Overview

Intent classification studies how to infer what a person wants from text, speech, dialogue, or behavioral signals, often as part of task-oriented systems that also identify entities through slot filling. Common approaches include fine-tuned Transformers, embedding-based classifiers, synthetic-data augmentation, and zero-shot or few-shot large language models; benchmarks such as CLINC150, Banking77, and MASSIVE test performance across domains and languages. Research also addresses out-of-scope queries, combinations of intents, noisy or multilingual input, and changes in intent across conversation turns. A central challenge is making systems reliable beyond familiar benchmarks: broad language understanding must be balanced with calibration, long-tail coverage, robustness to wording and distribution shifts, and accurate tracking of users’ current goals.

Latest papers

CardsList

65 papers4 papers in the last four weeks, down 20% on the four weeks before. 0.0% of all new papers.

  1. IntPro: Context-Aware Intent Inference via Selective Retrieval from User History

    Feb 10, 2026Guanming Liu, Meng Wu, Peng Zhang +8Intent ClassificationIn-Context Retrieval

  2. Understanding Role Switching in Human-AI Collaboration through Multimodal Behavioral Signals

    Sep 25, 2025Avinash Ajit Nargund, Arthur Caetano, Kevin Yang +7Intent ClassificationHuman-AI Decision Making

  3. Measuring Intent Comprehension in LLMs

    Jun 19, 2025Nadav Kunievsky, James A. EvansPrompt SensitivityLLM Evaluation

  4. A Survey of Intent Classification and Slot-Filling Datasets for Task-Oriented Dialog

    Jul 26, 2022Stefan Larson, Kevin LeachIntent ClassificationConversational Agents