Anindya Iqbal

Toxic interactions during code reviews can undermine teamwork and hinder productivity in software engineering (SE) teams. While prior studies explore toxicity detection and empirical investigation, they lack real-time detoxification tools to support the SE community. To address this gap, we present ToxiShield, a browser extension for GitHub pull requests that is built using three modules: i) Toxicity Filter -- to identify whether a text is toxic, ii) Communication coach -- to facilitate just-in-time fine-grained toxicity categorization with explanations, and iii) The Reframer -- that generates a revised, constructive alternative of a toxic text. For each module, we trained and evaluated multiple deep learning and Large Language Models (LLMs) to identify the best choice. A BERT-based binary detection model, trained on 38,761 code review samples, achieves 98% accuracy and an F1-score of 97% and is the selected one for the Toxicity Filter module. For the Communication Coach, prompt-tuned Claude 3.5 Sonnet achieved the best performance with 39% MCC and 42% F1 in multiclass toxicity classification with detailed reasoning. For Reframer, we evaluated five LLMs using a fine-tuning strategy on a dataset of 10,120 code review comments. The fine-tuned Llama 3.2 model achieves 95.27% style transfer accuracy, 97.03% fluency, 67.07% content preservation, and an 84% J-score. We further validated ToxiShield through a human evaluation using the Technology Acceptance Model with 10 participants, confirming its perceived usefulness and ease of adoption. ToxiShield sets a benchmark for advancing constructive communication in software engineering, driving inclusivity and healthier collaboration in open-source communities.

Generating accurate code review comments remains a significant challenge due to the inherently diverse and non-unique nature of the task output. Large language models pretrained on both programming and natural language data tend to perform well in code-oriented tasks. However, large-scale pretraining is not always feasible due to its environmental impact and project-specific generalizability issues. In this work, first we fine-tune open-source Large language models (LLM) in parameter-efficient, quantized low-rank (QLoRA) fashion on consumer-grade hardware to improve review comment generation. Recent studies demonstrate the efficacy of augmenting semantic metadata information into prompts to boost performance in other code-related tasks. To explore this in code review activities, we also prompt proprietary, closed-source LLMs augmenting the input code patch with function call graphs and code summaries. Both of our strategies improve the review comment generation performance, with function call graph augmented few-shot prompting on the GPT-3.5 model surpassing the pretrained baseline by around 90% BLEU-4 score on the CodeReviewer dataset. Moreover, few-shot prompted Gemini-1.0 Pro, QLoRA fine-tuned Code Llama and Llama 3.1 models achieve competitive results (ranging from 25% to 83% performance improvement) on this task. An additional human evaluation study further validates our experimental findings, reflecting real-world developers' perceptions of LLM-generated code review comments based on relevant qualitative metrics.

Since the beginning of 2020, the novel coronavirus has begun to sweep across the globe. Given the prevalence of smartphones everywhere, many countries across continents also developed COVID-19 contract tracing apps that users can install to get a warning of potential contacts with infected people. Unlike regular apps that undergo detailed requirement analysis, carefully designed development, rigorous testing, contact tracing apps were deployed after rapid development. Therefore such apps may not reach expectations for all end users. Users share their opinions and experience of the usage of the apps in the app store. This paper aims to understand the types of topics users discuss in the reviews of the COVID-19 contact tracing apps across the continents by analyzing the app reviews. We collected all the reviews of 35 COVID-19 contact tracing apps developed by 34 countries across the globe. We group the app reviews into the following geographical regions: Asia, Europe, North America, Latin America, Africa, Middle East, and Australasia (Australia and NZ). We run topic modeling on the app reviews of each region. We analyze the produced topics and their evolution over time by categorizing them into hierarchies and computing the ratings of reviews related to the topics. While privacy could be a concern with such apps, we only find privacy-related topics in Australasia, North America, and Middle East. Topics related to usability and performance of the apps are prevalent across all regions. Users frequently complained about the lack of features, user interface and the negative impact of such apps on their mobile batteries. Still, we also find that many users praised the apps because they helped them stay aware of the potential danger of getting infected. The finding of this study is expected to help app developers utilize their resources to address the reported issues in a prioritized way.

We present a novel approach - CLAA - for API aspect detection in API reviews that utilizes transformer models trained with a supervised contrastive loss objective function. We evaluate CLAA using performance and impact analysis. For performance analysis, we utilized a benchmark dataset on developer discussions collected from Stack Overflow and compare the results to those obtained using state-of-the-art transformer models. Our experiments show that contrastive learning can significantly improve the performance of transformer models in detecting aspects such as Performance, Security, Usability, and Documentation. For impact analysis, we performed empirical and developer study. On a randomly selected and manually labeled 200 online reviews, CLAA achieved 92% accuracy while the SOTA baseline achieved 81.5%. According to our developer study involving 10 participants, the use of 'Stack Overflow + CLAA' resulted in increased accuracy and confidence during API selection. Replication package: https://github.com/disa-lab/Contrastive-Learning-API-Aspect-ASE2023

Background: As improving code review (CR) effectiveness is a priority for many software development organizations, projects have deployed CR analytics platforms to identify potential improvement areas. The number of issues identified, which is a crucial metric to measure CR effectiveness, can be misleading if all issues are placed in the same bin. Therefore, a finer-grained classification of issues identified during CRs can provide actionable insights to improve CR effectiveness. Although a recent work by Fregnan et al. proposed automated models to classify CR-induced changes, we have noticed two potential improvement areas -- i) classifying comments that do not induce changes and ii) using deep neural networks (DNN) in conjunction with code context to improve performances. Aims: This study aims to develop an automated CR comment classifier that leverages DNN models to achieve a more reliable performance than Fregnan et al. Method: Using a manually labeled dataset of 1,828 CR comments, we trained and evaluated supervised learning-based DNN models leveraging code context, comment text, and a set of code metrics to classify CR comments into one of the five high-level categories proposed by Turzo and Bosu. Results: Based on our 10-fold cross-validation-based evaluations of multiple combinations of tokenization approaches, we found a model using CodeBERT achieving the best accuracy of 59.3%. Our approach outperforms Fregnan et al.'s approach by achieving 18.7% higher accuracy. Conclusion: Besides facilitating improved CR analytics, our proposed model can be useful for developers in prioritizing code review feedback and selecting reviewers.

Firebase is the leading Platform-as-a-Service (PaaS) solution in the application development industry. It is globally used by millions of mobile applications and websites. Being a very popular and robust development platform, Firebase has introduced and developed many features and components over the years. Various services and modules are being added to the platform. So, the developers who use Firebase in their applications, need to have versatile knowledge of the platform. It is difficult to acquire an overview of the trends and discussions of any vast and dynamic platform such as Firebase. Therefore, learners, enthusiasts, and often developers have to put in a lot of effort to figure out topics they should explore and focus on. As a large number of application developers frequently use it and engage in various discussions about it in popular community discussion forums for developers, it has become a very promising source for identifying the topics and trends related to Firebase. Stack Overflow is the most popular community for developers’ discussions. It has grown to be a vast ocean of knowledge to explore. In this study, a systematic approach is followed to explore, extract, and analyze Stack Overflow discussions on Firebase to present an overview of the platform. Natural language processing and machine learning methods were used for extracting information from Stack Overflow data. Information collected was reviewed, analyzed, and compiled into this paper by the research group who have practical experiences in the Firebase platform and other relevant software engineering fields.