Known Intents, New Combinations: Clause-Factorized Decoding for Compositional Multi-Intent Detection

2026-03-30

AI Summary

This work addresses the limited generalization of existing models to unseen intent combinations—a common challenge in real-world applications. To tackle this, the authors propose ClauseCompose, a method leveraging clause-level factorized decoding that trains lightweight decoders exclusively on single-intent data yet effectively recognizes novel intent compositions. The study also introduces CoMIX-Shift, the first benchmark specifically designed for compositional generalization in multi-intent understanding, featuring controlled data construction and zero-shot triplet evaluation protocols. Experimental results demonstrate that ClauseCompose achieves a 95.7% exact match accuracy on unseen intent pairs, substantially outperforming strong baselines such as full-sentence multilabel classification and fine-tuned BERT models.

Abstract

Multi-intent detection papers usually ask whether a model can recover multiple intents from one utterance. We ask a harder and, for deployment, more useful question: can it recover new combinations of familiar intents? Existing benchmarks only weakly test this, because train and test often share the same broad co-occurrence patterns. We introduce CoMIX-Shift, a controlled benchmark built to stress compositional generalization in multi-intent detection through held-out intent pairs, discourse-pattern shift, longer and noisier wrappers, held-out clause templates, and zero-shot triples. We also present ClauseCompose, a lightweight decoder trained only on singleton intents, and compare it to whole-utterance baselines including a fine-tuned tiny BERT model. Across three random seeds, ClauseCompose reaches 95.7 exact match on unseen intent pairs, 93.9 on discourse-shifted pairs, 62.5 on longer/noisier pairs, 49.8 on held-out templates, and 91.1 on unseen triples. WholeMultiLabel reaches 81.4, 55.7, 18.8, 15.5, and 0.0; the BERT baseline reaches 91.5, 77.6, 48.9, 11.0, and 0.0. We also add a 240-example manually authored SNIPS-style compositional set with five held-out pairs; there, ClauseCompose reaches 97.5 exact match on unseen pairs and 86.7 under connector shift, compared with 41.3 and 10.4 for WholeMultiLabel. The results suggest that multi-intent detection needs more compositional evaluation, and that simple factorization goes surprisingly far once evaluation asks for it.

Problem

• compositional generalization
• multi-intent detection
• unseen intent combinations
• benchmarking
• zero-shot triples

Innovation

• compositional generalization
• multi-intent detection
• clause-factorized decoding
• zero-shot intent combination
• controlled benchmark