Natural Language Processing to Extract Contextual Structure from Requirements

Maximilian Vierlboeck, Daniel Dunbar, Roshanak Nilchiani

    Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

    6 Scopus citations

    Abstract

    The automatic extraction of structure from text can be difficult for machines. Yet, the elicitation of this information can provide many benefits and opportunities for various applications. Such benefits have been identified amongst others for the area of Requirements Engineering. By assessing the Natural Language Processing for Requirement Engineering status quo and literature, a necessity for an automatic and universal approach to elicit structure from requirement and specification documents was identified. This paper outlines the first steps and results towards a modularized approach that splits the core algorithm from the text corpus as an input and underlying rule/knowledge base. This separation of functions allows for individual modification of the included parts and eases or potentially removes restrictions as well as limitations, such as input rules or the necessity for human supervision. Furthermore, contextual information and links via ontology inference can be considered that are not explicit on a textual level. The initial results of the approach show the successful extraction of structural information from requirement text, which was validated by comparing the results to human interpretations for small and public sample sets. In addition, the contextual consideration and inference via ontologies is described conceptually. At the current stage, limitations still exist regarding scalability and handling of text ambiguities, but solutions for these caveats have been developed and are being tested. Overall, the approach and results presented will be integrated and are part of a novel requirement complexity assessment framework.

    Original languageEnglish
    Title of host publicationSysCon 2022 - 16th Annual IEEE International Systems Conference, Proceedings
    ISBN (Electronic)9781665439923
    DOIs
    StatePublished - 2022
    Event16th Annual IEEE International Systems Conference, SysCon 2022 - Virtual, Online, Canada
    Duration: 25 Apr 202223 May 2022

    Publication series

    NameSysCon 2022 - 16th Annual IEEE International Systems Conference, Proceedings

    Conference

    Conference16th Annual IEEE International Systems Conference, SysCon 2022
    Country/TerritoryCanada
    CityVirtual, Online
    Period25/04/2223/05/22

    Keywords

    • complexity
    • contextual information
    • natural language processing
    • ontology
    • requirements engineering
    • structure

    Fingerprint

    Dive into the research topics of 'Natural Language Processing to Extract Contextual Structure from Requirements'. Together they form a unique fingerprint.

    Cite this