OwlerLite: Scope- and Freshness-Aware Web Retrieval for LLM Assistants

TL;DR

OwlerLite addresses the challenge of controllable, up-to-date web information for browser-based LLM assistants by introducing a scope- and freshness-aware RAG layer. It extends a LightRAG backend with a freshness-aware crawler and semantic change detection to selectively re-ingest updated content, and it fuses semantic relevance, scope priors, and recency in a single retrieval score $h(q,p) = \alpha\cdot sim_{vec}(q,p) + (1-\alpha)\cdot score_{graph}(q,p) + \beta\cdot log g(p; S_q) + \delta\cdot fresh(p)$. The paper formalizes the problem with metrics for scope fidelity $SF@k$, scope leakage $SL@k$, and stale-answer risk $R(q,t)$, and proposes a transparent explanation interface that shows scope and version provenance. An offline evaluation on the MS MARCO/TREC 2024 RAG corpus demonstrates gains in scope fidelity with modest reductions in traditional relevance, illustrating the potential of per-user, scope-aware, versioned indexing to improve trust and control in web-assisted retrieval.

Abstract

Browser-based language models often use retrieval-augmented generation (RAG) but typically rely on fixed, outdated indices that give users no control over which sources are consulted. This can lead to answers that mix trusted and untrusted content or draw on stale information. We present OwlerLite, a browser-based RAG system that makes user-defined scopes and data freshness central to retrieval. Users define reusable scopes-sets of web pages or sources-and select them when querying. A freshness-aware crawler monitors live pages, uses a semantic change detector to identify meaningful updates, and selectively re-indexes changed content. OwlerLite integrates text relevance, scope choice, and recency into a unified retrieval model. Implemented as a browser extension, it represents a step toward more controllable and trustworthy web assistants.