Paper Daily · AI Search (English)

EmbeddingGemma 2: Unified and Efficient Multimodal Representations

Thu, 29 Oct 2026 08:00:00 +0800

To search text, images, video and audio with one embedding model, the usual route is to transcribe audio and caption images first, then hand everything to a text model. That costs extra compute and throws away tone of voice and visual detail. Google's EmbeddingGemma 2 is a 744M-parameter model that maps these inputs directly into one vector space, and each modality encoder can be loaded only when…

AI Search · Score 4.65 · Deep dive · 25 min · Henrique Schechter Vera, Sahil Dua, Jovan Andonov, Sara Smoot, Sebastian Russo, Ivan Llanos, Max Ma… · Source

A Dense Subset Index for Collective Query Coverage

Thu, 29 Oct 2026 08:00:00 +0800

In classic retrieval, documents compete for the top ranks. In multi-hop QA or text-to-SQL, a single document is often incomplete and the answer must be assembled from several. In dense retrieval terms, the task is to find a small set of documents whose token vectors together cover every token vector of the query. Coverage here means every token vector of the query finds a close token vector somew…

AI Search · Score 4.40 · Kartik Nair, Pritish Chakraborty, Atharva Abhijit Tambat, Indradyumna Roy, Soumen Chakrabarti, Anir… · Source

GRAPE: Let GRPO Supervise Query Rewriting by Ranking for Retrieval

Thu, 29 Oct 2026 08:00:00 +0800

Retrievers like CLIP degrade noticeably on Chinese queries, long-form queries, or combined image-and-text queries, yet retraining the retriever and re-embedding the whole corpus is too expensive. This paper leaves the retriever untouched and trains only an LLM that rewrites the query, using GRPO with retrieval rank as the reward for each rewrite. The authors find that if the reward only looks at…

AI Search · Score 4.40 · Deep dive · 28 min · Zhaohua Zhang, Jianhuan Zhuo, Muxi Chen, Chenchen Zhao, Wenyu Jiang, Mingyang Chen, Yu Tang, Qiuyon… · Source

LTRR: Learning To Rank Retrievers for LLMs

Thu, 29 Oct 2026 08:00:00 +0800

RAG systems usually wire in one fixed retriever, yet different questions suit different retrievers, and some are better answered with no retrieval at all. This paper treats choosing a retriever as a ranking problem. For each training question it records how much each retriever improves answer quality over no retrieval, then trains a ranking model to order the retrievers, with no retrieval as one…

AI Search · Score 4.40 · Deep dive · 29 min · To Eun Kim, Fernando Díaz · Source

Natural Hard-Negative Mining from Positives for Instruction-Following Retrieval

Thu, 29 Oct 2026 08:00:00 +0800

Instruction-following retrieval asks one query to rank documents differently under different instructions: once an instruction adds a requirement, the original positive should drop if it fails that requirement. Training data for this is scarce, and existing methods keep the positive fixed and have an LLM invent a negative passage that violates the instruction. This paper does the reverse. It pick…

AI Search · Score 4.40 · Deep dive · 23 min · Chaolang Li, Jingyu Li, Zhaocheng Du, Xuyang Xie, Gang Wang, Pengwen Dai · Source

Tessera: Structure-Completed Evidence Units for Web Retrieval-Augmented Generation

Thu, 29 Oct 2026 08:00:00 +0800

Ask Wikipedia who produced Willy Wonka and the Chocolate Factory, and the answer sits in the producer row of the page's infobox. Text extraction drops the two names in that cell, the reader model sees only the director, and it answers wrong. Tracing answer evidence from page source to reader input over all of English Wikipedia, this paper finds such loss is common: for 7.6% of questions the evide…

AI Search · Score 4.30 · Deep dive · 21 min · Peichun Hua, He Sun, Zirui Zhuang, Qi Qi, Yiming Qiu, Shinan Liu, Yunming Xiao · Source

QESChunker: A Single Objective Unifies Overlapping and Non-Overlapping Chunking for RAG

Thu, 29 Oct 2026 08:00:00 +0800

Most RAG chunking is rule of thumb: fixed length, by paragraph, or by semantic similarity, and none aims directly at whether a question's evidence can be retrieved complete. This paper uses existing questions and their evidence to decide where each document is cut, aiming to put as many evidence passages as possible into chunks that the question can actually retrieve. Without overlap the objectiv…

AI Search · Score 4.30 · Yifan Zhao, Qianyang Zhao, Xinyu Guan, kai wei, Yuming Deng · Source

TaskBridge: Technical Report for SIGIR 2026 eCommerce Challenge

Thu, 29 Oct 2026 08:00:00 +0800

Many e-commerce queries describe a task rather than a product. A user types make lasagna and actually needs a baking dish, ingredients and kitchenware, but product titles contain none of those words, so BM25 struggles to retrieve them. This winning solution to an e-commerce challenge first aligns the query with category and entity terms in the catalog, then has an LLM generate five complementary…

AI Search · Score 4.05 · Zhiheng Fu, Yupeng Hu, Zixu Li, Zhiwei Chen, Qinlei Huang, Shaokun Wang, Han Liu, Fan Liu, Weili Gu… · Source

RouteDRAG: Learning to Route and Control Agentic Execution for Multimodal Document RAG

Thu, 29 Oct 2026 08:00:00 +0800

In multimodal QA over long documents, agent-style repeated retrieval and reasoning is not always useful: some questions are answered right after one retrieval, and extra steps can turn a right answer wrong. The authors first show experimentally that one more tool call can improve, preserve or degrade the answer. RouteDRAG therefore learns two things: whether a question should go to the agent at a…

AI Search · Score 4.00 · Zekai Chen, Hao Zhu, Yu Liu, Xunkai Li, Zejian Kang, Rui Qian, Quanjiang Li, Rong-Hua Li · Source

When Connectivity Misleads: Guiding Agentic Search with Contrastive Evidence Paths

Thu, 29 Oct 2026 08:00:00 +0800

When a search agent looks for evidence on a document graph, candidates are often scored with Personalized PageRank. Well-connected documents are naturally easy to reach, so their scores are inflated even when their evidence is no more useful, and they crowd the agent's limited observation window. This paper scores each candidate by one strongest path only, the one from the initially retrieved doc…

AI Search · Score 4.00 · Junchen Li, Liang Xu, Rongzheng Wang, Qizhi Chen, Junhao Wen, Shuang Liang · Source

Retriever-Free Retrieval: Navigating Documents with Decision Models

Wed, 28 Oct 2026 08:00:00 +0800

Most RAG systems still retrieve with BM25 or vector similarity, and evidence the retriever misses cannot be recovered by any generator. DecisionRAG drops the retriever and has a decision model browse like a person in a library, from topics to pages to passages, judging each level's candidates together in one context. Over 2.68 million Wikipedia passages its Hit@1 is 17.4 points above Qwen3-Embedd…

AI Search · Score 4.40 · Deep dive · 31 min · Chao Wan, Paul Jünger, Beixi Du, Bodong Liu, Navaneetha Krishnan, Kilian Q. Weinberger · Source

ACARE: Agent-Context-Aware Reranker for Iterative Search

Wed, 28 Oct 2026 08:00:00 +0800

Search agents often send sub-queries of just a few keywords, so a sub-query-only reranker promotes documents that match the words but do not help the real question. ACARe gives the reranker the original question and the agent's reasoning too. The key control: without changing Qwen3-Reranker-4B's weights, adding the question alone lifts a small agent's accuracy from 48.2% to 54.8%, and fine-tuning…

AI Search · Score 4.40 · Deep dive · 25 min · Sahel Sharifymoghaddam, Yauhen Babakhin, Reza Esfandiarpoor, Oliver Holworthy, Jimmy Lin, Bo Liu · Source

CA-HNM: Constraint-Aware Hard Negative Mining for Dense Retrieval with Structured Domain Models

Wed, 28 Oct 2026 08:00:00 +0800

Dense retrievers usually take hard negatives from top-ranked documents that are not labeled positive, but ranking high only shows a document resembles the query, not why it should be excluded. CA-HNM checks each candidate against a course knowledge graph, wrong concept, wrong scope or wrong difficulty level, and keeps only those with a clear conflict as negatives. It beats a rank-matched control…

AI Search · Score 4.40 · Deep dive · 31 min · Ngoc Luyen Le, Marie-Hélène Abel, Bertrand Laforge · Source

SkillQR: Agentic Skill Discovery for Reason-Intensive Query Rewriting

Wed, 28 Oct 2026 08:00:00 +0800

LLM query rewriting usually applies one instruction to every query, yet the right direction for a rewrite depends on the corpus. SkillQR clusters the corpus offline and has an LLM write a few rewrite instructions for the document types, such as listing the required data structures and algorithms; at query time the same model picks one before rewriting, with no training data. Its mined skills matc…

AI Search · Score 4.30 · Deep dive · 23 min · Kenan Jiang, Meixiu Long, Zihao Zhang, Ran Xu, Eugene Agichtein, Fei Liu · Source

Can System-1-Style Decision Models Replace Dedicated Rerankers in RAG? A Systems Evaluation

Wed, 28 Oct 2026 08:00:00 +0800

If a team wants to replace its RAG reranker with a large model, this paper says to tune the existing reranker first. Comparing the 27B decision model PPLX Decider with the 0.6B Jina v3.5 on identical candidates, default PPLX returns valid results for only 44.1% of requests; it looks better on completed requests but clearly loses once failures count. Splitting the input budget evenly and keeping e…

AI Search · Score 4.15 · Deep dive · 26 min · Qiaomu Li, Ying Xie · Source

Ranking Is Not Deciding: Benchmark Shortcuts and Operating-Point Collapse in Query-Time RAG Poisoning Defenses

Wed, 28 Oct 2026 08:00:00 +0800

RAG poisoning defenses often report near-perfect detection, but this paper finds much of it comes from how poison sets are built: PoisonedRAG's code prepends the target question to every poison passage. Deleting only that prefix raises the published RAGuard filter's miss rate from 0.191 to 0.763, which means it mostly recognizes the prefix. The authors argue detection numbers mean little unless r…

AI Search · Score 4.10 · Pushkal Kumar, Arvin Sahaym · Source

Knowing When the Evidence Is Enough: Requirement-Directed Search with Horn-Closure Certificates

Wed, 28 Oct 2026 08:00:00 +0800

After each search, a search-augmented model must decide whether its evidence suffices: answer, search again, or abstain. Leaving this to holistic model judgement lets it answer when every entity is mentioned but one required fact is missing. Certify splits the question into fact requirements, stating which facts together yield the next, and searches for each; a passage satisfies a requirement onl…

AI Search · Score 4.00 · Xinyu Zhang, Weiwei Sun · Source

FormBench: A Benchmark for Fine-Grained Retrieval in Formulation Patents

Wed, 28 Oct 2026 08:00:00 +0800

In formulation patents, relevant and distracting passages often name the same chemicals and differ only in the role they play, which generic embedding models struggle with. FormBench tests this with over 55,000 queries, with distractors chosen through a materials taxonomy to share most chemicals with the answer but differ in function. The results are striking: OpenAI's commercial embedding scores…

AI Search · Score 3.85 · Vineeth Venugopal, Elsa Olivetti, Jonathan Welch · Source

Evolve on Graph: Self-Evolving Knowledge Graph Reasoning without Question Annotations

Wed, 28 Oct 2026 08:00:00 +0800

Knowledge graphs in finance or industry rarely have QA annotations, so there is no data to improve LLM multi-hop graph reasoning with RL. EoG has one model alternate between proposer and solver: it samples reasoning structures on the graph and executes logic queries for exact answers, and the proposer writes multi-hop questions from them. The solver answers each question several times, and the pr…

AI Search · Score 3.75 · Zongyuan Sun, Shuai Zhao, Ke Shu, Ziyue Zhu, Shangyang Wu, Yuanlin Lin, Tao Wang, Shiyue Li, Yingju… · Source

CalibraRL: Confidence-Calibrated Reinforcement Learning for Mitigating Hallucinations in RAG

Wed, 28 Oct 2026 08:00:00 +0800

When RL teaches a model to abstain when unsure, typical rewards penalize every wrong answer equally and give abstention a fixed reward, ignoring confidence. CalibraRL adds a Brier penalty based on confidence and correctness, so confident errors cost more. For abstention, it estimates from other rollouts of the same question whether the model can answer reliably, rewarding abstention more when it…

AI Search · Score 3.75 · Qianchi Zhang, Hainan Zhang, Liang Pang, Hongwei Zheng, Zhiming Zheng · Source

LightRetriever: A LLM-based Text Retrieval Architecture with Extremely Faster Query Inference

Tue, 27 Oct 2026 08:00:00 +0800

In LLM-based text retrieval, documents can be encoded offline, but queries arrive in real time, so every query runs the full LLM, which keeps throughput low and serving costly. LightRetriever breaks the symmetry: documents still go through the full LLM, while the query side is reduced to a table lookup. During training, a shared prompt plus a single token is fed through the LLM to get that token'…

AI Search · Score 4.65 · Deep dive · 17 min · Guangyuan Ma, Yongliang Ma, Xuanrui Gou, Zhenpeng Su, Ming Zhou, Songlin Hu · Source

System One models Are Not Better Rerankers

Tue, 27 Oct 2026 08:00:00 +0800

Using the new decision models to rerank search results does not beat a purpose-tuned reranker, concludes this University of Passau paper. Decision models, also called System One models, answer a question with a calibrated probability of yes instead of text, and within weeks of release they were used for reranking. The paper compares two fine-tunes of the same backbone: decision fine-tuning trains…

AI Search · Score 4.65 · Deep dive · 17 min · Michael Dinzinger, Kanishka Ghosh Dastidar, Michael Granitzer · Source

Closing the Modality Gap via Simplex-Constrained Representations

Tue, 27 Oct 2026 08:00:00 +0800

In contrastive models like CLIP and CLAP, the two modalities are trained to align yet their embeddings often sit in offset regions, a phenomenon called the modality gap. This paper leaves the backbone untouched and changes only the space the embeddings live in: the unit sphere is replaced by the probability simplex, where every dimension is nonnegative and all dimensions sum to one. In practice a…

AI Search · Score 4.40 · Deep dive · 19 min · Shubham Gupta, Siva Reddy, Perouz Taslakian, Valentina Zantedeschi, Cem Subakan · Source

56 scanned pages to a cited answer in 5 minutes: RAG pipeline for the documents search cannot read

Tue, 27 Oct 2026 08:00:00 +0800

Much enterprise knowledge sits in scanned PDFs that ordinary search cannot read. This Elastic tutorial builds a retrieval-augmented generation pipeline: upload a scanned 56-page plant guide, ask in plain language, and within minutes get an answer citing the page it came from. The key piece is jina-ocr-v1, which reads one page image per pass and outputs Markdown in reading order, keeping tables as…

AI Search · Score 4.30 · Deep dive · 15 min · Source

Beyond Post-Hoc Rewards: Privileged Reasoning Exploration for Multimodal Embeddings

Tue, 27 Oct 2026 08:00:00 +0800

Some multimodal embedding models first have the model write reasoning about the input and then embed it along with the input, producing richer vectors. But training data only says which pairs are relevant, never what reasoning to write, so exploring from the input alone rarely finds reasoning that actually helps retrieval. PILOT has one multimodal LLM play two roles: one writes reasoning from the…

AI Search · Score 4.30 · Deep dive · 17 min · Tianyu Yang, Yijun Tian, Lars Benedikt Kaesberg, Jan Philip Wahle, Bela Gipp, Terry Ruas, Daniel Ku… · Source

Yalda-Embedding: Open Development of a Competitive Persian Retrieval Model

Tue, 27 Oct 2026 08:00:00 +0800

Persian dense retrieval has lagged for lack of dedicated training data and compact embedding models. This paper continues training Qwen3-Embedding-0.6B on over a million Persian query-document pairs, producing the roughly 600-million-parameter Yalda Embedding, with training and data released openly. On eight Persian retrieval tasks from FaMTEB, it has the best average among models its size, beats…

AI Search · Score 4.10 · Yalda Taheri, Mohammad Hassan Heydari, Armon Rasooli · Source

Scalable PDF Retrieval with Cohere Embed 5 and Qdrant Asymmetric Quantization

Tue, 27 Oct 2026 08:00:00 +0800

Many answers in PDFs live in charts and tables that OCR mangles, so the usual approach stores around a thousand patch vectors per page, which is heavy. Qdrant found that Cohere's new Embed 5 Pro, with a single vector per page, already beats several multi-vector models on five public English document sets. Compressing further, they keep only the first 512 dimensions, store just the sign of each, k…

AI Search · Score 4.05 · Source

Mitigating Saliency Collapse: Robust Saliency-Aware Long-Text Image-Text Alignment

Tue, 27 Oct 2026 08:00:00 +0800

When retrieving images with long text, models tend to fixate on the few sentences describing prominent objects and barely use the surrounding context about setting and details. This paper calls it saliency collapse: removing part of the query's salient text cuts retrieval accuracy much faster than removing the same amount at random. RoSA splits each image with object-level visual saliency estimat…

AI Search · Score 4.00 · Qiuyu Kong, Zanxi Ruan, Marco Cristani, Yiming Wang · Source

UAE GovTrust Scope Filtering Gains and Routing Failures in Bilingual Public Service Retrieval

Tue, 27 Oct 2026 08:00:00 +0800

Public-service questions often mention several places, but only one of them decides which service applies, say someone living in one emirate licensing a company in another. This paper builds a controlled English-Arabic retrieval diagnostic for the UAE and compares twelve methods. Deciding which place controls the service, keeping only that place's service entries, then fusing keyword and vector r…

AI Search · Score 3.75 · Aleksandar Mihajlovic · Source

Cycle-of-Thought: Self-Evaluative Iterative Reasoning for Table Visual Question Answering in Document Images

Tue, 27 Oct 2026 08:00:00 +0800

Vision-language models answering questions about tables in document images often misread rows and columns or miscalculate, and fine-tuning needs costly expert labels. Self-RefineVQA needs no training and no labels: one model takes three turns, first answering, then checking like a reviewer whether the answer matches the table, then revising according to the critique, for several rounds. Small mod…

AI Search · Score 3.75 · Somraj Gautam, Gaurav Harit · Source

Semantically Rich, Metrically Affect-Blind: A Geometric Blind Spot in Modern Text Embedders

Mon, 26 Oct 2026 08:00:00 +0800

Search a vector index for what users dislike about a product, and the reviews you get back are usually on topic but a mix of praise and complaints. This paper measures why. Across 11 text embedding models, sentiment is present in the vectors and a simple linear classifier can read it, yet the cosine similarity that retrieval actually uses barely registers it. A small projection head on the frozen…

AI Search · Score 4.65 · Deep dive · 23 min · Mahdi Naser Moghadasi · Source

PersonalQuery-Benchmark:Evaluating Retrieval Robustness to Personalized User Expression Differences

Mon, 26 Oct 2026 08:00:00 +0800

Two shoppers want the same product. One asks for strawberry, lactose-free Greek yogurt in a short phrase; the other writes a longer sentence and misspells a word along the way. Will search return the same results? This paper builds a benchmark, PQB, from real user reviews in three Amazon categories to test whether results hold up when the same need is worded differently. They often do not. With t…

AI Search · Score 4.40 · Deep dive · 26 min · Wenyu Liang, Hongzhi Yin, Tongtong Wu, Fan Liu, Kexin Dai, Weiqing Wang · Source

What the Query Remembers: Query Abstraction Governs Lexical and Dense Retrieval of Classical Arabic Poetry

Mon, 26 Oct 2026 08:00:00 +0800

When searching a classical Arabic poetry archive, which works better: keyword search with BM25 or an embedding model? This paper finds that the answer depends mainly on how much of the original wording the user still remembers. When users half-remember a few lines, BM25 beats all nine embedding models. When users can only describe what the poem is about, BM25 drops to 8th of 10 systems. By rewrit…

AI Search · Score 4.40 · Deep dive · 24 min · Omar Farouk Zouak, Selma Mani, Larbi Said Chikh, Raid Abderrezak Ouahioune · Source

MICE to Meet You: From Re-ranking Only to End-to-End Retrieval

Mon, 26 Oct 2026 08:00:00 +0800

Cross-encoders rerank candidate documents most accurately, but every candidate must be fully processed for every query. This paper gives the faster MICE architecture a document storage scheme, MICE-P, which comes close to cross-encoder ranking quality at about a quarter of the per-query time. In MICE, the first layers encode the query and the document separately, so the document part can be preco…

AI Search · Score 4.30 · Deep dive · 23 min · Victor Morand, Yuxuan Zong, Mathias Vast, Josiane Mothe, Benjamin Piwowarski · Source

Collection-Date Assignment Redirects Answers in Temporal QA

Mon, 26 Oct 2026 08:00:00 +0800

Retrieval-augmented question-answering systems often attach a collection date to web snippets, and this paper finds that this field, which only records the fetch day, can steer an LLM's answer the wrong way, even though it says nothing about when the content changed. Ask which team an ice hockey player played for on a given day, and put both an older and a newer Wikipedia revision in the context.…

AI Search · Score 4.10 · Deep dive · 22 min · Minsu Oh, Seunghui Jwa, Soo Kyun Kim, Yeo-Chan Yoon · Source

From Noisy Retrieval to Decision Evidence: Silver Evidence Demonstrations for In-Context Learning

Mon, 26 Oct 2026 08:00:00 +0800

In retrieval-augmented few-shot decisions, such as making a call against a pile of policy documents, the key evidence is often buried in long, noisy retrieved passages, and the raw passages in the demonstrations add even more distraction. This paper replaces the raw retrieved passages in each demonstration with a compact evidence summary that an LLM writes for that decision, without seeing the ex…

AI Search · Score 4.00 · Masoud Monajatipoor, Fazlolah Mohaghegh, Hamid Reza Hassanzadeh, Akash Uttamrao Bhoite, Robert E. T… · Source

The Bottleneck Is Retrieval, Not Reading: Diagnosing Evidence-Based Generation with RAG for Long Context Question Answering

Mon, 26 Oct 2026 08:00:00 +0800

When retrieval-augmented generation (RAG) fails on long-document questions, is retrieval missing evidence or is the model misreading it? This paper asks models to return line-level citations with each answer and checks both whether the answer is right and whether the cited lines support it. Across 8 frontier proprietary and 3 open-weight models, the bottleneck is mainly retrieval: similarity sear…

AI Search · Score 3.90 · Debopriyo Banerjee, Owais Aijaz, Ahmer Tabassum, Zhuohan Xie, Preslav Nakov, Veselin Stoyanov · Source

Before the Ticket: Grounding HPC User Support in Documentation and Resolved Tickets with an Agentic LLM Assistant

Mon, 26 Oct 2026 08:00:00 +0800

Most questions reaching a supercomputing center's help desk were already answered in the user guides or in someone else's old ticket, but the guides are hard to navigate and users cannot search other people's tickets. This Purdue paper builds an assistant: one tool runs retrieval-augmented generation over the live documentation, another is a Llama-3-8B fine-tuned on historical ticket questions an…

AI Search · Score 3.85 · Elham J. Barezi, Primus C Kabuo, Niveditha Nerella · Source

GeoLEON-CLIP: Graph-Anchored Multimodal Contrastive Learning for Geographic Region Representation

Mon, 26 Oct 2026 08:00:00 +0800

A city area can be described by satellite imagery, by text, or by the road and amenity structure in OpenStreetMap, yet most existing methods align only images and text. This paper proposes GeoLEON-CLIP, which aligns all three views of each grid cell in one vector space and trains both satellite images and text to align with the same cell's map-structure embedding, which anchors images to text. Al…

AI Search · Score 3.45 · Radoslaw Malek, Szymon Soltysiak, Aleksandra Kawala-Sterniuk, Piotr Szymanski · Source

Synthetic-to-Real Text-Based Person Anomaly Search via Multi-Model Retrieval Fusion and Cross-Encoder Reranking

Mon, 26 Oct 2026 08:00:00 +0800

Finding a person in surveillance footage from one sentence means describing both appearance and abnormal behavior. The hard part is that training can use only synthetic image-text pairs, while testing uses real surveillance frames full of near-duplicate neighboring frames and look-alike distractors. This paper fine-tunes a Qwen3-VL embedding model on synthetic descriptions, retrieves by a weighte…

AI Search · Score 3.30 · Eugene Denteh, Blessing Agyei Kyem, Andrews Danyo, Joshua Kofi Asamoah, Richard Dzinyela, Solomon O… · Source

When Agentic Retrieval Fails Silently: Framework and Production Evidence from a Creative AI Assistant