Hybrid retrieval—fusing lexical BM25 with dense neural retrievers—is the default recommendation in modern search stacks, usually justified by aggregate benchmark gains. This paper challenges that consensus by diagnosing what it calls the "fusion tax": the frequency and severity with which fusion hurts individual queries relative to the better of its two components or to a deployable dense-only baseline. Through exhaustive experiments across seven BEIR corpora, five encoders spanning 22M–596M parameters, and multiple fusion functions, the authors show the tax is material, pervasive, and undiminished by encoder scale, and they derive practical guidance for when to fuse and which method to choose.

This survey tackles long-document visual question answering, where answers require locating and integrating evidence scattered across many pages and modalities including charts, tables, and fine-grained layout. Rather than grouping papers by model architecture, the authors decompose the task into four functional components: document parsing, evidence organization and retrieval, generator organization and learning, and evaluation. The core argument is that task bottlenecks—visual fidelity, evidence dispersion, and budget limits—should drive taxonomy design, not just technical novelty. This reframing matters because the field has grown explosively since 2024 and prior surveys have organized literature by approach rather than by the functional requirements that long documents impose.

The paper studies how much embedding work is wasted when documents in a RAG knowledge base are revised, comparing fixed-size, recursive paragraph-first, and content-defined chunking on 120 adjacent Wikipedia revision pairs in Chinese and English. It finds that fixed-size windows force the most re-embedding (57–68\% of the embedding-input workload), while content-aware methods cut this by double-digit percentage points. Surprisingly, the extra recomputation buys almost no retrieval recall advantage under a fixed 2048-token budget. The authors argue that update cost, measured under byte-identical reuse, should be reported alongside static retrieval metrics when choosing a chunker for an evolving knowledge base.

PAIR tackles a fundamental mismatch in professional network retrieval: relevance is complementary rather than symmetric. LinkedIn proposes a triple-tower retrieval framework that learns from exposure-conditioned preference signals via a multi-negative contrastive objective, combines static profile embeddings with lightweight dynamic history updates in embedding space, and supports multiple serving budgets through Matryoshka training. The work matters because demonstrated retrieval gains translated into real business impact at billion-user scale.

As Mathlib grows past 400,000 theorems, effective semantic search is critical for both human formalization and automated proving, yet real-world queries are short and ambiguous while existing engines train on lengthy, context-rich informalizations. Lean Finder V2 closes this gap with an adversarial generator-retriever loop that synthesizes over 600,000 faithful, challenging short queries, then uses them for both retriever fine-tuning and GEPA-optimized input rewriting. The result is a new state of the art on human-curated and real-world benchmarks, plus measurable gains in downstream theorem-proving tasks.

Automated survey generation has relied on flat document retrieval, which misses the evolutionary structure of research fields. SurveyGraph tackles this by constructing a Methods-Entity Knowledge Graph (MEKG) that encodes typed method-level relations—such as extends, improves, and applies—across papers. The graph then serves as a unified backbone for core-paper discovery, graph-aware retrieval, hybrid outline generation, and evolutionary narrative synthesis, aiming to make survey generation more auditable and structure-aware.

Current agentic systems often hallucinate or drift during multi-turn reasoning because they rely on unstructured token-level generation. This paper proposes SGUR, a training-free framework that grounds multi-step inference in structured hypothesis spaces using partial graph alignment and active information gain. Evaluated on a sequential medical-diagnosis benchmark derived from PrimeKG, the results suggest that tighter structural discipline can partially offset the need for ever-larger language models.

Graph collaborative filtering models collapse in strict cold-start settings because zero-degree items sever the message-passing chain. DualGCN tackles this with an asymmetric two-branch distillation framework: a teacher network anchors item representations with frozen semantic priors smoothed over the graph topology, and a lightweight student learns to project unseen semantic features into that collaborative space via cosine distillation and residual mapping. If the claims hold up, it offers a much leaner alternative to heavy multi-modal fusion modules for industrial retrieval.

When multiple LLM retrieval actions are generated jointly, downstream systems typically return a single reward for the merged result set, making it impossible to tell which action helped or hurt. CASI closes this information gap through snapshot-grounded set intervention: it caches per-action retrievals, evaluates all alternative subsets under a frozen training snapshot, and aggregates the resulting marginals into Shapley-based credit. This matters because broadcasting the same scalar reward ignores redundancy and competition among actions, which misleads policy optimization in capacity-limited regimes such as sponsored search.

Existing retrieval-augmented MT applies retrieval uniformly, yet LLM parametric knowledge suffices for many inputs and external evidence is only conditionally useful. This paper proposes SRAT, an agentic RL framework that treats reasoning, retrieval, and translation as learnable actions optimized via GRPO for final translation quality. It learns when to retrieve, what to query, and how to integrate evidence, supported by two retriever-alignment strategies and evaluated across idiom, knowledge-intensive, and document-level settings. The work is important because it challenges the default assumption that retrieval is always beneficial and shows that adaptive, quality-aligned evidence acquisition consistently outperforms fixed pipelines.

Advertising delivery still relies heavily on manual coordination across budgeting, bidding, and diagnosis, limiting scale and performance. AdCoEvo automates this pipeline by assigning Planner, Optimizer, and Inspector roles to an 8B-parameter language model and updating their LoRA adapters in rotation using settled business feedback, so the agents mutually adapt during continuous simulated delivery. The authors report gains of roughly 14% in advertiser value and 7% lower cost per acquisition over rule-based baselines in the AuctionNet simulator.

Personalization encoders compress lengthy user interaction histories into a fixed preference state for downstream ranking or generation, but this compression can suppress evidence that remains available in cached per-timestep representations. This paper proposes REPAIR, a lightweight module that identifies and aggregates corrective signals from a frozen encoder's own cache across long, short, and episodic temporal views, then adds the correction back to the compressed state before the task head. By keeping the host encoder and task head frozen, REPAIR offers a practical way to improve deployed personalization systems without retraining them.

Generative recommenders represent items as short discrete Semantic IDs (SIDs) that are decoded autoregressively, yet most prior work derives these codes by quantising isolated content embeddings. This paper argues that SID construction is fundamentally a recursive clustering problem over a collaborative item graph, and proposes GrIS, a framework that separates graph construction from hierarchical partition to balance semantic and behavioural signals. By exposing these two axes, the authors show that content-only tokenisers are merely empty-graph instances of a broader design space, and they demonstrate new instantiations that improve SID quality for downstream recommendation.

RPTune addresses product search for small merchants whose catalogs fit inside long-context LLMs. Rather than relying on retrieval pipelines, the authors show that feeding the full catalog works only when items are learned to be pruned, ranked, and positioned with downstream LLM feedback, after which the LLM itself is post-trained on these curated contexts using purely synthetic supervision. The result is a fully automated pipeline that achieves large accuracy gains and latency reductions across 7 merchants and 9 model families.

A long-term assistant must recall memories whose cues are associatively related to, yet not semantically similar to, the current user turn. Madeleine reframes this involuntary recall as a learnable relevance—the pointwise mutual information of memories under simulated life trajectories—and distills it into a query-side residual adapter, $s_\theta(q,m)=\cos(E(q),E(m))+\cos(E_\theta(q),E(m))$. The system is trained offline on synthetic (trigger, cue) pairs written by an LLM life simulator, and at runtime it plugs into existing vector memories with zero LLM calls, cutting answer context by roughly $20\times$ while setting a new state of the art on LoCoMo-Plus.

JoinGR attacks a real bottleneck in Text-to-SQL pipelines: dense retrievers rank tables independently and miss ones only identifiable through multi-hop join relationships. The method builds a column-level join graph with typed intra-table and foreign-key edges, selects semantic anchor tables via dense similarity, then propagates query-conditioned relevance deposits along edges using a lightweight learned scorer. Final table scores aggregate these deposits, making schema structure a first-class retrieval signal rather than an afterthought, which matters most for enterprise settings with cryptic table names and implicit relational dependencies.

This paper addresses interactive text-to-image retrieval (I-TIR) where diffusion models generate visual proxies from dialogue, but static additive fusion of text and generated images introduces harmful noise. The core idea is ADaFuSE, a lightweight plug-in module combining adaptive gating (to dynamically weight modalities per instance) with a semantic-aware mixture-of-experts branch (to capture fine-grained cross-modal cues). The work matters because it challenges the assumption that diffusion-augmented retrieval always benefits from generated images, showing that up to 55.62% of queries suffer degradation under static fusion.

This paper identifies "semantic shift"—the intrinsic evolution of meaning within a text—as the root cause of embedding pathologies like anisotropy and length-induced collapse. The authors argue that pooling-based aggregation forces "semantic smoothing," where diverse sentences compromise into a diluted representation. They formalize semantic shift as the product of local evolution and global dispersion ($\mathrm{Shift}(k) = \mathrm{Local}(k) \cdot \mathrm{Disp}(k)$), showing through controlled concatenation experiments that it predicts embedding concentration and retrieval degradation better than text length alone. The work reframes geometric pathologies not as inherent model defects but as consequences of content structure interacting with pooling mechanics.

Code retrieval currently relies on dense embeddings, but this paper proposes SPLADE-Code, the first large-scale learned sparse retrieval (LSR) family for code search (600M–8B parameters). The authors address unique challenges including subword fragmentation, semantic gaps between natural language and code, and latency issues from long code documents. Their lightweight single-stage training achieves 75.4 nDCG@10 on MTEB Code under 1B parameters (state-of-the-art for that size) and 79.0 with 8B parameters, while enabling sub-millisecond retrieval via inverted indices.

This paper addresses cross-lingual knowledge graph fusion, where heterogeneous KGs in different languages must be unified without expensive manually-curated seed alignments. The core idea is to use Large Language Models as a universal semantic bridge by linearizing graph triplets into natural language sequences and sequentially agglomerating multiple graphs. This matters because it promises zero-shot alignment capability for low-resource languages where traditional embedding-based methods fail due to lack of training data.