DaQuaMRec welcomes contributions that advance our understanding of data quality issues in multimodal recommender systems. As recommendation models increasingly rely on combinations of text, images, audio, video, and other heterogeneous signals, they enable richer personalization and more accurate recommendations across domains such as fashion, food, music, e-commerce, and digital media. At the same time, these systems are highly sensitive to the quality of the data they consume. Issues such as noise, missing modalities, misalignment, duplication, weak supervision, and embedded bias can substantially affect performance, robustness, explainability, and fairness.

This workshop aims to provide a dedicated venue for discussing how data quality shapes multimodal recommendation pipelines, from data collection and preprocessing to modeling, evaluation, and deployment. We welcome submissions presenting original research, reproducibility studies, tools, datasets, as well as visionary or critical perspectives on the field.

DaQuaMRec welcomes submissions that fall in the following three categories:

Research Papers: should present original work that has not been previously published, is not under review, and will not be submitted elsewhere during the review process.

• Long papers (up to 8 pages, excluding references) should make a clear and novel contribution and be positioned with respect to the state of the art.

• Short papers (up to 4 pages, excluding references) may present early-stage research, promising ideas, negative results, or thought-provoking perspectives that can stimulate discussion and future work.
  
  

Reproducibility and Resource Papers: this category includes submissions focused on tools, datasets, benchmarks, and reproducibility studies, including newly developed resources, significant updates to existing tools, or systematic evaluations of published work.

• Long papers (up to 8 pages, excluding references) should present substantial contributions, such as comprehensive tools, large-scale datasets, or in-depth reproducibility analyses.

• Short papers (up to 4 pages, excluding references) may describe smaller-scale resources, focused tool descriptions, or preliminary reproducibility efforts of interest to the community.
  
  

Position Papers: are intended for short, critical, or visionary contributions that highlight future directions, emerging challenges, or reflective perspectives on the field. Position papers may be up to 2 pages, excluding references, and should aim to spark discussion and inspire future research, even in the absence of experimental results.

All submissions must be written in English, formatted as PDF files, and follow the CEUR-WS single-column conference format, available as a compressed archive and an Overleaf template.

Submissions will undergo a double-blind peer review process. Review criteria include relevance to the workshop, originality, significance of the contribution, technical soundness, clarity of presentation, quality of references, and reproducibility.
Authors are encouraged to share code and supplementary material via an anonymous repository, such as https://anonymous.4open.science/, to support reproducibility.

Submissions that are not properly anonymized, fail to follow the required formatting, or disregard these guidelines may be rejected without review.

Accepted long and short papers will be published in the CEUR Workshop Proceedings and presented in the main workshop program. Position papers will also be included in the proceedings; a selection of these may be invited for oral presentations.

Materials and workshop discussions will be available on this website.

Please note that at least one author of each accepted paper must register for and attend the workshop in order to present the work. We expect authors, reviewers, and organizers to adhere to the ACM Conflict of Interest Policy and the ACM Code of Ethics and Professional Conduct.