A Lexical Resource for the Identification of “Weak Words” in German Specification Documents

A Lexical Resource for the Identification of “Weak Words” in German Specification Documents Jennifer Krisch1,2, Melanie Dick2, Ronny Jauch2 and Ulrich Heid2 1Daimler AG Stuttgart, 2Universit¨at Hildesheim – Institute for Information Science and Natural Language Processing, jennifer.krisch@daimler.com, {ronny.jauch, melanie.dick, ulrich.heid}@uni-hildesheim.de Abstract We report on the creation of a lexical resource for the identification of potentially unspecific or imprecise constructions in German requirements documentation from the car manufacturing industry. In requirements engineering, such expressions are called “weak words”: they are not sufficiently precise to ensure an unambiguous interpretation by the contractual partners, who for the definition of their cooperation, typically rely on specification documents (Melchisedech, 2000); an example are dimension adjectives, such as kurz or lang (‘short’, ‘long’) which need to be modified by adverbials indicating the exact duration, size etc. Contrary to standard practice in requirements engineering, where the identification of such weak words is merely based on stopword lists, we identify weak uses in context, by querying annotated text. The queries are part of the resource, as they define the conditions when a word use is weak. We evaluate the recognition of weak uses on our development corpus and on an unseen evaluation corpus, reaching stable F1-scores above 0.95. Keywords: requirements engineering, weak words, corpus-based methods 1. Introduction In the automotive industry, requirements specifications are used as a communication medium between contractor and supplier. Requirements specifications contain textual re- quirements which are “conditions or capabilities that must be met or possessed by a system or system component”. These texts serve “to achieve a consensus among the stake- holders [...] to minimize the risk of delivering a system that does not meet the contractor’s desires and needs” (Pohl and Rupp, 2011). Requirements texts should be unambiguous and easy to understand and they should allow consensual interpretation. The requirements are produced by many different authors (typically domain experts), without pos- sibilities of using controlled language and author control systems; consequently, checks for ambiguities, for under- specified constructions etc. are needed before the require- ments are made available to suppliers. Requirements that pose problems of ambiguity, underspecification or under- standability are thus given back to the author, for correc- tion, before they are made official. Our resource is part of a system for such checks. A problem in the writing of requirements documents, which is well-known in the field of requirements engineering is the use of potentially vague adjectives. In requirements en- gineering they are called ”weak words”. Whether a weak word may lead to different interpretations depends on the syntactic and lexical context in which it is used. The class of weak words includes scalable adjectives (for example klein (‘small’)) and limit adjectives (for example: unter- schiedlich (‘different’)), cf. Paradis (1997). Bierwisch (1989) further subdivides scalable adjectives into dimen- sional adjectives and evaluative adjectives. In this work we concentrate on dimensional and limit adjectives, as evala- tive adjectives are rare in our data. Another example of a weak word is the dimensional adjec- tive lang (‘long’). When an author uses the …

Content truncated to 500 words