Maxime Warnier

Linguistic Analysis of Requirements of a Space Project and Their Conformity with the Recommendations Proposed by a Controlled Natural Language

The long term aim of the project carried out by the French National Space Agency (CNES) is to design a writing guide based on the real and regular writing of requirements. As a first step in the project, this paper proposes a linguistic analysis of requirements written in French by CNES engineers. The aim is to determine to what extent they conform to two rules laid down in INCOSE, a recent guide for writing requirements. Although CNES engineers are not obliged to follow any Controlled Natural Language in their writing of requirements, we believe that language regularities are likely to emerge from this task, mainly due to the writers’ experience. The issue is approached using natural language processing tools to identify sentences that do not comply with INCOSE rules. We further review these sentences to understand why the recommendations cannot (or should not) always be applied when specifying largescale projects.

Mapping French terms in a Belgian guideline on heart failure to international classifications and nomenclatures: the devil is in the detail

INTRODUCTION With growing sophistication of eHealth platforms, medical information is increasingly shared across patients, health care providers, institutions and across borders. This implies more stringent demands on the quality of data entry at the point-of-care. Non-native English-speaking general practitioners (GPs) experience difficulties in interacting with international classification systems and nomenclatures to facilitate the secondary use of their data and to ensure semantic interoperability. AIM To identify words and phrases pertaining to the heart failure domain and to explore the difficulties in mapping to corresponding concepts in ICPC-2, ICD-10, SNOMED-CT and UMLS. METHODS The medical concepts in a Belgian guideline for GPs in its French version were extracted manually and coded first in ICPC-2, then ICD-10 by a physician, an expert in classification systems. In addition, mappings were sought with SNOMED-CT and UMLS concepts, using the UMLS SNOMED-CT browser. RESULTS We identified 143 words and phrases, of which 128 referred to a single concept (1-to-1 mapping), while 15 referred to two or more concepts (1-to-n mapping to ICPC rubrics or to the other nomenclatures). In the guideline, words or phrases were often too general for specific mapping to a code or term. Marked discrepancy between semantic tags and types was found. CONCLUSION This article shows the variability of the various international classifications and nomenclatures, the need for structured guidelines with more attention to precise wording and the need for classification expertise embedded in sophisticated terminological resources. End users need support to perform their clinical work in their own language, while still assuring standardised and semantic interoperable medical registration. Collaboration between computational linguists, knowledge engineers, health informaticians and domain experts is needed.

Towards the creation of a CNL adapted to requirements writing by combining writing recommendations and spontaneous regularities: example in a space project

The Quality Department of the French National Space Agency (CNES, Centre National d’Études Spatiales) wishes to design a writing guide based on the real and regular writing of requirements. As a first step in this project, the present article proposes a linguistic analysis of requirements written in French by CNES engineers. One of our goals is to determine to what extent they conform to several rules laid down in two existing Controlled Natural Languages (CNLs), namely the Simplified Technical English developed by the AeroSpace and Defense Industries Association of Europe and the Guide for Writing Requirements proposed by the International Council on Systems Engineering. Indeed, although CNES engineers are not obliged to follow any controlled language in their writing of requirements, we believe that language regularities are likely to emerge from this task, mainly due to the writers’ experience. We are seeking to identify these regularities in order to use them as a basis for a new CNL for the writing of requirements. The issue is approached using natural language processing tools to identify sentences that do not comply with the rules or contain specific linguistic phenomena. We further review these sentences to understand why the recommendations cannot (or should not) always be applied when specifying large-scale projects.

Improving Requirement Boilerplates Using Sequential Pattern Mining

In the field of requirements engineering, the use of the so-called boilerplates (i.e. standard phrases and sentences containing gaps to be filled in) is a popular solution to reduce variation among requirements and writers, and thus to improve the clarity of technical specifications. However, the examples of boilerplates found in the literature are often very general, as they need to be applicable to projects as varied as computer software and aircraft or industrial machines. As a result, they only partially fulfill their role, leaving a lot of freedom to the writers in charge of filling in the gaps. Instead, we would like to propose a bottom-up approach for discovering more specific sequences that could constitute either boilerplates or elements to instantiate these boilerplates. To this end, we investigate whether sequential data mining techniques can be used on a small corpus of genuine requirements written in French at CNES (Centre National d’Études Spatiales), the French Space Agency.

How can corpus linguistics help improve requirements writing? Specifications of a space project as a case study

The specific purpose of this doctoral research is to improve the writing of requirements at the French Space Agency (CNES) by proposing a set of linguistic rules - referred to as a Controlled Natural Language (CNL) - that engineers should follow when writing out specifications in French. CNLs for technical writing do already exist, but if they are reviewed from a linguistic point of view, they are found unsatisfactory and too constraining, because some of the rules they impose lack relevance or are not compatible with the way engineers actually specify large-scale systems. In this research abstract, we will present a methodology based on corpus analysis aimed at improving existing rules and suggesting new ones that are inspired by existing data. We will also consider requirements extracted from specifications written at CNES to demonstrate its feasibility.