Ambiguity in requirement specifications causes numerous problems; for example in defining customer /supplier contracts, ensuring the integrity of safety-critical systems, and analysing the implications of system change requests. A direct appeal to formal specification has not solved these problems, partly because of the restrictiveness and lack of habitability of formal languages. An alternative approach, described in this paper, is to use natural language processing (NLP) techniques to aid the development of formal descriptions from requirements expressed in controlled natural language. While many problems in NLP remain unsolved, we show that suitable extensions to existing tools provide a useful platform for detecting and resolving ambiguities. Our system is demonstrated through a case-study on a simple requirements specification. 1. Introduction System Requirement Specifications (SRSs) often form the basis of the contract between the customer and the system supplier. Consequently,...

We introduce a learner capable of automatically extend- ing large, manually written natural language Definite Clause Grammars with missing syntactic rules. It is based upon the Minimum Description Length principle, and can be trained upon either just raw text, or else raw text additionally annotated with parsed corpora. As a demonstration of the learner, we show how full Noun Phrases (NPs that might contain pre or post- modifying phrases and might also be recursively nested) ca be identified in raw text. Preliminary results obtained by varying the amount of syntactic information in the training set suggests that raw text is less useful than additional NP bracketing information. However, using all syntactic information in the training set does not produce a significant improvement over just brack- eting information.

Practical text processing systems need wide covering grammars. When parsing unrestricted language, such grammars often fail to generate all of the sentences that humans would judge to be grammatical. This problem undermines successful parsing of the text and is known as undergeneration. There are two main ways of dealing with undergeneration: either by sentence correction, or by grammar correction. This thesis concentrates upon automatic grammar correction (or machine learning of grammar) as a solution to the problem of undergeneration. Broadly speaking, grammar correction approaches can be classified as being either datadriven, or model-based. Data-driven learners use data-intensive methods to acquire grammar. They typically use grammar formalisms unsuited to the needs of practical text processing and cannot guarantee that the resulting grammar is adequate for subsequent semantic interpretation. That is, data-driven learners acquire grammars that generate strings that humans would jud...

We use Inductive Logic Programming (ILP) within a chart-parsing framework for grammar learning. Given an existing grammar G, together with some sentences which G can not parse, we use ILP to find the "missing " grammar rules or lexical items. Our aim is to exploit the inductive capabilities of chart parsing, i.e. the ability to efficiently determine what is needed for a parse. For each unparsable sentence, we find actual edges and needed edges: those which are needed to allow a parse. The former are used as background knowledge for the ILP algorithm (P-Progol) and the latter are used as examples for the ILP algorithm. We demonstrate our approach with a number of experiments using context-free grammars and a feature grammar.

In this paper, we describe the basic Alvey Natural Language Toolkit and a set of modifications we have made to it to enhance its robustness. Following this, we report on a series of experiments that show the performance of the robust ANLT for tasks that involve the parsing of software manuals. The main findings, with respect to the robust ANLT, are that the shorter the sentence, the greater the chance of that sentence being parsed, and that the major barrier to parsing unrestricted, naturally occuring language is the development of an appropriate lexicon, and not the coverage of the grammar.

The quality of learnt natural language grammars can be enhanced by exploiting the linguistic devices that comprise a corpus. This paper considers one such device, namely punctuation. After briefly considering the linguistics of punctuation, a model capturing some of these properties is presented. Following this, a series of experiments learning unificationbased natural language grammars, using the Spoken English Corpus as data, demonstrate that even a simple model of punctuation increases the plausibility of learnt grammars over grammars learnt without the use of punctuation. Introduction Natural Language Processing (NLP) systems have for many years suffered from what Magerman terms "the toy problem syndrome" (Magerman 1994). Systems are built, often with a small lexicon or modest grammar, that are usually demonstrational rather than being operational. That is, there has been little work in developing systems capable of dealing with unrestricted, naturally occurring langua...

The problem of analysing the effects of changing requirements imposes strict demands on system representations, particularly in safety-critical domains. We argue that solving this problem will require structured representations that highlight the interaction between requirements, and record the rationale for decisions made during the development process. As a means of providing and analysing this information, we propose a framework, called goal-structured analysis, that integrates three components: a goal-oriented model for imposing the desired structure on requirements, natural language processing for useability of the system and avoidance of change via the creation of clear specifications, and formal reasoning for performing impact and sensitivity analaysis. 1 Supported by the DTI/SERC "PROTEUS" PROJECT IED4/1/9304 Contents 1 Introduction 1 2 Overview of the Goal-based Framework for Requirements Analysis 2 2.1 Goals and Effects . . . . . . . . . . . . . . . . . . . . ....

this paper are obtained by manually examining the output and classify the errors into the categories in Table 7.2. For the above sentence, the first error is an incorrect attachment of the clause "using ..." and cannot be classified into any of the categories in Table 7.2. The second error is classified as a F-error (incorrect prepositional phrases attachment).