Software Robot Requirements: Automating requirement extraction and Analysis

Informazioni relative al progetto

Requirement Robot

ID dell’accordo di sovvenzione: 739705

DOI

10.3030/739705

Progetto chiuso

Data della firma CE 27 Febbraio 2017

Data di avvio 1 Settembre 2017

Data di completamento 31 Agosto 2018

Finanziato da

• INDUSTRIAL LEADERSHIP - Innovation In SMEs

Costo totale

€ 91 250,00

Contributo UE

€ 91 250,00

Coordinato da

OSUUSKUNTA DYNAVIO
Finland

Periodic Reporting for period 1 - Requirement Robot (Software Robot Requirements: Automating requirement extraction and Analysis)

Periodo di rendicontazione: 2017-09-01 al 2018-08-31

Sintesi del contesto e degli obiettivi generali del progetto

What is the problem being addressed?
Needs for complex systems are first presented in natural language format. Those needs are processed manually by highly skilled engineers to form technical requirements. The technical requirements are forming the basis for starting developing any kind of man made artifacts and services. The project is developing an automatic approach to extract requirements from needs and to assess and cluster those requirements in categories.
Why is it important for society?
As mentioned the requirements are forming the basis for all the man made developments regarding complex systems, infrastructures and services of all kind. Getting better requirements is a must to develop better systems for societies. There exist multiple example of projects that have failed in Europe and worldwide due to poor requirements. The final success of the project is consequently source of potentially huge impact for EU in general.

What are the overall objectives?
Natural Language Processing is usually forming the basis for any kind of language processing. The methods developed in literature are usually based on massive amount of training data and use of classical Machine learning techniques. The current project is trying to take benefit of formal logic as a tool to reduce the size of required training data sets for requirements extraction and analysis. Since data is not always a cheap resource and often is hard to obtain from companies to its strategic sensitivity, reducing the size of the dataset is needed.

Lavoro eseguito dall’inizio del progetto fino alla fine del periodo coperto dalla relazione e principali risultati finora ottenuti

Work performed from the beginning of the project to the end of the period covered by the report:
During the project a formal logic based approach has been implemented, trained and tested on various test cases of industrial requirements to evaluate the capabilities of the formal logic approach in comparison with classical NLP methods. At the end of the period covered by the report, a promising usage of formal logic for classification of requirements has been validated.

Main results achieved:
A formal logic method and software kit has been implemented, successfully tested on industrial case studies and currently the approach is under deployment as a supplementary brick of the REA software. The overview of the results, their exploitation and dissemination are summarized in the final report of the project.

Progressi oltre lo stato dell’arte e potenziale impatto previsto (incluso l’impatto socioeconomico e le implicazioni sociali più ampie del progetto fino ad ora)

Progress beyond the state of the art:
The project aims at introducing a formal logic analysis module coming as a supplement to classical Natural Language Processing (NLP) techniques as well as Artificial Intelligence methods used in the NLP domain.

Expected results until the end of the project:
Validating the formal logic development and starting deploying this module in the REA software.

Potential impacts:
The project aims at automatizing the requirement extraction, clustering and analysis. The impact for society is important because the project should contribute to produce better requirements leading to better systems, services and infrastructures. This should be achieved by removing different requirements defects due to limited human capabilities to analyze massive amount of language descriptions, their interconnections, contradictions and redundancies as well as defaults such as fuzziness, ambiguities and correctness.

REA Requirement process