Search papers, labs, and topics across Lattice.
To overcome the noise and inefficiency of standard attribute extraction over unstructured e-commerce catalogs, this work implements a two-stage pipeline that discovers purchase-discriminative attribute schemas before extracting values via a compact model. By pairing category-level schema distillation with Hyper-Parallel Decoding on a fine-tuned Qwen3-4B model, the framework addresses the high latency and cost barriers that typically prevent production-scale extraction. The resulting system matches the 85% extraction accuracy of its foundational teacher LLM while slashing inference costs by 92%.
Hyper-parallel decoding on a distilled 4B LLM matches frontier model information extraction accuracy while operating at an 8% compute footprint.
Customers rely on specific product attributes to compare products and make purchasing decisions, but e-commerce catalogs are messy and unstructured, making it difficult to identify which attributes matter most and extract them at scale. Standard Attribute Value Extraction (AVE) systems treat all attributes equally, producing large, inconsistent attribute sets that do not reflect the factors consumers use to differentiate products. We introduce a two-stage LLM pipeline that first discovers a compact, ranked schema of purchase-discriminative attributes for each product category, then extracts their values from catalog text using a fine-tuned compact LLM (Qwen3-4B) with Hyper-Parallel Decoding (HPD). This pipeline achieves 85% extraction accuracy, on par with the foundational LLM it was distilled from, while reducing inference costs by 92% over foundational LLMs, enabling production-scale use for product discovery and catalog enrichment. The resulting category-level structured representations effectively constitute automatically constructed product knowledge bases, providing consistent, comparable attributes across varied product categories that can ground downstream knowledge-intensive applications.
