Mika Mäntylä

Murali Sridharan, Leevi Rantala, Mika Mäntylä

Most Self-Admitted Technical Debt (SATD) research utilizes explicit SATD features such as 'TODO' and 'FIXME' for SATD detection. A closer look reveals several SATD research uses simple SATD ('Easy to Find') code comments without the contextual data (preceding and succeeding source code context). This work addresses this gap through PENTACET (or 5C dataset) data. PENTACET is a large Curated Contextual Code Comments per Contributor and the most extensive SATD data. We mine 9,096 Open Source Software Java projects with a total of 435 million LOC. The outcome is a dataset with 23 million code comments, preceding and succeeding source code context for each comment, and more than 250,000 comments labeled as SATD, including both 'Easy to Find' and 'Hard to Find' SATD. We believe PENTACET data will further SATD research using Artificial Intelligence techniques.

Alexander Naakka, Yuqing Wang, Mika V Mäntylä

Recent advances in large language models (LLMs) have enabled early attempts to automate root cause analysis (RCA) in microservice-based systems (MSS). Yet, prior works typically rely on a linear reasoning process that proceeds along a single diagnostic path. In this paper, we propose LATS-RCA, an LLM-based multi-agent framework for RCA in MSS. LATS-RCA formulates RCA as a reflection-guided tree-structured search using a Language Agent Tree Search algorithm. In LATS-RCA, multiple LLM-driven agents iteratively perform RCA for each microservice by reasoning over its execution logs and performance metrics to collect operational evidence for root cause exploration. Reflection scores derived from intermediate diagnostic states are used to guide the search toward the most likely root cause based on accumulated evidence. We evaluate LATS-RCA on the open-source industrial MSS, Light-OAuth2 (LO2), using a publicly available dataset and in a production microservice environment (Prod) in a case company with substantially higher operational complexity. LO2 is a small-team Java system with a homogeneous technology stack. The results on LO2 show that LATS-RCA achieves high diagnostic accuracy, and we further benchmark its associated computational costs. Compared to LO2, Prod attains lower diagnostic accuracy and incurs higher computational cost. The Prod deployment demonstrates the practical applicability of LATS-RCA in real-world MSS and reflects the challenges introduced by polyglot tech stack, varied logging practices of source components, and multi-factor root-causes by production-scale MSS.

Mika Mäntylä, Yuqing Wang, Jesse Nyyssölä

This paper introduces LogLead, a tool designed for efficient log analysis benchmarking. LogLead combines three essential steps in log processing: loading, enhancing, and anomaly detection. The tool leverages Polars, a high-speed DataFrame library. We currently have Loaders for eight systems that are publicly available (HDFS, Hadoop, BGL, Thunderbird, Spirit, Liberty, TrainTicket, and GC Webshop). We have multiple enhancers with three parsers (Drain, Spell, LenMa), Bert embedding creation and other log representation techniques like bag-of-words. LogLead integrates to five supervised and four unsupervised machine learning algorithms for anomaly detection from SKLearn. By integrating diverse datasets, log representation methods and anomaly detectors, LogLead facilitates comprehensive benchmarking in log analysis research. We show that log loading from raw file to dataframe is over 10x faster with LogLead compared to past solutions. We demonstrate roughly 2x improvement in Drain parsing speed by off-loading log message normalization to LogLead. Our brief benchmarking on HDFS indicates that log representations extending beyond the bag-of-words approach offer limited additional benefits. Tool URL: https://github.com/EvoTestOps/LogLead

Yuqing Wang, Ying Song, Xiaozhou Li et al.

Recent deep learning (DL) methods for log anomaly detection increasingly rely on semantic log representation methods that convert the textual content of log events into vector embeddings as input to DL models. However, these DL methods are typically evaluated as end-to-end pipelines, while the impact of different semantic representation methods is not well understood. In this paper, we benchmark widely used semantic log representation methods, including static word embedding methods (Word2Vec, GloVe, and FastText) and the BERT-based contextual embedding method, across diverse DL models for log-event level anomaly detection on three publicly available log datasets: BGL, Thunderbird, and Spirit. We identify an effectiveness--efficiency trade off under CPU deployment settings: the BERT-based method is more effective, but incurs substantially longer log embedding generation time, limiting its practicality; static word embedding methods are efficient but are generally less effective and may yield insufficient detection performance. Motivated by this finding, we propose QTyBERT, a novel semantic log representation method that better balances this trade-off. QTyBERT uses SysBE, a lightweight BERT variant with system-specific quantization, to efficiently encode log events into vector embeddings on CPUs, and leverages CroSysEh to enhance the semantic expressiveness of these log embeddings. CroSysEh is trained unsupervisedly using unlabeled logs from multiple systems to capture the underlying semantic structure of the BERT model's embedding space. We evaluate QTyBERT against existing semantic log representation methods. Our results show that, for the DL models, using QTyBERT-generated log embeddings achieves detection effectiveness comparable to or better than BERT-generated log embeddings, while bringing log embedding generation time closer to that of static word embedding methods.

Aleksi Huotala, Miikka Kuutila, Mika Mäntylä

Context: Systematic reviews (SRs) summarize state-of-the-art evidence in science, including software engineering (SE). Objective: Our objective is to evaluate how SRs report research artifacts and to provide a comprehensive list of these artifacts. Method: We examined 537 secondary studies published between 2013 and 2023 to analyze the availability and reporting of research artifacts. Results: Our findings indicate that only 31.5% of the reviewed studies include research artifacts. Encouragingly, the situation is gradually improving, as our regression analysis shows a significant increase in the availability of research artifacts over time. However, in 2023, just 62.0% of secondary studies provide a research artifact while an even lower percentage, 30.4% use a permanent repository with a digital object identifier (DOI) for storage. Conclusion: To enhance transparency and reproducibility in SE research, we advocate for the mandatory publication of research artifacts in secondary studies.

Nana Reinikainen, Mika Mäntylä, Yuqing Wang

Assessing the effectiveness of REST API tests in black-box settings can be challenging due to the lack of access to source code coverage metrics and polyglot tech stack. We propose three metrics for capturing average, minimum, and maximum log coverage to handle the diverse test generation results and runtime behaviors over multiple runs. Using log coverage, we empirically evaluate three REST API test generation strategies, Evolutionary computing (EvoMaster v5.0.2), LLMs (Claude Opus 4.6 and GPT-5.2-Codex), and human-written Locust load tests, on Light-OAuth2 authorization microservice system. On average, Claude Opus 4.6 tests uncover 28.4% more unique log templates than human-written tests, whereas EvoMaster and GPT-5.2-Codex find 26.1% and 38.6% fewer, respectively. Next, we analyze combined log coverage to assess complementarity between strategies. Combining human-written tests with Claude Opus 4.6 tests increases total observed log coverage by 78.4% and 38.9% in human-written and Claude tests respectively. When combining Locust tests with EvoMaster the same increases are 30.7% and 76.9% and when using GPT-5.2-Codex 26.1% and 105.6%. This means that the generation strategies exercise largely distinct runtime behaviors. Our future work includes extending our study to multiple systems.

Murali Sridharan, Mikel Robredo, Leevi Rantala et al.

Context. Detecting Self-Admitted Technical Debt (SATD) is crucial for proactive software maintenance. Previous research has primarily targeted detecting and prioritizing SATD, with little focus on the source code afflicted with SATD. Our goal in this work is to connect the SATD comments with source code constructs that surround them. Method. We leverage the extensive SATD dataset PENTACET, containing code comments from over 9000 Java Open Source Software (OSS) repositories. We quantitatively infer where SATD most commonly occurs and which code constructs/statements it most frequently affects. Results and Conclusions. Our large-scale study links over 225,000 SATD comments to their surrounding code, showing that SATD mainly arises in inline code near definitions, conditionals, and exception handling, where developers face uncertainty and trade-offs, revealing it as an intentional signal of awareness during change rather than mere neglect.

Murali Sridharan, Mika Mäntylä, Leevi Rantala

Context: Previous research on software aging is limited with focus on dynamic runtime indicators like memory and performance, often neglecting evolutionary indicators like source code comments and narrowly examining legacy issues within the TD context. Objective: We introduce the concept of Aging Debt (AD), representing the increased maintenance efforts and costs needed to keep software updated. We study AD through Self-Admitted Aging Debt (SAAD) observed in source code comments left by software developers. Method: We employ a mixed-methods approach, combining qualitative and quantitative analyses to detect and measure AD in software. This includes framing SAAD patterns from the source code comments after analysing the source code context, then utilizing the SAAD patterns to detect SAAD comments. In the process, we develop a taxonomy for SAAD that reflects the temporal aging of software and its associated debt. Then we utilize the taxonomy to quantify the different types of AD prevalent in OSS repositories. Results: Our proposed taxonomy categorizes temporal software aging into Active and Dormant types. Our extensive analysis of over 9,000+ Open Source Software (OSS) repositories reveals that more than 21% repositories exhibit signs of SAAD as observed from our gold standard SAAD dataset. Notably, Dormant AD emerges as the predominant category, highlighting a critical but often overlooked aspect of software maintenance. Conclusion: As software volume grows annually, so do evolutionary aging and maintenance challenges; our proposed taxonomy can aid researchers in detailed software aging studies and help practitioners develop improved and proactive maintenance strategies.

Yuqing Wang, Mika Mäntylä, Zihao Liu et al.

The popularity of continuous integration (CI) is increasing as a result of market pressure to release product features or updates frequently. The ability of CI to deliver quality at speed depends on reliable test automation. In this paper, we present an empirical study to observe the effect of test automation maturity (assessed by standard best practices in the literature) on product quality, test automation effort, and release cycle in the CI context of open source projects. We run our test automation maturity survey and got responses from 37 open source java projects. We also mined software repositories of the same projects. The main results of regression analysis reveal that, higher levels of test automation maturity are positively associated with higher product quality (p-value=0.000624) and shorter release cycle (p-value=0.01891); There is no statistically significant evidence of increased test automation effort due to higher levels of test automation maturity and product quality. Thus, we conclude that, a potential benefit of improving test automation maturity (using standard best practices) is product quality improvement and release cycle acceleration in the CI context of open source projects. We encourage future research to extend our findings by adding more datasets with different programming languages and CI tools, closed source projects, and large-scale industrial projects. Our recommendation to practitioners (in the similar CI context) is to utilize standard best practices to improve test automation maturity.

Alireza Haghighatkhah, Mika Mäntylä, Markku Oivo et al.

Two heuristics namely diversity-based (DBTP) and history-based test prioritization (HBTP) have been separately proposed in the literature. Yet, their combination has not been widely studied in continuous integration (CI) environments. The objective of this study is to catch regression faults earlier, allowing developers to integrate and verify their changes more frequently and continuously. To achieve this, we investigated six open-source projects, each of which included several builds over a large time period. Findings indicate that previous failure knowledge seems to have strong predictive power in CI environments and can be used to effectively prioritize tests. HBTP does not necessarily need to have large data, and its effectiveness improves to a certain degree with larger history interval. DBTP can be used effectively during the early stages, when no historical data is available, and also combined with HBTP to improve its effectiveness. Among the investigated techniques, we found that history-based diversity using NCD Multiset is superior in terms of effectiveness but comes with relatively higher overhead in terms of method execution time. Test prioritization in CI environments can be effectively performed with negligible investment using previous failure knowledge, and its effectiveness can be further improved by considering dissimilarities among the tests.

Alireza Haghighatkhah, Mika Mäntylä, Markku Oivo et al.

A classical heuristic in software testing is to reward diversity, which implies that a higher priority must be assigned to test cases that differ the most from those already prioritized. This approach is commonly known as similarity-based test prioritization (SBTP) and can be realized using a variety of techniques. The objective of our study is to investigate whether SBTP is more effective at finding defects than random permutation, as well as determine which SBTP implementations lead to better results. To achieve our objective, we implemented five different techniques from the literature and conducted an experiment using the defects4j dataset, which contains 395 real faults from six real-world open-source Java programs. Findings indicate that running the most dissimilar test cases early in the process is largely more effective than random permutation (Vargha-Delaney A [VDA]: 0.76-0.99 observed using normalized compression distance). No technique was found to be superior with respect to the effectiveness. Locality-sensitive hashing was, to a small extent, less effective than other SBTP techniques (VDA: 0.38 observed in comparison to normalized compression distance), but its speed largely outperformed the other techniques (i.e., it was approximately 5-111 times faster). Our results bring to mind the well-known adage, "don't put all your eggs in one basket". To effectively consume a limited testing budget, one should spread it evenly across different parts of the system by running the most dissimilar test cases early in the testing process.

Maëlick Claes, Mika Mäntylä, Miikka Kuutila et al.

Open source development contains contributions from both hired and volunteer software developers. Identification of this status is important when we consider the transferability of research results to the closed source software industry, as they include no volunteer developers. While many studies have taken the employment status of developers into account, this information is often gathered manually due to the lack of accurate automatic methods. In this paper, we present an initial step towards predicting paid and unpaid open source development using machine learning and compare our results with automatic techniques used in prior work. By relying on code source repository meta-data from Mozilla, and manually collected employment status, we built a dataset of the most active developers, both volunteer and hired by Mozilla. We define a set of metrics based on developers' usual commit time pattern and use different classification methods (logistic regression, classification tree, and random forest). The results show that our proposed method identify paid and unpaid commits with an AUC of 0.75 using random forest, which is higher than the AUC of 0.64 obtained with the best of the previously used automatic methods.

Aleksi Huotala, Miikka Kuutila, Paul Ralph et al.

Systematic review (SR) is a popular research method in software engineering (SE). However, conducting an SR takes an average of 67 weeks. Thus, automating any step of the SR process could reduce the effort associated with SRs. Our objective is to investigate if Large Language Models (LLMs) can accelerate title-abstract screening by simplifying abstracts for human screeners, and automating title-abstract screening. We performed an experiment where humans screened titles and abstracts for 20 papers with both original and simplified abstracts from a prior SR. The experiment with human screeners was reproduced with GPT-3.5 and GPT-4 LLMs to perform the same screening tasks. We also studied if different prompting techniques (Zero-shot (ZS), One-shot (OS), Few-shot (FS), and Few-shot with Chain-of-Thought (FS-CoT)) improve the screening performance of LLMs. Lastly, we studied if redesigning the prompt used in the LLM reproduction of screening leads to improved performance. Text simplification did not increase the screeners' screening performance, but reduced the time used in screening. Screeners' scientific literacy skills and researcher status predict screening performance. Some LLM and prompt combinations perform as well as human screeners in the screening tasks. Our results indicate that the GPT-4 LLM is better than its predecessor, GPT-3.5. Additionally, Few-shot and One-shot prompting outperforms Zero-shot prompting. Using LLMs for text simplification in the screening process does not significantly improve human performance. Using LLMs to automate title-abstract screening seems promising, but current LLMs are not significantly more accurate than human screeners. To recommend the use of LLMs in the screening process of SRs, more research is needed. We recommend future SR studies publish replication packages with screening data to enable more conclusive experimenting with LLM screening.

Yuqing Wang, Mika V. Mäntylä, Serge Demeyer et al.

Microservice-based systems (MSS) may fail with various fault types. While existing AIOps methods excel at detecting abnormal traces and locating the responsible service(s), human efforts are still required for diagnosing specific fault types and failure causes.This paper presents TraFaultDia, a novel AIOps framework to automatically classify abnormal traces into fault categories for MSS. We treat the classification process as a series of multi-class classification tasks, where each task represents an attempt to classify abnormal traces into specific fault categories for a MSS. TraFaultDia leverages meta-learning to train on several abnormal trace classification tasks with a few labeled instances from a MSS, enabling quick adaptation to new, unseen abnormal trace classification tasks with a few labeled instances across MSS. TraFaultDia's use cases are scalable depending on how fault categories are built from anomalies within MSS. We evaluated TraFaultDia on two MSS, TrainTicket and OnlineBoutique, with open datasets where each fault category is linked to faulty system components (service/pod) and a root cause. TraFaultDia automatically classifies abnormal traces into these fault categories, thus enabling the automatic identification of faulty system components and root causes without manual analysis. TraFaultDia achieves 93.26% and 85.20% accuracy on 50 new classification tasks for TrainTicket and OnlineBoutique, respectively, when trained within the same MSS with 10 labeled instances per category. In the cross-system context, when TraFaultDia is applied to a MSS different from the one it is trained on, TraFaultDia gets an average accuracy of 92.19% and 84.77% for the same set of 50 new, unseen abnormal trace classification tasks of the respective systems, also with 10 labeled instances provided for each fault category per task in each system.

Mika Mäntylä, Martín Varela, Shayan Hashemi

As stability testing execution logs can be very long, software engineers need help in locating anomalous events. We develop and evaluate two models for scoring individual log-events for anomalousness, namely an N-Gram model and a Deep Learning model with LSTM (Long short-term memory). Both are trained on normal log sequences only. We evaluate the models with long log sequences of Android stability testing in our company case and with short log sequences from HDFS (Hadoop Distributed File System) public dataset. We evaluate next event prediction accuracy and computational efficiency. The LSTM model is more accurate in stability testing logs (0.848 vs 0.865), whereas in HDFS logs the N-Gram is slightly more accurate (0.904 vs 0.900). The N-Gram model has far superior computational efficiency compared to the Deep model (4 to 13 seconds vs 16 minutes to nearly 4 hours), making it the preferred choice for our case company. Scoring individual log events for anomalousness seems like a good aid for root cause analysis of failing test cases, and our case company plans to add it to its online services. Despite the recent surge in using deep learning in software system anomaly detection, we found limited benefits in doing so. However, future work should consider whether our finding holds with different LSTM-model hyper-parameters, other datasets, and with other deep-learning approaches that promise better accuracy and computational efficiency than LSTM based models.

Yuqing Wang, Mika V. Mäntylä, Zihao Liu et al.

Mature test automation is key for achieving software quality at speed. In this paper, we present a multivocal literature review with the objective to survey and synthesize the guidelines given in the literature for improving test automation maturity. We selected and reviewed 81 primary studies, consisting of 26 academic literature and 55 grey literature sources. From primary studies, we extracted 26 test automation best practices (e.g., Define an effective test automation strategy, Set up good test environments, Develop high-quality test scripts) and collected many pieces of advice (e.g., in forms of implementation/improvement approaches, technical techniques, concepts, experience-based heuristics) on how to conduct these best practices. We made main observations: (1) There are only 6 best practices whose positive effect on maturity improvement have been evaluated by academic studies using formal empirical methods; (2) Several technical related best practices in this MLR were not presented in test maturity models; (3) Some best practices can be linked to success factors and maturity impediments proposed by other scholars; (4) Most pieces of advice on how to conduct proposed best practices were identified from experience studies and their effectiveness need to be further evaluated with cross-site empirical evidence using formal empirical methods; (5) In the literature, some advice on how to conduct certain best practices are conflicting, and some advice on how to conduct certain best practices still need further qualitative analysis.

Miikka Kuutila, Mika Mäntylä, Maëlick et al.

Reports of poor work well-being and fluctuating productivity in software engineering have been reported in both academic and popular sources. Understanding and predicting these issues through repository analysis might help manage software developers' well-being. Our objective is to link data from software repositories, that is commit activity, communication, expressed sentiments, and job events, with measures of well-being obtained with a daily experience sampling questionnaire. To achieve our objective, we studied a single software project team for eight months in the software industry. Additionally, we performed semi-structured interviews to explain our results. The acquired quantitative data are analyzed with generalized linear mixed-effects models with autocorrelation structure. We find that individual variance accounts for most of the $R^2$ values in models predicting developers' experienced well-being and productivity. In other words, using software repository variables to predict developers' well-being or productivity is challenging due to individual differences. Prediction models developed for each developer individually work better, with fixed effects $R^2$ value of up to 0.24. The semi-structured interviews give insights into the well-being of software developers and the benefits of chat interaction. Our study suggests that individualized prediction models are needed for well-being and productivity prediction in software development.

Shayan Hashemi, Mika Mäntylä

With the growth of online services, IoT devices, and DevOps-oriented software development, software log anomaly detection is becoming increasingly important. Prior works mainly follow a traditional four-staged architecture (Preprocessor, Parser, Vectorizer, and Classifier). This paper proposes OneLog, which utilizes a single Deep Neural Network (DNN) instead of multiple separate components. OneLog harnesses Convolutional Neural Networks (CNN) at the character level to take digits, numbers, and punctuations, which were removed in prior works, into account alongside the main natural language text. We evaluate our approach in six message- and sequence-based data sets: HDFS, Hadoop, BGL, Thunderbird, Spirit, and Liberty. We experiment with Onelog with single-, multi-, and cross-project setups. Onelog offers state-of-the-art performance in our datasets. Onelog can utilize multi-project datasets simultaneously during training, which suggests our model can generalize between datasets. Multi-project training also improves Onelog performance making it ideal when limited training data is available for an individual project. We also found that cross-project anomaly detection is possible with a single project pair (Liberty and Spirit). Analysis of model internals shows that one log has multiple modes of detecting anomalies and that the model learns manually validated parsing rules for the log messages. We conclude that character-based CNNs are a promising approach toward end-to-end learning in log anomaly detection. They offer good performance and generalization over multiple datasets. We will make our scripts publicly available upon the acceptance of this paper.

Murali Sridharan, Mika Mantyla, Leevi Rantala et al.

A high imbalance exists between technical debt and non-technical debt source code comments. Such imbalance affects Self-Admitted Technical Debt (SATD) detection performance, and existing literature lacks empirical evidence on the choice of balancing technique. In this work, we evaluate the impact of multiple balancing techniques, including Data level, Classifier level, and Hybrid, for SATD detection in Within-Project and Cross-Project setup. Our results show that the Data level balancing technique SMOTE or Classifier level Ensemble approaches Random Forest or XGBoost are reasonable choices depending on whether the goal is to maximize Precision, Recall, F1, or AUC-ROC. We compared our best-performing model with the previous SATD detection benchmark (cost-sensitive Convolution Neural Network). Interestingly the top-performing XGBoost with SMOTE sampling improved the Within-project F1 score by 10% but fell short in Cross-Project set up by 9%. This supports the higher generalization capability of deep learning in Cross-Project SATD detection, yet while working within individual projects, classical machine learning algorithms can deliver better performance. We also evaluate and quantify the impact of duplicate source code comments in SATD detection performance. Finally, we employ SHAP and discuss the interpreted SATD features. We have included the replication package and shared a web-based SATD prediction tool with the balancing techniques in this study.

Shayan Hashemi, Mika Mäntylä

Logs are semi-structured text files that represent software's execution paths and states during its run-time. Therefore, detecting anomalies in software logs reflect anomalies in the software's execution path or state. So, it has become a notable concern in software engineering. We use LSTM like many prior works, and on top of LSTM, we propose a novel anomaly detection approach based on the Siamese network. This paper also provides an authentic validation of the approach on the Hadoop Distributed File System (HDFS) log dataset. To the best of our knowledge, the proposed approach outperforms other methods on the same dataset at the F1 score of 0.996, resulting in a new state-of-the-art performance on the dataset. Along with the primary method, we introduce a novel training pair generation algorithm that reduces generated training pairs by the factor of 3000 while maintaining the F1 score, merely a modest decay from 0.996 to 0.995. Additionally, we propose a hybrid model by combining the Siamese network with a traditional feedforward neural network to make end-to-end training possible, reducing engineering effort in setting up a deep-learning-based log anomaly detector. Furthermore, we examine our method's robustness to log evolutions by evaluating the model on synthetically evolved log sequences; we got the F1 score of 0.95 at the noise ratio of 20%. Finally, we dive deep into some of the side benefits of the Siamese network. Accordingly, we introduce a method of monitoring the evolutions of logs without label requirements at run-time. Additionally, we present a visualization technique that facilitates human administrations of log anomaly detection.

Leevi Rantala, Mika Mäntylä, David Lo

When developers use different keywords such as TODO and FIXME in source code comments to describe self-admitted technical debt (SATD), we refer it as Keyword-Labeled SATD (KL-SATD). We study KL-SATD from 33 software repositories with 13,588 KL-SATD comments. We find that the median percentage of KL-SATD comments among all comments is only 1,52%. We find that KL-SATD comment contents include words expressing code changes and uncertainty, such as remove, fix, maybe and probably. This makes them different compared to other comments. KL-SATD comment contents are similar to manually labeled SATD comments of prior work. Our machine learning classifier using logistic Lasso regression has good performance in detecting KL-SATD comments (AUC-ROC 0.88). Finally, we demonstrate that using machine learning we can identify comments that are currently missing but which should have a SATD keyword in them. Automating SATD identification of comments that lack SATD keywords can save time and effort by replacing manual identification of comments. Using KL-SATD offers a potential to bootstrap a complete SATD detector.

Yuqing Wang, Maaret Pyhäjärvi, Mika V. Mäntylä

How to successfully conduct test automation process improvement (TAPI) for continuous development, consisting of iterative software development, continuous testing, and delivery, is the challenge faced by many software organizations. In this paper, we present an experience report on TAPI in one DevOps team in F-Secure (a Finnish software company). The team builds Windows application software and exists in F-Secure's TAPI culture. The team self-reports high satisfaction and maturity in test automation for continuous development. To study their TAPI, we reviewed a collection of experience notes, team reflection reports and telemetry result reports. Then several meetings were held to discuss the details. We found that based on the understanding of the team, test automation maturity for continuous development is defined as a set of indicators, e.g., the increasing speed to release, improving the productivity of the team, high test efficiency. Second, the team indicated that a set of critical success factors have a major impact on successfully carrying out its TAPI, e.g., incremental approach, the whole team effort, test tool choice and architecture, telemetry. Third, we compare the TAPI practices in the observed team with the practices described in prior literature. The team believes that the existing test automation maturity approaches should include the identified practices like the whole team effort to build a more comprehensive test automation improvement model for the software industry.

Maëlick Claes, Mika Mäntylä

Data of long-lived and high profile projects is valuable for research on successful software engineering in the wild. Having a dataset with different linked software repositories of such projects, enables deeper diving investigations. This paper presents 20-MAD, a dataset linking the commit and issue data of Mozilla and Apache projects. It includes over 20 years of information about 765 projects, 3.4M commits, 2.3M issues, and 17.3M issue comments, and its compressed size is over 6 GB. The data contains all the typical information about source code commits (e.g., lines added and removed, message and commit time) and issues (status, severity, votes, and summary). The issue comments have been pre-processed for natural language processing and sentiment analysis. This includes emoticons and valence and arousal scores. Linking code repository and issue tracker information, allows studying individuals in two types of repositories and provide more accurate time zone information for issue trackers as well. To our knowledge, this the largest linked dataset in size and in project lifetime that is not based on GitHub.

Vahid Garousi, Michael Felderer, Mika V. Mäntylä et al.

Researchers generally place the most trust in peer-reviewed, published information, such as journals and conference papers. By contrast, software engineering (SE) practitioners typically do not have the time, access or expertise to review and benefit from such publications. As a result, practitioners are more likely to turn to other sources of information that they trust, e.g., trade magazines, online blog-posts, survey results or technical reports, collectively referred to as Grey Literature (GL). Furthermore, practitioners also share their ideas and experiences as GL, which can serve as a valuable data source for research. While GL itself is not a new topic in SE, using, benefitting and synthesizing knowledge from the GL in SE is a contemporary topic in empirical SE research and we are seeing that researchers are increasingly benefitting from the knowledge available within GL. The goal of this chapter is to provide an overview to GL in SE, together with insights on how SE researchers can effectively use and benefit from the knowledge and evidence available in the vast amount of GL.

Mika Mäntylä, Vahid Garousi

Many factors could affect the number of citations to a paper. Citations have an important role in research policy and in measuring the excellence of research and researchers. This work is the first study in software engineering (SE) to assess multiple factors affecting the number of citations to SE papers. We use (a) negative binomial regression and (b) quantile regression to study arithmetic mean and median expected citations of a paper. Our dataset includes all the 25,113 papers which have been published in a set of 16 main SE journals, between 1970 and 2018. Our results indicate that publication venue, author team's past citations, paper length, the number of references, and the recency of references are the most influential factors on the number of citations to SE papers. From our empirical findings, we present several implications and advice to researchers for getting higher citations on their papers, which are in addition to the obvious case of conducting high-quality technical research, e.g. (1) Aim for high-profile venues, (2) Build a high-quality author team with highly cited past papers, and (3) Aim for high-quality work that has comprehensive content (thus longer paper length and reference list).

Miikka Kuutila, Mika Mäntylä, Umar Farooq et al.

Large project overruns and overtime work have been reported in the software industry, resulting in additional expense for companies and personal issues for developers. The present work aims to provide an overview of studies related to time pressure in software engineering; specifically, existing definitions, possible causes, and metrics relevant to time pressure were collected, and a mapping of the studies to software processes and approaches was performed. Moreover, we synthesize results of existing quantitative studies on the effects of time pressure on software development, and offer practical takeaways for practitioners and researchers, based on empirical evidence. Our search strategy examined 5,414 sources, found through repository searches and snowballing. Applying inclusion and exclusion criteria resulted in the selection of 102 papers, which made relevant contributions related to time pressure in software engineering. The majority of high quality studies report increased productivity and decreased quality under time pressure. Frequent categories of studies focus on quality assurance, cost estimation, and process simulation. It appears that time pressure is usually caused by errors in cost estimation. The effect of time pressure is most often identified during software quality assurance. The majority of empirical studies report increased productivity under time pressure, while the most cost estimation and process simulation models assume that compressing the schedule increases the total needed hours. We also find evidence of the mediating effect of knowledge on the effects of time pressure, and that tight deadlines impact tasks with an algorithmic nature more severely. Future research should better contextualize quantitative studies to account for the existing conflicting results and to provide an understanding of situations when time pressure is either beneficial or harmful.

Mika Mäntylä, Maëlick Claes, Umar Farooq

Background: Unstructured and textual data is increasing rapidly and Latent Dirichlet Allocation (LDA) topic modeling is a popular data analysis methods for it. Past work suggests that instability of LDA topics may lead to systematic errors. Aim: We propose a method that relies on replicated LDA runs, clustering, and providing a stability metric for the topics. Method: We generate k LDA topics and replicate this process n times resulting in n*k topics. Then we use K-medioids to cluster the n*k topics to k clusters. The k clusters now represent the original LDA topics and we present them like normal LDA topics showing the ten most probable words. For the clusters, we try multiple stability metrics, out of which we recommend Rank-Biased Overlap, showing the stability of the topics inside the clusters. Results: We provide an initial validation where our method is used for 270,000 Mozilla Firefox commit messages with k=20 and n=20. We show how our topic stability metrics are related to the contents of the topics. Conclusions: Advances in text mining enable us to analyze large masses of text in software engineering but non-deterministic algorithms, such as LDA, may lead to unreplicable conclusions. Our approach makes LDA stability transparent and is also complementary rather than alternative to many prior works that focus on LDA parameter tuning.

Miikka Kuutila, Mika Mäntylä, Maëlick Claes et al.

Background: The experience sampling method studies everyday experiences of humans in natural environments. In psychology it has been used to study the relationships between work well-being and productivity. To our best knowledge, daily experience sampling has not been previously used in software engineering. Aims: Our aim is to identify links between software developers self-reported affective states and work well-being and measures obtained from software repositories. Method: We perform an experience sampling study in a software company for a period of eight months, we use logistic regression to link the well-being measures with development activities, i.e. number of commits and chat messages. Results: We find several significant relationships between questionnaire variables and software repository variables. To our surprise relationship between hurry and number of commits is negative, meaning more perceived hurry is linked with a smaller number of commits. We also find a negative relationship between social interaction and hindered work well-being. Conclusions: The negative link between commits and hurry is counter-intuitive and goes against previous lab-experiments in software engineering that show increased efficiency under time pressure. Overall, our work is an initial step in using experience sampling in software engineering and validating theories on work well-being from other fields in the domain of software engineering.

Mika V. Mäntylä, Fabio Calefato, Maelick Claes

The use of natural language processing (NLP) is gaining popularity in software engineering. In order to correctly perform NLP, we must pre-process the textual information to separate natural language from other information, such as log messages, that are often part of the communication in software engineering. We present a simple approach for classifying whether some textual input is natural language or not. Although our NLoN package relies on only 11 language features and character tri-grams, we are able to achieve an area under the ROC curve performances between 0.976-0.987 on three different data sources, with Lasso regression from Glmnet as our learner and two human raters for providing ground truth. Cross-source prediction performance is lower and has more fluctuation with top ROC performances from 0.913 to 0.980. Compared with prior work, our approach offers similar performance but is considerably more lightweight, making it easier to apply in software engineering text mining pipelines. Our source code and data are provided as an R-package for further improvements.

Vahid Garousi, Michael Felderer, Mika V. Mäntylä

Context: A Multivocal Literature Review (MLR) is a form of a Systematic Literature Review (SLR) which includes the grey literature (e.g., blog posts and white papers) in addition to the published (formal) literature (e.g., journal and conference papers). MLRs are useful for both researchers and practitioners since they provide summaries both the state-of-the art and -practice in a given area. Objective: There are several guidelines to conduct SLR studies in SE. However, given the facts that several phases of MLRs differ from those of traditional SLRs, for instance with respect to the search process and source quality assessment. Therefore, SLR guidelines are only partially useful for conducting MLR studies. Our goal in this paper is to present guidelines on how to conduct MLR studies in SE. Method: To develop the MLR guidelines, we benefit from three inputs: (1) existing SLR guidelines in SE, (2), a literature survey of MLR guidelines and experience papers in other fields, and (3) our own experiences in conducting several MLRs in SE. All derived guidelines are discussed in the context of three examples MLRs as running examples (two from SE and one MLR from the medical sciences). Results: The resulting guidelines cover all phases of conducting and reporting MLRs in SE from the planning phase, over conducting the review to the final reporting of the review. In particular, we believe that incorporating and adopting a vast set of recommendations from MLR guidelines and experience papers in other fields have enabled us to propose a set of guidelines with solid foundations. Conclusion: Having been developed on the basis of three types of solid experience and evidence, the provided MLR guidelines support researchers to effectively and efficiently conduct new MLRs in any area of SE.

Miikka Kuutila, Mika V. Mäntylä, Maëlick Claes et al.

During the past years, psychological diseases related to unhealthy work environments, such as burnouts, have drawn more and more public attention. One of the known causes of these affective problems is time pressure. In order to form a theoretical background for time pressure detection in software repositories, this paper combines interdisciplinary knowledge by analyzing 1270 papers found on Scopus database and containing terms related to time pressure. By clustering those papers based on their abstract, we show that time pressure has been widely studied across different fields, but relatively little in software engineering. From a literature review of the most relevant papers, we infer a list of testable hypotheses that we want to verify in future studies in order to assess the impact of time pressures on software developers mental health.

Miikka Kuutila, Mika Mäntylä, Päivi Raulamo-Jurvanen

Context: Selenium is claimed to be the most popular software test automation tool. Past academic works have mainly neglected testing tools in favor of more methodological topics. Objective: We investigated the performance of web-testing tools, to provide empirical evidence supporting choices in software test tool selection and configuration. Method: We used 4*5 factorial design to study 20 different configurations for testing a web-store. We studied 5 programming language bindings (C#, Java, Python, and Ruby for Selenium, while Watir supports Ruby only) and 4 browsers (Google Chrome, Internet Explorer, Mozilla Firefox and Opera). Performance was measured with execution time, memory usage, length of the test scripts and stability of the tests. Results: Considering all measures the best configuration was Selenium with Python language binding for Chrome. Selenium with Python bindings was the best option for all browsers. The effect size of the difference between the slowest and fastest configuration was very high (Cohens d=41.5, 91% increase in execution time). Overall Internet Explorer was the fastest browser while having the worst results in the stability. Conclusions: We recommend benchmarking tools before adopting them. Weighting of factors, e.g. how much test stability is one willing to sacrifice for faster performance, affects the decision.

Mika Mäntylä, Bram Adams, Giuseppe Destefanis et al.

Similar to other industries, the software engineering domain is plagued by psychological diseases such as burnout, which lead developers to lose interest, exhibit lower activity and/or feel powerless. Prevention is essential for such diseases, which in turn requires early identification of symptoms. The emotional dimensions of Valence, Arousal and Dominance (VAD) are able to derive a person's interest (attraction), level of activation and perceived level of control for a particular situation from textual communication, such as emails. As an initial step towards identifying symptoms of productivity loss in software engineering, this paper explores the VAD metrics and their properties on 700,000 Jira issue reports containing over 2,000,000 comments, since issue reports keep track of a developer's progress on addressing bugs or new features. Using a general-purpose lexicon of 14,000 English words with known VAD scores, our results show that issue reports of different type (e.g., Feature Request vs. Bug) have a fair variation of Valence, while increase in issue priority (e.g., from Minor to Critical) typically increases Arousal. Furthermore, we show that as an issue's resolution time increases, so does the arousal of the individual the issue is assigned to. Finally, the resolution of an issue increases valence, especially for the issue Reporter and for quickly addressed issues. The existence of such relations between VAD and issue report activities shows promise that text mining in the future could offer an alternative way for work health assessment surveys.