H+ Embedding is introduced, a unified multi-granularity retriever that predicts variable-length phrase partitions, preserves uncovered tokens as singletons, and applies importance-guided unit selection with weighted MaxSim interaction.
Abstract
Terminology-intensive retrieval, especially in medical settings, depends on preserving multi-word entities, abbreviations, numerical constraints, and compositional concepts. However, existing representations lie at two extremes: single-vector retrievers often over-compress local relevance signals, while token-level late interaction retains every tokenizer subword at substantial indexing, storage, and scoring cost. This mismatch raises a natural question: can context-dependent phrases provide a useful retrieval unit between global vectors and tokens? We introduce H+ Embedding, a unified multi-granularity retriever that predicts variable-length phrase partitions, preserves uncovered tokens as singletons, and applies importance-guided unit selection with weighted MaxSim interaction. Across 16 scientific, medical, and bilingual tasks, its phrase retrieval branch exceeds the global retrieval branch by 6.91 macro nDCG@10. It also nearly matches Token while using 13.7% fewer document vectors and outperforms content-independent grouping rules under moderate vector budgets. Context-dependent phrase interaction therefore provides an intermediate quality-cost point between global compression and token-level interaction for practical retrieval systems.
SALT, a model-agnostic extractive framework that organizes per-sentence keywords into a trie ordered by sentence frequency (SF), a lightweight, reusable proxy for document thematic structure, reduces the prefill computation and memory cost of long-context prompts while remaining composable with KV-cache methods that target decoding-time latency and memory.
Oteo Mamo, Hyunji Yi, Joydhriti Choudhury et al.· 0 citations
Lexical retrieval (BM25) captures exact keyword matches and weights terms by corpus-wide significance, but it is blind to the semantic vocabulary gap: when a relevant document phrases an answer differently from the query, BM25 never retrieves it, and no amount of downstream reranking or fusion can recover a document that was never in the candidate set. We present Cross-Encoder Query Expansion (CE-QE), which reads the per-token relevance attributions of a cross-encoder applied to top semantic search results, selects the terms the cross-encoder treats as decisive, and appends them to the BM25 query. Unlike classical pseudo-relevance feedback, which reuses BM25's own (possibly wrong) top results, CE-QE seeds expansion from the semantic retriever's results, avoiding self-reinforcing query drift. Unlike recent generative query expansion (HyDE, Query2doc), which prompts a large language model to hallucinate text from its parametric knowledge, every CE-QE expansion term is copied verbatim from a retrieved passage, so it cannot introduce vocabulary the corpus does not contain, and its only added cost is attribution extraction on a cross-encoder a hybrid pipeline already runs for reranking. On seven BEIR datasets, CE-QE improves lexical recall substantially where query and answer vocabulary diverge (e.g., NQ Recall@100 from 0.32 to 0.47), and its score-fusion variant (SESF) beats cross-encoder score fusion by 2.5% on Recall@100 and beats SPLADEv2 and ColBERTv2 by 5.3% and 4.6% on nDCG@10, while leaving the underlying BM25 index completely unmodified.
Adam Kahirov, Umesh Deshpande, S. Sundararaman· 0 citations
In highly specialized and terminology-dense scientific and technical texts, existing word sense disambiguation (WSD) models struggle to adequately model the contextual semantic dependencies of polysemous words, especially in engineering domains where the same term may carry different technical meanings across contexts. To address this issue, this paper proposes a robust WSD model that integrates a bidirectional long shortterm memory network (BiLSTM ) with an attention mechanism, specifically designed for Chinese patent texts. First, a two-layer BiLSTM is used for bidirectional context modeling to capture long-range dependencies. Then, a multi-head attention mechanism uses dynamic weighting to highlight key semantic components, generating highly discriminative context vectors. Finally, a paraphrase alignment mechanism employs bilinear matching to align the context vectors with candidate paraphrase embeddings, thereby reducing semantic confusion. Experiments show that the model achieves a Top-1 accuracy of 88.6 % on high-frequency words, with an average paraphrase alignment similarity of 0.920. In perturbation tests, the average robustness index is 0.159, representing reductions of 62.7%, 53.9%, and 39.5% compared to Word2Vec+CNN, BiLSTM, and BERT, respectively. The method presented in this paper helps to enhance the accuracy and stability of word meaning recognition in technical texts, providing reliable support for knowledge mining and intelligent text processing in technical domains. Its terminology alignment is also useful for engineering corpora where antenna, wavepropagation and materials terms require context-sensitive interpretation.
L. X. Gao, T. Dong, M. H. Yang· Advanced Electromagnetics· 0 citations
KoVRE: Korean Visual Document Retrieval Embedding is introduced, a single-vector retriever for Korean visual documents, alongside a comprehensive training recipe, demonstrating that targeted bilingual supervision and carefully designed training strategies can produce a highly effective Korean VDR model across diverse document domains, without requiring a scaled-up backbone or multi-vector representations.
This work defines model-specific Token and Phrase Representativeness Scores (TRS/PRS) to discover tokens and key phrases that appear in a document but are poorly expressed by its embedding, which improve retrieval on a constructed hard-positive benchmark and can also improve standard in-domain retrieval benchmarks.