WIT Press

NL-OOPS: A Requirements Analysis Tool Based On Natural Language Processing


Free (open access)







746 kb

Paper DOI



WIT Press


L Mich & R Garigliano


The main goal of the NL-OOPS (acronym for Natural Language – Object-Oriented Production System) project is to develop a tool supporting object-oriented analysis using natural language (NL) processing. Requirements documents are analysed with LOLITA, a large-scale NL processing system, developed at Durham University and 3F Ltd. Both, the knowledge in the documents and that already stored in the knowledge base of LOLITA can then be used to produce requirements models at different levels of detail. Object oriented modelling is based on a two-phase algorithm for the identification of classes and associations. Moreover, the latest version of NL-OOPS supports traceability between the original input texts, their representation in LOLITA and the final models. To illustrate the main features and the performance of the tools we refer to the Automated Teller Machine study case described by the authors of the Object Modeling Technique. 1. Introduction The NL-OOPS project started in 1994 as a collaboration between the University of Trento (I) and the University of Durham (UK). The main goal of this project is to develop a CASE tool supporting requirements analysis by generating object oriented conceptual models from requirements documents in NL. Many authors have proposed the idea of extracting object from NL documents, starting from Abbot in 1983, and for entity-relationship diagrams it is older still. Our approach is based on the consideration that requirements are often written in unrestricted