🤖💬 What if we moved AI from question answering to task support? AI chatbots and search engines answer questions - but what if they understood the task you're trying to accomplish, not just your query? 🤔
Our CHIIR 2026 paper, "Rules, Resources, and Restrictions: A Taxonomy of Task-Based Information Request Intents" by Melanie A. Kilian and David Elsweiler, takes a task-based perspective on query intents to better align search and LLM-based systems with complex, real-world activities.
🕳️ The gap:
Existing query intent taxonomies mostly capture isolated information needs derived from log data, while the broader context of the user's underlying task remains invisible.
As LLMs and AI assistants now expand our expectations from simple query answering towards comprehensive task support - for example in purchasing decisions or travel planning - understanding the user's task context becomes crucial.
🛠️ What we did:
We interviewed airport information clerks with decades of experience to understand how real-world tasks shape information needs. Based on this, we developed a task-based taxonomy of request intents.
🧩 Our query intent taxonomy:
- aims to bridge the gap between traditional query-focused classifications and the emerging demands of (AI-driven) task-oriented search.
- captures diverse request intents in a multi-faceted task context, from rules for action to personal recommendations, status updates, and how-to instructions.
- provides a high-level, task-oriented perspective on request intents that supports generalisation beyond the airport setting
💡 Key insight: Task context matters.
The same question can imply different needs depending on the task context.
For example, we found that "Where is...?" typically calls for a simple location during planning tasks - but can also mean precise wayfinding directions when someone is about to act.
👉📄 Learn how information clerks handle this and explore our request intent taxonomy here: https://arxiv.org/abs/2601.12985 (externer Link, öffnet neues Fenster) (pre-print)
We are excited to share and discuss this work with the CHIIR community in Seattle. See you there! 👋
#QueryIntents #UserGoals
#TaskBasedSearch #TaskBasedIR
#DigitalAssistants
#InformationSeeking
#InformationRetrieval #IR
#EmpiricalStudy #QualitativeStudy
#GroundedTheory #InterviewStudy
#SearchResearch #AIResearch #PhDResearch
#ACM #SIGIR #ConferenceOnHumanInformationInteractionAndRetrieval #CHIIR #CHIIR2026
#ResearchSuccess #ResearchPaperAccepted
#InformationScienceRegensburg #StayInformed
