PROMISE 2025

The International Conference on Predictive Models and Data Analytics in Software Engineering (PROMISE) is an annual forum for researchers and practitioners to present, discuss and exchange ideas, results, expertise and experiences in construction and/or application of predictive models, artificial intelligence, and data analytics in software engineering. PROMISE encourages researchers to publicly share their data in order to provide interdisciplinary research between the software engineering and data mining communities, and seek for verifiable and repeatable experiments that are useful in practice.

Keynote 1 by Dr. Jacques Klein, University of Luxembourg, Luxembourg

Title: Datasets, AI, and Static Analysis for Efficient Mobile App Analysis

Abstract: Users can today download a wide variety of apps ranging from simple toy games to sophisticated business-critical apps. They rely on these apps daily to perform diverse tasks, some of them related to sensitive information such as their finance or health. Ensuring high-quality, reliable, and secure software is thus key. In the TruX research group of the Interdisciplinary Center for Security, Reliability, and Trust (SnT) of the University of Luxembourg, we have been working for over 15 years to deliver practical techniques, tools, and other artifacts (such as repositories), making the analysis of Android apps possible. In this talk, I will briefly introduce our key contributions to Android app analysis, leveraging techniques such as static analysis and artificial intelligence.

Bio: Dr. Jacques Klein is a full professor in software engineering and software security within the Interdisciplinary Centre for Security, Reliability and Trust (SnT) at the University of Luxembourg. Prof. Klein co-leads a team named TruX of about 35 researchers developing innovative approaches and tools for helping the research and practice communities build trustworthy and secure software. Prof. Klein received a Ph.D. degree in Computer Science from the University of Rennes, France, in 2006. His main areas of expertise are threefold: (1) Software Security, (2) Software Reliability, and (3) Data Analytics. Prof. Klein received multiple most influential papers and has published over 200 papers, often in top venues such as ICSE, FSE, ASE, ISSTA, PLDI, etc. In addition to academic achievements, Prof. Klein also has long-standing experience and expertise in successfully running industrial projects with several industrial partners in various domains by applying AI, software engineering, information retrieval, etc., to their research problems.

Keynote 2 by Dr. Haipeng Cai, University at Buffalo, The State University of New York (SUNY), USA

Title: The Data Quest in Software Vulnerability Analysis: Hope, Challenges, and New Horizon

Abstract: In the past decade, machine‐learning techniques have transformed software vulnerability analysis, driven by manually curated datasets—from CVE records to mined fix pairs—and boosted further through synthetic sample augmentation. Yet despite these advances, persistent data‐quality issues (label noise, oversimplified injections, limited CWE scope, and narrow code contexts) leave models struggling to generalize to complex, real‐world or zero‐day vulnerabilities. At the same time, large language models (LLMs) have demonstrated surprising promises in this space, prompting fresh questions about where and how data curation still matters.

In this talk, I will first highlight the hope born of early dataset successes and augmentation techniques that improved vulnerability detection and repair (e.g., using automatically synthesized vulnerable code samples). Next, I’ll reflect on the challenges posed by label inaccuracies, lack of diversity, and poor representativeness that undermine current deep‐learning-based vulnerability analysis models (e.g., suffering poor replicability). Finally, I’ll touch on the new horizon of LLM driven vulnerability analysis—exploring how task specific fine tuning and the next generation of high quality, diverse datasets can unlock truly generalizable, reliable security tools (e.g., general-purpose LLMs often underperform in specialized coding tasks, such as vulnerability analysis, while harnessing task-specific LLMs through fine-tuning still necessitates high-quality, diverse, and contextualized datasets).

Bio: Dr. Haipeng Cai is an Associate Professor in the Department of Computer Science and Engineering at University at Buffalo, SUNY. His research generally lies in software engineering, program analysis, and software security, with a current focus on adaptive/data-driven static and dynamic analysis for security applications to mobile apps, distributed systems, and multilingual software. His research has been recognized by a couple of distinguished/outstanding paper awards, including the inaugural TOSEM Outstanding Paper (TOP) award for his pioneering introduction of the malware detection sustainability concept and its impact in extending evaluation practices in mobile security research. His professional services to the software engineering and computer security communities have been acknowledged by several distinguished reviewer awards (for TOSEM’20, TSE’24, NDSS’24, NDSS’25, and FSE’25).

Accepted Papers

Call for Papers

The International Conference on Predictive Models and Data Analytics in Software Engineering (PROMISE) welcomes four types of submissions:

Technical papers (10 pages)

• PROMISE accepts a wide range of papers where AI tools have been applied to SE such as predictive modeling and other AI methods. Both positive and negative results are welcome, though negative results should still be based on rigorous research and provide details on lessons learned.

Industrial papers (2-4 pages)

• Results, challenges, lessons learned from industrial applications of software analytics.

New idea papers (2-4 pages)

• Novel insights or ideas that may yet to be fully tested.

Journal First

• Selected papers will be invited for journal first presentations at PROMISE. Details to follow.

Topics of Interest

PROMISE papers can explore any of the following topics (or more).

Application-oriented papers:

• prediction of cost, effort, quality, defects, business value;

• quantification and prediction of other intermediate or final properties of interest in software development regarding people, process or product aspects;

• using predictive models and data analytics in different settings, e.g. lean/agile, waterfall, distributed, community-based software development;

• dealing with changing environments in software engineering tasks;

• dealing with multiple-objectives in software engineering tasks;

• using predictive models and software data analytics in policy and decision-making.

Ethically-aligned papers:

• Can we apply and adjust our AI-for-SE tools (including predictive models) to handle ethical non-functional requirements such as inclusiveness, transparency, oversight and accountability, privacy, security, reliability, safety, diversity and fairness?

Theory-oriented papers:

• model construction, evaluation, sharing and reusability;

• interdisciplinary and novel approaches to predictive modelling and data analytics that contribute to the theoretical body of knowledge in software engineering;

• verifying/refuting/challenging previous theory and results;

• combinations of predictive models and search-based software engineering;

• the effectiveness of human experts vs. automated models in predictions.

Data-oriented papers:

• data quality, sharing, and privacy;

• curated data sets made available for the community to use;

ethical issues related to data collection and sharing;

• metrics;

• tools and frameworks to support researchers and practitioners to collect data and construct models to share/repeat experiments and results.

Validity-oriented papers:

• replication and repeatability of previous work using predictive modelling and data analytics in software engineering;

• assessment of measurement metrics for reporting the performance of predictive models;

• evaluation of predictive models with industrial collaborators.

Submissions

PROMISE 2025 submissions must meet the following criteria:

• be original work, not published or under review elsewhere while being considered;

• conform to the submission format requirements of the FSE 2025 Companion proceedings;

• not exceed 10 (4) pages for technical (industrial, new-ideas) papers including references;

• be written in English;

• be prepared for double blind review

  • Exception: for data-oriented papers, authors may elect not to use double blind by placing a footnote on page 1 saying “Offered for single-blind review”.

• be submitted via HotPRC.

• on submission, please choose the paper category appropriately, i.e., technical (main track, 10 pages max); industrial (4 pages max); and new idea papers (4 pages max).

• for Industrial papers and New Idea papers, please clearly indicate the paper category in the keywords below the abstract.

To satisfy the double blind requirement submissions must meet the following criteria:

• no author names and affiliations in the body and metadata of the submitted paper;

• self-citations are written in the third person;

• no references to the authors personal, lab, or university website;

• no references to personal accounts on GitHub, bitbucket, Google Drive, etc.

Evaluation

Submissions will be peer reviewed by at least three experts from the international program committee. Submissions will be evaluated on the basis of their originality, importance of contribution, soundness, evaluation, quality, and consistency of presentation, and appropriate comparison to related work.

Important Dates

• Abstracts due: Feb 18th, 2025 AoE

• Submissions due: Feb 25th, 2025 AoE Feb 28th, 2025 AoE

• Author notification: Mar 24th, 2025 AoE

• Camera ready: Apr 24th, 2025 AoE

• Conference Date: Jun 26th, 2025 AoE

Green Open Access

Similar to other leading SE conferences, PROMISE supports and encourages Green Open Access, i.e., self-archiving. Authors can archive their papers on their personal home page, an institutional repository of their employer, or at an e-print server such as arXiv (preferred). Also, given that PROMISE papers heavily rely on software data, we would like to draw authors that leverage data scraped from GitHub of GitHub’s Terms of Service, which require that “publications resulting from that research are open access”.

We also strongly encourage authors to submit their tools and data to Zenodo, which adheres to FAIR (findable, accessible, interoperable and re-usable) principles and provides DOI versioning.

Journal Special Section

Following the conference, the authors of the best papers will be invited to submit extended versions of their papers for consideration in a special section in the journal Empirical Software Engineering (EMSE).

Publication and Attendance

Accepted papers will be published in the ACM Digital Library within its International Conference Proceedings Series and will be available electronically via ACM Digital Library.

Each accepted paper needs to have one registration at the full conference rate and be presented in person at the conference.