Comparing Pre-Training Schemes for Luxembourgish BERT Models

in Proceedings of the 19th Conference on Natural Language Processing (KONVENS 2023) (2023, September)

Despite the widespread use of pre-trained models in NLP, well-performing pre-trained models for low-resource languages are scarce. To address this issue, we propose two novel BERT models for the ... [more ▼]

Despite the widespread use of pre-trained models in NLP, well-performing pre-trained models for low-resource languages are scarce. To address this issue, we propose two novel BERT models for the Luxembourgish language that improve on the state of the art. We also present an empirical study on both the performance and robustness of the investigated BERT models. We compare the models on a set of downstream NLP tasks and evaluate their robustness against different types of data perturbations. Additionally, we provide novel datasets to evaluate the performance of Luxembourgish language models. Our findings reveal that pre-training a pre-loaded model has a positive effect on both the performance and robustness of fine-tuned models and that using the German GottBERT model yields a higher performance while the multilingual mBERT results in a more robust model. This study provides valuable insights for researchers and practitioners working with low-resource languages and highlights the importance of considering pre-training strategies when building language models. [less ▲]

Artificial Intelligence-enabled Automation For Ambiguity Handling And Question Answering In Natural-language Requirements

Doctoral thesis (2022)

Requirements Engineering (RE) quality control is a crucial step for a project’s success. Natural Language (NL) is by far the most commonly used means for capturing requirement specifications. Despite ... [more ▼]

Requirements Engineering (RE) quality control is a crucial step for a project’s success. Natural Language (NL) is by far the most commonly used means for capturing requirement specifications. Despite facilitating communication, NL is prone to quality defects, one of the most notable of which is ambiguity. Ambiguous requirements can lead to misunderstandings and eventually result in a system that is different from what is intended, thus wasting time, money, and effort in the process. This dissertation tackles selected quality issues in NL requirements: • Using Domain-specific Corpora for Improved Handling of Ambiguity in Requirements: Syntactic ambiguity types occurring in coordination and prepositional-phrase attachment structures are prevalent in requirements (in our document collection, as we discuss in Chapter 3, 21% and 26% of the requirements are subject to coordination and prepositional-phrase attachment ambiguity analysis, respectively). We devise an automated solution based on heuristics and patterns for improved handling of coordination and prepositional-phrase attachment ambiguity in requirements. As a prerequisite for this research, we further develop a more broadly applicable corpus generator that creates a domain-specific knowledge resource by crawling Wikipedia. • Automated Handling of Anaphoric Ambiguity in Requirements: A Multi-solution Study: Anaphoric ambiguity is another prevalent ambiguity type in requirements. Estimates from the RE literature suggest that nearly 20% of industrial requirements contain anaphora [1, 2]. We conducted a multi-solution study for anaphoric ambiguity handling. Our study investigates six alternative solutions based on three different technologies: (i) off-the-shelf natural language processing (NLP), (ii) recent NLP methods utilizing language models, and (iii) machine learning (ML). • AI-based Question Answering Assistant for Analyzing NL Requirements: Understanding NL requirements requires domain knowledge that is not necessarily shared by all the involved stakeholders. We develop an automated question-answering assistant that supports requirements engineers during requirements inspections and quality assurance. Our solution uses advanced information retrieval techniques and machine reading comprehension models to answer questions from the same requirement specifications document and/or an external domain-specific knowledge resource. All the research components in this dissertation are tool-supported. Our tools are released with open-source licenses to encourage replication and reuse. [less ▲]

TAPHSIR: Towards AnaPHoric Ambiguity Detection and ReSolution In Requirements

in ACM SIGSOFT CONFERENCE ON THE FOUNDATIONS OF SOFTWARE ENGINEERING (2022)

We introduce TAPHSIR -- a tool for anaphoric ambiguity detection and anaphora resolution in requirements. TAPHSIR facilities reviewing the use of pronouns in a requirements specification and revising ... [more ▼]

We introduce TAPHSIR -- a tool for anaphoric ambiguity detection and anaphora resolution in requirements. TAPHSIR facilities reviewing the use of pronouns in a requirements specification and revising those pronouns that can lead to misunderstandings during the development process. To this end, TAPHSIR detects the requirements which have potential anaphoric ambiguity and further attempts interpreting anaphora occurrences automatically. TAPHSIR employs a hybrid solution composed of an ambiguity detection solution based on machine learning and an anaphora resolution solution based on a variant of the BERT language model. Given a requirements specification, TAPHSIR decides for each pronoun occurrence in the specification whether the pronoun is ambiguous or unambiguous, and further provides an automatic interpretation for the pronoun. The output generated by TAPHSIR can be easily reviewed and validated by requirements engineers. TAPHSIR is publicly available on Zenodo (https://doi.org/10.5281/zenodo.5902117). [less ▲]

Using Domain-specific Corpora for Improved Handling of Ambiguity in Requirements

in In Proceedings of the 43rd International Conference on Software Engineering (ICSE'21), Madrid 25-28 May 2021 (2021, May)

Ambiguity in natural-language requirements is a pervasive issue that has been studied by the requirements engineering community for more than two decades. A fully manual approach for addressing ambiguity ... [more ▼]

Ambiguity in natural-language requirements is a pervasive issue that has been studied by the requirements engineering community for more than two decades. A fully manual approach for addressing ambiguity in requirements is tedious and time-consuming, and may further overlook unacknowledged ambiguity – the situation where different stakeholders perceive a requirement as unambiguous but, in reality, interpret the requirement differently. In this paper, we propose an automated approach that uses natural language processing for handling ambiguity in requirements. Our approach is based on the automatic generation of a domain-specific corpus from Wikipedia. Integrating domain knowledge, as we show in our evaluation, leads to a significant positive improvement in the accuracy of ambiguity detection and interpretation. We scope our work to coordination ambiguity (CA) and prepositional-phrase attachment ambiguity (PAA) because of the prevalence of these types of ambiguity in natural-language requirements [1]. We evaluate our approach on 20 industrial requirements documents. These documents collectively contain more than 5000 requirements from seven distinct application domains. Over this dataset, our approach detects CA and PAA with an average precision of 80% and an average recall of 89% (90% for cases of unacknowledged ambiguity). The automatic interpretations that our approach yields have an average accuracy of 85%. Compared to baselines that use generic corpora, our approach, which uses domain-specific corpora, has 33% better accuracy in ambiguity detection and 16% better accuracy in interpretation. [less ▲]