Hierarchical Semantic Retrieval with Cobweb

Anant Gupta, Karthik Singaravadivelan, Zekun Wang

Keywords: Information Retrieval, Cobweb, Interpretability, Language Embeddings, Transformers

TL;DR: We use Cobweb to turn sentence embeddings into a prototype tree for coarse-to-fine retrieval; it matches FAISS on strong encoders and stays robust & interpretable when FAISS fails (e.g., with GPT-2).

Abstract: Neural document retrieval often treats a corpus as a flat cloud of vectors scored at a single granularity, leaving corpus structure underused and explanations opaque. We use Cobweb--a hierarchy-aware framework--to organize sentence embeddings into a prototype tree and rank documents via coarse-to-fine traversal. Internal nodes act as concept prototypes, providing multi-granular relevance signals and transparent rationale through retrieval paths. We instantiate two inference approaches: a generalized best-first search and a lightweight path-sum ranker. We evaluate our approaches on MS MARCO and QQP with encoder (e.g., BERT/T5) and decoder (GPT-2) representations. Our results show that our retrieval approaches match the dot product search on strong encoder embeddings while remaining robust when kNN degrades: with GPT-2 vectors, dot product performance collapses whereas our approaches still retrieve relevant results. Overall, our experiments suggest that Cobweb provides competitive effectiveness, improved robustness to embedding quality, scalability, and interpretable retrieval via hierarchical prototypes.

Paper Track: Technical paper

Submission Number: 54