Application of reinforcement learning to requirements engineering: requirements tracing

Sultanov, Hakim; Hayes, Jane Huffman

doi:10.1109/re.2013.6636705

Cited by 29 publications

(15 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ML has been utilized as a way to provide computerized assistance for several requirements engineering tasks, e.g., trace link generation (Asuncion et al 2010;Cleland-Huang et al 2010;Sultanov and Hayes 2013;Guo et al 2017;Wang et al 2019), requirements identification and classification (Cleland-Huang et al 2007;Winkler and Vogelsang 2016;Kurtanović and Maalej 2017a;Dalpiaz et al 2019;Winkler et al 2019), prioritization (Perini et al 2013), ambiguity detection (Yang et al 2010;Yang et al 2012), relevance analysis (Arora et al 2019), and review classification (Maalej et al 2016;Kurtanovic and Maalej 2017b). The application of ML over textual requirements is almost always preceded by some form of NLP.…”

Section: Related Workmentioning

confidence: 99%

Automated demarcation of requirements in textual specifications: a machine learning-based approach

et al. 2020

View full text Add to dashboard Cite

A simple but important task during the analysis of a textual requirements specification is to determine which statements in the specification represent requirements. In principle, by following suitable writing and markup conventions, one can provide an immediate and unequivocal demarcation of requirements at the time a specification is being developed. However, neither the presence nor a fully accurate enforcement of such conventions is guaranteed. The result is that, in many practical situations, analysts end up resorting to after-the-fact reviews for sifting requirements from other material in a requirements specification. This is both tedious and time-consuming. We propose an automated approach for demarcating requirements in free-form requirements specifications. The approach, which is based on machine learning, can be applied to a wide variety of specifications in different domains and with different writing styles. We train and evaluate our approach over an independently labeled dataset comprised of 33 industrial requirements specifications. Over this dataset, our approach yields an average precision of 81.2% and an average recall of 95.7%. Compared to simple baselines that demarcate requirements based on the presence of modal verbs and identifiers, our approach leads to an average gain of 16.4% in precision and 25.5% in recall. We collect and analyze expert feedback on the demarcations produced by our approach for industrial requirements specifications. The results indicate that experts find our approach useful and efficient in practice. We developed a prototype tool, named DemaRQ, in support of our approach. To facilitate replication, we make available to the research community this prototype tool alongside the non-proprietary portion of our training data.

show abstract

Section: Related Workmentioning

confidence: 99%

Automated demarcation of requirements in textual specifications: a machine learning-based approach

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Their approach achieves high accuracy on average using both types of features but there are still a high number of miss-classified links. Sultanov et al [42] use reinforcement learning and improve the results compared to VSM. Niu and Mahmoud [6] use clustering to group links in high-quality and low-quality clusters respectively to improve accuracy.…”

Section: Related Workmentioning

confidence: 99%

ATLaS: A Framework for Traceability Links Recovery Combining Information Retrieval and Semi-Supervised Techniques

Bella

Creff

Gervais

et al. 2019

2019 IEEE 23rd International Enterprise Distributed Object Computing Conference (EDOC)

View full text Add to dashboard Cite

Current Model-Based Systems Engineering (MBSE) practices to design and implement complex systems require modeling and analysis based on many representations: structure, dynamics, safety, security, etc. This induces a large volume of overlapping heterogeneous artefacts which are subject to frequent changes during the project life cycle. In order to verify and validate systems requirements and ensure that models meet user's needs, MBSE techniques shall rely on consistent traceability management. In this paper, we investigate the benefits of Information Retrieval (IR) techniques and the latest advances in Natural Language Processing (NLP) approaches to suggest stakeholders with candidate semantic links generated from the processing of structured and unstructured contents. We illustrate our approach called ATLaS (Aggregation Trace Links Support) through an application on the design and analysis of a mobility service gathering several industrial partners. We provide an empirical evaluation regarding its limitations as part of an industrial MBSE process. Most importantly, we highlight how our method drastically reduces the false positive links generated compared to current IR techniques. The results obtained suggest a good synergy between the presented approach and MBSE techniques.

show abstract

“…Sultanov et al [48] find traceability candidates from highlevel to low-level requirements by the use of reinforcement learning. They use textual high and low-level requirements documents as input and try to find the candidate traces.…”

Section: Requirements Validationmentioning

confidence: 99%

A Bird's Eye View on Requirements Engineering and Machine Learning

Iqbal

Elahidoost

Levi

2018

2018 25th Asia-Pacific Software Engineering Conference (APSEC)

View full text Add to dashboard Cite

Machine learning (ML) has demonstrated practical impact in a variety of application domains. Software engineering is a fertile domain where ML is helping in automating different tasks. In this paper, our focus is the intersection of software requirement engineering (RE) and ML. To obtain an overview of how ML is helping RE and the research trends in this area, we have surveyed a large number of research articles. We found that the impact of ML can be observed in requirement elicitation, analysis and specification, validation and management. Furthermore, in these categories, we discuss the specific problem solved by ML, the features and ML algorithms used as well as datasets, when available. We outline lessons learned and envision possible future directions for the domain.

show abstract

Application of reinforcement learning to requirements engineering: requirements tracing

Cited by 29 publications

References 12 publications

Automated demarcation of requirements in textual specifications: a machine learning-based approach

Automated demarcation of requirements in textual specifications: a machine learning-based approach

ATLaS: A Framework for Traceability Links Recovery Combining Information Retrieval and Semi-Supervised Techniques

A Bird's Eye View on Requirements Engineering and Machine Learning

Contact Info

Product

Resources

About