiments have as yet not been able to reveal a substantial advantage in this respect. However, the grammatical approach may become essential as soon as the application is extended in such a way that more complicated grammatical constructions need to be recognized. In that case, simple concept spotting may not be able to correctly process all constructions, whereas the capabilities of the grammatical approach extend much further.
The structure of this paper is as follows. In section 2 we describe the grammar for OVIS2. We present the grammar in some detail, since we believe it constitutes an interesting compromise between linguistic and computational considerations. Readers interested in processing issues rather than the details of linguistic analysis might prefer to skip section 2 (possibly except the first paragraph) and jump to section 3 immediately. That section describes the robust parsing algorithm. Section 4 reports test results, showing that grammatical analysis allows fast and accurate processing of spoken input.