An Improved Quantum Software Challenges Classification Approach using Transfer Learning and Explainable AI

*, *, *, *, *, *, * · 2025

Read An Improved Quantum Software Challenges Classification Approach using Transfer Learning and Explainable AI by *, *, *, *, *, *, * (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-21068.txt.

Introduction

Quantum software developers air their problems on Stack Overflow, and those posts reveal where the field struggles. Nek Dil Khan and six co-authors classify those challenges in a 2025 paper on quantum software engineering. They aim to improve automatic classification using transfer learning and explainable AI.

The team extracted nearly three thousand questions using quantum-related tags. Reading the posts, they built a grounded theory of challenge types: tooling, theoretical, learning, conceptual, errors, and API usage, among others. Existing tags often name technologies rather than the developer's actual difficulty. A classifier trained on the new scheme could route questions and expose frequent pain points. The authors argue better categorization helps researchers and tool builders serve a young, fast-moving developer community.

Catalog record

CREATOR*, *, *, *, *, *, *
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE105 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-21068

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An Improved Quantum Software Challenges Classification Approach using Transfer Learning and Explainable AI *, *, *, *, *, *, * Abstract Quantum Software Engineering (QSE) has emerged as a research direction practiced by tech joints. However, quantum developers face challenges in optimizing quantum computing and QSE concepts. Quantum developers use Stack Overflow (SO) to discuss quantum-related challenges and use specialized quantum tags to label posts in developers’ forums. These tags often refer to technical quantum aspects rather than developer posts. Categorizing quantum practitioners’ questions based on quantum concepts can help developers identify frequent QSE challenges. We conducted qualitative and quantitative studies to classify quantum developers’ questions into various challenges. We extracted 2829 developers’ questions from Q&A platforms using quantum-related tags. The posts were analyzed to identify frequent quantum-related challenges and develop a novel grounded theory. The challenges identified include Tooling, Theoretical, Learning, Conceptual, Errors, and API Usage. Through content analysis and grounded theory, the developers’ discussions were annotated with commonly reported quantum challenges to develop a ground truth dataset. ChatGPT was used to validate human annotations and resolve disagreements. Various fine-tuned transformer algorithms, including BERT, DistilBERT, and RoBERTa, were used to classify developer discussions into commonly reported quantum challenges. We achieved an average accuracy of 95% with BERT DistilBERT algorithms, compared to fine-tuned Deep and Machine Learning (D&ML) classifiers, including Feedforward Neural Networks (FNN), Convolutional Neural Networks (CNN), and Long Short-Term Memory networks (LSTM), which achieved accuracies of 89%, 86%, and 84%, respectively. The proposed Transformer-based approach outperforms the previous D&ML-based approach with a 6% increase in accuracy by processing actual developer discussions, i.e., without data augmentation. Furthermore, we applied SHAP (SHapley Additive exPlanations) to provide model interpretability, revealing how specific linguistic features drive predictions and enhancing transparency in the classification process. These improved research findings can help quantum vendors and developers’ discussion forums to better organize developers’ discussions for improved access and readability. However, it further needs empirical evaluation studies with actual quantum developers and vendors. keywords Quantum Software Engineering, Repository mining, developer forums, Stack overflow, Transfer Learning, Natural language processing. † † fcssetup: corr-email = [email protected], [1]Nek Dil Khan [2]Javed Ali Khan [3]Mobashir Husain [3]Muhammad Sohail Khan [4]Arif Ali Khan [5]Muhammad Azeem Akbar [6]Shahid Hussain 1 Introduction Quantum computing leverages quantum mechanics principles to describe the behavior of particles at the atomic and subatomic levels. Unlike classical computers, which use bits (0s and 1s), quantum computers use quantum bits or ”qubits” to process information [ 1 ] . Superposition allows qubits to exist in a combination of 0 and 1 states simultaneously, increasing the computational power of quantum systems [ 2 ] . Google [ 3 ] , IBM [ 4 ] , and Rigetti [ 5 ] have demonstrated the ability to create qubits in controlled environments, enabling practical quantum computing. This provides opportunities to solve complex computational problems requiring parallel computation, which is challenging for classical computers. It can resolve complex issues in big data analytics, financial systems, national security, chemistry, cryptography, health analytics, and medicine [ 6 ] . Recently, there has been an increasing interest from the research community in proposing various approaches across different QSE activities [ 7 ] . Where QSE is a discipline emphasizing designing, developing, and optimizing software for a wide range of applications [ 8 ] . QSE processes must be aligned with hardware advancements to leverage the capabilities of quantum computing effectively. Developing methodologies, processes, tools, and algorithms is crucial for quantum computing domains [ 9 ] . Quantum programming languages like Q# [ 10 ] , Scaffold [ 11 ] , and Qiskit [ 12 ] exist for developing quantum applications. Microsoft, Google, and IBM have developed cloud platforms that support the development of quantum software. IBM’s platform enables the execution of applications on quantum systems. Developers create quantum software for domains such as chemistry and machine learning [ 13 ] . Given the importance of applications, software engineering processes are required for large projects. The QSE was coined to meet the demands for large applications [ 9 ] . Initial approaches for quantum requirements engineering [ 14 ] , quantum software architecture [ 15 ] , Quantum programming [ 16 ] , Quantum Development [ 17 ] , and Quantum Testing [ 18 ] are emerging. Software practitioners use developers’ discussion forums to extract information and address challenges associated with software evolution [ 19 ] [ 20 ] . For example, researchers have analyzed Q&A platform discussions to identify challenges related to software development methodologies [ 21 ] . Also, De Dieu et al. [ 20 ] explore Q&A for extracting software architecture related information. Similarly, Li et al. [ 22 ] explore Q&A for extracting requirements-related information. Moreover, El-Aoun used Q&A to identify frequently occurring challenges to the quantum developers community [ 13 ] . These forums offer resources, communities, and opportunities for problem-solving. Q&A are valuable for sharing experiences in less-resourced areas, such as QSE, blockchain, etc. They enable collaboration, updates on trends, and collective problem-solving. The developers’ posts in these Q&A platforms are categorized based on specific tags, which we believe they hard and challenging for developers to remember. Particularly, the more emerging and comparatively new domains, i.e., QSE. Therefore, this motivated us to contextually analyze the developers’ questions to identify their best possible category and add it to the existing tags associated with it from improved searching and visibility in the Q&A forums, which is aligned with previous research [ 23 ] . In this study, we evaluate developer discussions on QSE across various Q&A platforms like Stack Overflow (SO), Quantum Computing Stack Exchange (QCSE), Computer Science Stack Exchange (CSSE), and Artificial Intelligence Stack Exchange (AISE) to understand their contextual category, which are often reported as challenges by software quantum practitioners. This aids quantum developers in finding possible solutions to specific QSE-related challenges on these platforms that are reported by other quantum practitioners and are frequently organized by tags such as qiskit, qcl, qutip, qubit, and TensorFlow Quantum. For this purpose, we extracted 2829 developers’ questions using quantum-related tags, following El-Aoun et al.’s method [ 13 ] . These developers’ posts were manually critically analyzed to identify frequently discussed quantum challenges and develop a novel grounded theory, which was cross-validated with that of El-Aoun et al. [ 13 ] who identified QSE challenges using topic modeling. The frequent challenges include Tooling, Theoretical, Learning, Conceptual, Errors, and API Usage. Next, using content analysis and the novel QSE grounded theory document, the developers’ discussions in the dataset were annotated with common quantum challenges to create a ground-truth dataset for the D&ML and transfer learning (TL) algorithms. Moreover, to validate the annotation process, ChatGPT was utilized to annotate developers’ discussions. Additionally, another cycle was run with ChatGPT to resolve disagreements between the human annotators and ChatGPT, resulting in a conflict-free dataset. Finally, in this approach, in addition to the various fine-tuned D&ML classifiers, fine-tuned transformer learning algorithms, including BERT, DistilBERT, and RoBerta, were used to contextually analyze developer discussions and classify them into common quantum challenges. We extended the work published in the Proceedings of the 33rd ACM International Conference on the Foundations of Software Engineering [ 24 ] , with the following additions. •

Read the complete text