From High Recall to High Utility: Dataset-Adaptive Post-Processing of LLM-Generated Customer Intents

Key points

• Large language models can extract useful signals from heterogeneous enterprise data, but high-recall extraction often produces outputs that are duplicated, uneven in granularity, semantically overlapping, or too numerous for downstream systems and human reviewers to use effectively.1
• Source-specific preprocessing first isolates evidence from multimodal plans, sparse operational records, and structured opportunity data.1
• Finally, constrained language-model aggregation produces one concise intent per cluster without introducing unsupported concepts, and the resulting unit retains provenance, clustering, embedding, and generation metadata.1
• We also describe two downstream applications: Machine-Generated Intents, which infer likely objectives for customers lacking direct evidence from peer customers with similar profiles, and intent-guided semantic retrieval and mapping, which uses the stable intent as a query against a downstream decision space, illustrated here by mapping customer intents to business outcomes.1

Sources (1)

• [1]From High Recall to High Utility: Dataset-Adaptive Post-Processing of LLM-Generated Customer Intents
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 07:41 PM

  We present a dataset-adaptive post-processing architecture developed for Customer Intent Extraction (CIE), where unstructured customer language is transformed into stable, traceable intent units.

  Large language models can extract useful signals from heterogeneous enterprise data, but high-recall extraction often produces outputs that are duplicated, uneven in granularity, semantically overlapping, or too numerous for downstream systems and human reviewers to use effectively.

Extractive summary: sentences quoted from the sources.