Industry Day will take place on the first day of the conference, Monday, February 15, 2027. It will bring together researchers, engineers, practitioners, and technology leaders to share technically substantive work at the intersection of research and real-world deployment.

The WSDM Industry Chairs invite proposals for technical talks on how web search, recommendation, data mining, and artificial intelligence are designed, built, deployed, evaluated, and operated in industrial settings. We welcome novel systems, applied research, large-scale experiments, deployment case studies, and candid accounts of technical challenges, tradeoffs, failures, and lessons learned. Submissions from industry teams, academia-industry collaborations, and other organizations with genuine deployment experience are encouraged.

WSDM 2027 Industry Day invites talk proposals under two complementary themes. Together, they highlight both the rapid industrialization of foundation models and agents and the broader technological, societal, and market forces shaping the future of web search, data mining, and digital platforms.

This theme focuses on how foundation models and agentic systems are adapted, grounded, evaluated, deployed, operated, and governed at industrial scale

This theme welcomes forward-looking industrial work that may use AI but need not be AI-centric. Its primary contribution may concern systems, data, users, markets, governance, or digital ecosystems rather than model capability itself.
Proposals may focus on either theme or bridge the two. Both themes will receive equal consideration, and strong proposals on other timely industrial challenges within the broader scope of WSDM are also welcome.

Across both themes, we especially welcome experience reports, case studies, system innovations, rigorous evaluations, lessons from deployment—including unsuccessful approaches—and perspectives that connect technical choices to business, user, and societal outcomes.

• Agent architectures and workflows. Planning and reasoning; tool and API use; web and GUI agents; memory; orchestration; state management; single- and multi-agent coordination; human approval; and reliable long-horizon execution
• Grounding, retrieval, personalization, and knowledge integration. Retrieval-augmented generation; search and ranking; knowledge graphs; structured and unstructured data integration; context management; long-term and episodic memory; freshness; provenance; citation; and domain-specific personalization.
• Model adaptation and optimization.Data curation; domain-specific pre-training and post-training; instruction tuning and preference optimization; alignment; distillation; multilingual and multimodal adaptation; inference efficiency; latency; and cost
• Evaluation, observability, and continuous improvement. Realistic task suites; task completion and utility; factuality and robustness; trajectory and human evaluation; simulation; A/B testing; causal measurement; tracing and diagnosis; feedback loops; and continual learning.
• Safe and responsible deployment. Hallucination and uncertainty; prompt injection, jailbreaks, and unsafe tool use; permissions and access control; data leakage; privacy and fairness; transparency; red teaming; auditability; governance; and meaningful human oversight.
• Industrial deployment and measurable impact. Real-world applications in search, recommendation, advertising, commerce, content intelligence, and knowledge systems; responsible rollout; failure analyses; LLMOps and AgentOps; build-versus-buy decisions; organizational adoption; and measurable user, business, or societal value.

• New paradigms for search, discovery, recommendation, and information access. Emerging ranking and matching methods; vertical, cross-lingual, and multimodal search; new interaction models across devices and contexts; long-term value; diversity and discovery; and meaningful user control.
• Web mining, data, and knowledge infrastructure. Crawling and indexing; extraction and content analysis; knowledge bases and large-scale graphs; traffic and log analysis; real-time and streaming data; data quality and provenance; benchmarking; and web measurement.
• Digital marketplaces and platform economics. Ranking and matching; auctions and advertising; pricing; incentives; e-commerce and creator ecosystems; competition; market design; and measures of platform and market health.
• Trust, safety, and information integrity. Fraud and abuse prevention; spam and coordinated behavior; misinformation; content quality and provenance; moderation; reputation and credibility; ecosystem integrity; and protection of vulnerable users.
• Privacy-aware data, personalization, and measurement. Privacy-preserving personalization; identity and consent; secure data collaboration; attribution; online experimentation; causal inference; counterfactual evaluation; and measurement under data constraints.
• Resilient, efficient, and sustainable web-scale systems. Distributed and real-time data and serving infrastructure; reliability and observability; edge and mobile systems; low-latency operation; cost and energy efficiency; responsible growth; and globally inclusive infrastructure.
• Open, interoperable, and decentralized ecosystems. Open standards; data portability; responsible data sharing; federated services; open-source infrastructure; interoperability; transparency; accountability; and new technical and regulatory governance models.
• Human-centered and emerging digital experiences. Accessibility and user well-being; long-term behavioral and societal impact; social, mobile, geospatial, ubiquitous, and connected-device systems; digital twins; immersive environments; and new applications in climate, education, healthcare, and science.

Proposals must be submitted as a PDF of at most two pages, including references, and must follow the ACM proceedings format. Please use the "sigconf" proceedings template in two-column format. Formatting instructions and templates are available on the ACM proceedings template page.
Submissions are not anonymous and must identify the authors and proposed speaker or speakers. Proposals will be reviewed using a single-blind process. The proposal should be self-contained, written in English, and contain no confidential or proprietary information.

• Title.
• Primary theme: Theme 1, Theme 2, or cross-theme.
• Abstract describing the problem, industrial context, technical contribution, and principal results or lessons.
• Outline of the presentation, including the key takeaways for the WSDM audience.
• A brief company or organization portrait (approximately 60 words).
• A short biography of the main presenter (approximately 100 words).

Proposals must be submitted electronically through the WSDM 2027 Industry Day submission portal. The portal URL will be published on this page when submissions open.

The Industry Day program committee will consider the following factors when selecting talks:

• Relevance. Fit with WSDM and clear value to the Industry Day audience
• Technical substance. Depth, novelty, and specificity of the methods, systems, evaluation, or operational lessons.
• Evidence. Credible results, experiments, deployment experience, or well-supported lessons, including informative negative results.
• Generalizability. Depth, novelty, and specificity of the methods, systems, evaluation, or operational lessons.
• Clarity. A focused, well-structured proposal that can become an engaging presentation.

Final selections may also take overall program balance into account, including representation across the two themes and the diversity of topics, organizations, application domains, and perspectives represented.

The presentation format will be determined after the submissions are reviewed and with the interests of the wider WSDM community in mind. The program is expected to include a mix of short and long presentations and may also include panels. Presentation length and format will be communicated to accepted speakers.

WSDM 2027 is an in-person conference. At least one author of each accepted proposal must register for the conference and present the work in person at Industry Day on February 15, 2027. Remote presentations will not be accommodated.

Authors of accepted proposals will be invited to submit a camera-ready abstract for inclusion in the conference proceedings. All accepted material must comply with applicable ACM publishing and open-access requirements. Detailed camera-ready and registration instructions will be provided with the acceptance notification.

Authors and coauthors are subject to the ACM Publications Policies, including the ACM policy on research involving human participants and subjects. Authors and coauthors of published abstracts should obtain an ORCID iD and connect it to their ACM profile.

For questions about Industry Day, please contact:

• Jiang Bian, Microsoft Research Asia , jiang.bian.prc@gmail.com
• Omar Alonso, Amazon, USA , omralon@amazon.com