From Explicit Elements to Implicit Intent: A Predefined Library for Auditable Behavioral Inference

In three linesSemantiClean is a modular framework extracting structured semantic signals from e-commerce session data to infer purchase intent, customer segmentation, and product affinity. Built on the OSPI dataset, it organizes 24 behavioral elements across 4 layers and integrates an LLM-driven inference engine with deterministic reproducibility (sigma=0).

Summary generated by Claude — human-verified