Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2611.08014

Probabilistic Relevance Decisions for Semantic Retrieval

Bahadır Tatar·Cevat Şener
View original source ↗Read source document ↗Export metadata ↓

Abstract

Dense similarity supplies reusable corpus representations, while probabilistic relevance decisions offer query-dependent document scoring. We compare dense-only BGE-M3 with TypeSafe Jev Noul and OpenAI Decisions Predicate scores on the official BEIR SciFact test benchmark. The study evaluates 300 official test queries using identical positive-enriched 100-document pools and full-corpus dense retrieval followed by reranking at candidate depths of 50 and 100. Unlike the controlled pools, neither dense prefix receives positive injection. Jev has the highest observed Recall@5 and nDCG@10 under the evaluated conditions. At depth 50, full-corpus nDCG@10 is 0.6413 for BGE-M3, 0.7545 with OpenAI and 0.7805 with Jev. Extending depth from 50 to 100 increases known-positive candidate recall but produces small, uncertain final-effectiveness changes. All four paired-bootstrap depth intervals include zero. Unique-condition API scoring cost approximately doubles, while compatible cached decisions reduce incremental expenditure. OpenAI returns many exact-zero scores, and secondary exact-tie randomization reveals ranking sensitivity without replacing deterministic primary results. Top-100 latency mixes cached and new execution. Historical hybrid latency is a saved-component sum under different execution environments and timer boundaries. These exploratory, unadjusted analyses support relevance decisions as a complement to dense retrieval within this protocol, without establishing general superiority, probability calibration or equivalence between depths.

Explore related subjects

cs.IR

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Bahadır Tatar, Cevat Şener. 2026-11-05. Probabilistic Relevance Decisions for Semantic Retrieval. https://arxiv.org/abs/2611.08014

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

MCGI: Manifold-Consistent Graph Indexing for Billion-Scale Disk-Resident Vector Search

Graph-based Approximate Nearest Neighbor (ANN) search often suffers from performance degradation in high-dimensional spaces due to the Euclidean-Geodesic mismatch, where greedy routing diverges from the underlying data manifold. To address this challenge, this paper presents Manifold-Consistent Graph Indexing (MCGI), a geometry-aware and disk-resident indexing method that leverages Local Intrinsic Dimensionality (LID) to allocate the search effort of a graph index according to the intrinsic geometry of data. Specifically, MCGI estimates the LID around each query from the distances that its own search has already computed and resizes the candidate list of the query once, so that queries in high-LID regions receive longer lists while queries in low-LID regions terminate earlier; the estimate itself costs no additional disk reads. Theoretical analysis shows that the success probability of any fixed search list decays exponentially with the LID of the query. At matched high recall, MCGI raises the query throughput of DiskANN on the same standard Vamana index by up to $1.43\times$ on three million-scale benchmarks and by up to $1.24\times$ on the billion-scale SIFT1B and Deep1B.

cs.IR↗
arXiv2026

Unified Heterogeneous Feature Interaction via Effective-Rank Scaling for Recommender Systems

Large-scale industrial ranking models must jointly capture heterogeneous feature interactions, long-term user behaviors, and multi-scenario dependencies under stringent online serving constraints. Existing architectures typically process sequential and non-sequential features with separate modules, limiting bidirectional information exchange. Directly applying homogeneous self-attention to all semantic tokens can also be inefficient when sequential and non-sequential signals exhibit different effective-ranks (eRanks). Meanwhile, shared representations can overlook scenario-specific patterns and induce negative transfer between scenarios. In this paper, we propose HiRank, a unified framework for multi-scenario recommendation that jointly models Heterogeneous feature Interactions via effective-Rank scaling. HiRank integrates gated sequential attention with heterogeneous feature interaction within a unified backbone, enabling bidirectional fusion between sequential and non-sequential representations. HiRank further uses scenario-aware aggregation to compress heterogeneous low-order tokens into a compact representation. Distinct learned projections transform the full compact representation into scenario-aware composite tokens with complementary global feature views and more uniform information density, enabling more effective high-order self-attention. Extensive experiments on industrial-scale datasets demonstrate favorable performance scaling with parameter count and consistent improvements over competitive baselines. Online A/B tests further show a 2.27% increase in orders per user.

cs.IR↗
arXiv2026

SkinMind: Equipping Agentic RAG with Human-Inspired Graph Probing for Reliable Dermatology AI

Reliable dermatology AI requires access to verifiable evidence distributed across heterogeneous textual and visual sources. However, conventional retrieval-augmented generation (RAG) typically relies on one-shot retrieval from flat corpora, limiting its ability to adapt retrieval granularity, explore multi-hop relations, and integrate multimodal clinical evidence. We introduce SkinMind, a multimodal agentic RAG framework that equips policy models with human-inspired graph probing for autonomous evidence seeking. SkinMind organizes retrieval as an iterative interaction between an orchestrator and a graph environment. Through eight hierarchical graph-probing tools, the orchestrator progressively identifies relevant entities, inspects node-level information, traverses relations and paths, and retrieves source-text and visual evidence. This progressive-disclosure design allows the model to adaptively determine the retrieval modality, exploration direction, and evidence granularity while constructing a query-specific evidence subgraph. To support autonomous graph probing, we construct SkinMind-KG from a 62M-token corpus of dermatology guidelines and textbooks and 423K images, guided by a dermatologist-informed schema and a conflict-resolution mechanism. The resulting graph contains over 1.0M nodes and 2.0M edges, unifying textual, relational, and visual evidence for retrieval. Across six dermatology VQA benchmarks and two policy-model settings, SkinMind improves the average score over direct answering by 8.70 and 8.89 points with GPT-5.4-mini and Gemini-2.5-Flash, respectively, while consistently outperforming competing RAG methods. Together, SkinMind and SkinMind-KG provide an effective foundation for evidence-grounded multimodal reasoning in dermatology.

cs.IR↗