This section has its own page. Continue to the call for papers.
This section has its own page. Continue to the call for papers.
This section has its own page. Continue to the call for papers.
This section has its own page. Continue to the call for papers.
This section has its own page. Continue to the invited speakers.
This section has its own page. Continue to the invited speakers.
This section has its own page. Continue to the program.
This section has its own page. Continue to the program.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the invited speakers.
This section has its own page. Continue to the invited speakers.
This section has its own page. Continue to the invited speakers.
This section has its own page. Continue to the invited speakers.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
This section has its own page. Continue to the organizers and program committee.
First Workshop at NAACL 2027
Retrieval Beyond Documents
Capability Retrieval for Language Agents
San Francisco, CaliforniaWorkshop date to be announced
Overview
Retrieval Beyond Documents (RBD) studies how language agents find the resources they need to carry out a task. Alongside documents and passages that provide information, an agent may need a tool to query a database or run code, a reusable skill or workflow, memory from an earlier attempt, a specialized model, or another agent with relevant expertise. We use capabilities as a collective term for these resources.
Capability retrieval means finding and selecting suitable resources for an agent’s goal and current state. Relevance depends on what a resource can do, whether it is usable in the current context, and how it works with other resources. Interfaces, dependencies, permissions, cost, reliability, and execution history can all affect the choice.
RBD brings together researchers in natural language processing, information retrieval, and agent systems. The workshop spans representation and indexing, retrieval and ranking, selection and composition, and evaluation through use. We study how these choices adapt as a task progresses, how resources can be combined, and where failures arise between finding a resource and using it successfully.
Join the program committee: Help review submissions for RBD 2027. Program Committee Signup.
Contact
For questions about submissions, scope, or participation, please contact hitesh.laxmichand.patel@oracle.com.