Papers for

enterprise search developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

AutoOntic builds organized knowledge graphs automatically from text

AutoOntic: Automatic Ontology Induction from Text for Knowledge Graph Construction

Abstract: Graph-based retrieval-augmented generation depends on the quality of the knowledge graph (KG) it retrieves over, yet schema-free LLM-based extraction pipelines typically produce flat graphs: deduplicated entities and relations with no type hierarchy and no domain/range constraints. We present AutoOntic, a pipeline that automatically constructs an ontology from text by canonicalizing open-vocabulary entity types and relations, inducing a hierarchy over the canonicalized types, and mining relation-level domain and range constraints. Its hierarchy induction procedure prioritizes types using corpus-derived relation-argument profiles and places them through local, LLM-adjudicated parent/child checks, allowing newly inserted types to trigger reassignment of existing parent-child links. The resulting hierarchy constrains entity deduplication and supports the induction of relation-level domain/range constraints. For intrinsic evaluation, we introduce a reference-based protocol that classifies induced ontology types as exact, broader, narrower, or unrelated to reference domain and range classes. On the Wikidata-TekGen and DBpedia-WebNLG subsets of Text2KGBench, over 90% of AutoOntic's induced endpoint types are semantically aligned with the reference classes across two LLM backbones: about half match exactly, while most remaining differences concern granularity. On MINE, knowledge graphs organized using an ontology built by AutoOntic achieve 88% retrieval accuracy. AutoOntic is available at https://github.com/screemix/AutoOntic.

Wed 11 NovComputation and LanguageArtificial IntelligenceMachine Learning
The gist
Building good knowledge graphs from text usually requires organizing the types of things and their relationships carefully. Existing tools often create simple, flat groupings without clear categories or rules. The authors developed AutoOntic, a system that reads text and automatically builds a structured hierarchy of types and their relations, making the knowledge graph more organized. Their method checks which types relate to each other using language models, improving accuracy. Tests show AutoOntic creates types closely matching known classifications and helps knowledge graphs retrieve information more accurately.
Open → 2611.15467v1

Controlled agents improve taxonomy building with adaptive tools and policies

Controlled Knowledge Engineering for Taxonomy Induction with Tool-Grounded LLM Agents and Runtime-Adaptive Policies

Abstract: Taxonomies are structured knowledge representations that support semantic search, recommendation, product catalog organization, and retrieval-augmented generation. Their automated construction from a fixed set of candidate entities requires more than plausible text generation: the resulting hierarchy must be complete, structurally valid, and semantically coherent. Large Language Models (LLMs) are promising for this task, but iterative free-form generation remains vulnerable to omitted or hallucinated entities, invalid hierarchical relations, and inconsistent organizational principles. We introduce TIGA (Taxonomy Induction with Tool-Grounded Agents), a knowledge-engineering architecture that formulates taxonomy induction as an iterative decision process over an explicit state and constrained action space. TIGA combines executable taxonomy-editing tools, an memory that records tool calls, rationales, and outcomes, and a runtime-adaptive organizational policy that the agent revises during inference. An optional global reviewer evaluates the constructed taxonomy against the candidate set and the active policy. We evaluate TIGA on four benchmark taxonomies using two LLM backbones and compare it with single-shot and iterative prompting-based baselines. TIGA achieves the strongest structural-quality results in most evaluated configurations, but does not dominate on every precision measure. The gains are accompanied by substantially higher token consumption, API-call volume, and inference time. Ablation results show consistent benefits from the policy and interaction-memory components, whereas the effect of global review depends on the dataset and the available revision budget. These findings characterize tool grounding and runtime policy adaptation as promising but costly mechanisms for improving the controlled construction of structured knowledge representations.

Tue 10 NovInformation RetrievalArtificial Intelligence
The gist
Organizing information into taxonomies helps computers understand and find things better, but making these structures automatically is difficult because mistakes like missing items or wrong connections can happen. The authors created a system called TIGA that uses language AI models guided by special tools and rules to build better taxonomies step-by-step, checking its own work along the way. This approach leads to taxonomies with better structure and consistency, though it takes more time and computing resources. Their tests show that having memory and adaptable rules during the process helps most, while an overall reviewer sometimes adds value depending on the situation.
Open → 2611.13966v1

Instruction following retrieval improves with natural hard negatives

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

Abstract: Instruction-following information retrieval (IFIR) requires retrievers to rank documents by both a query and its instance-specific instructions, yet instruction-conditioned contrastive supervision remains scarce. Existing synthetic methods typically fix a positive passage and generate additional negatives, overlooking the hard-negative potential of the original positive. We propose Natural Hard-Negative Mining (NHNM), which promotes convertible negatives under query-compatible instructions and repurposes original positives as natural instruction hard negatives. Its automated Candidate Cleaning, Reversal Construction, and Geometry Filtering pipeline produces compact, high-contrast supervision aligned with the pretrained retrieval space. Applied to MSMARCO, NHNM yields NHNM-MARCO, an IFIR training set of nearly 7K instances. On InstructIR and FollowIR, models with 568M to 4B parameters fine-tuned on NHNM-MARCO achieve strong instruction-following performance and match or surpass substantially larger retrievers in several settings. Controlled analyses and ablations attribute these gains to natural hard-negative supervision rather than model scale or training-data volume.

Tue 27 OctArtificial Intelligence
The gist
Finding relevant documents based on specific instructions is hard because there are few ways to teach machines to distinguish subtle differences. The authors found that using original documents as tricky negative examples, rather than just generating synthetic ones, helps models learn better. They created an automated process to find these natural hard negatives and used it to produce a new training set. Models trained on this data performed better at following detailed search instructions, even competing with much larger systems.
Open → 2610.36900v1

Programmable search improves complex question answering and reasoning

Learning to Program Retrievals: Composing Search Primitives via Reinforcement Learning

Abstract: Retrieval-augmented language models increasingly use multi-step search to solve knowledge-intensive tasks, yet existing systems typically expose retrieval through fixed pipelines or collections of atomic tools. We introduce Retrieval Program Learning (RPL), a programmable search framework that allows a policy to compose search primitives into executable programs with local state and control flow. This enables adaptive search that can retrieve broadly, process intermediate results, and selectively construct model context without tying retrieval breadth to context consumption. However, the resulting flexibility creates a difficult exploration problem: useful programs may require several coordinated decisions, while reinforcement learning often supplies only sparse feedback. To address this challenge, we propose program mutation, a grammar-constrained and execution-verified procedure that generates local variants of suboptimal programs sampled during RL, converts them into training trajectories, and distills them back into the policy. We evaluate RPL on open-domain question answering and corpus-level reasoning tasks, comparing it with a range of retrieval and reasoning baselines. RPL consistently improves open-domain QA performance across model sizes and in- and out-of-domain benchmarks, with particularly large gains on multi-hop tasks. Notably, a 4B model trained with RPL outperforms all 8B model baselines. On corpus-level reasoning, RPL achieves the highest answer F1 and document-coverage F1 for both model sizes while substantially reducing context consumption and retrieval turns. Our results show that programmable search expands the retrieval strategies that agents can learn, while program mutation provides an effective mechanism for exploring useful program structures beyond direct on-policy sampling.

Sat 17 OctComputation and LanguageInformation Retrieval
The gist
Answering complex questions often requires searching through a lot of information step-by-step. The authors created a method that teaches computer models how to make flexible, step-by-step search plans like writing a small program. This helps the model look broadly, use intermediate findings smartly, and be more efficient. They also developed a way to improve learning when feedback is rare by tweaking partial plans and learning from them. Their method works better than existing approaches, especially on questions that need multiple steps to answer.
Open → 2610.24854v1

Memory retrieval improves by using neighborhood-aware residuals

NEAR: Neighborhood-Aware Residuals for Memory Retrieval

Abstract: Memory retrieval aims to identify evidence from stored records to answer a query. Stored records with different factual details may share topics, entities, or events. Such shared contents would make records without the required evidence more likely to receive high similarity scores and outrank the correct evidence. In this paper, we use the information shared by these records to construct local references for query-memory matching. We propose NEighborhood-Aware Residuals (NEAR), a training-free framework that uses these references to construct residual representations that complement the original embeddings. Specifically, each memory derives a reference direction from its neighbors, and each query obtains its reference from a precomputed table. Using these references, NEAR constructs projection residuals for both queries and memories. These residuals are combined with the original embeddings for a single approximate nearest-neighbor search over a fixed memory index. Experiments on eight long-context retrieval benchmarks show consistent improvements across nine embedding models. Downstream evaluations show that NEAR improves question-answering performance.

Fri 9 OctInformation Retrieval
The gist
Memory retrieval systems search stored records to find answers to questions, but sometimes similar records with different details confuse the search. The authors propose a method called NEAR that looks at related records nearby to adjust how queries and records are compared. This adjustment helps the system better recognize the right evidence by refining the data representations. Tests show NEAR consistently improves memory retrieval and question answering without needing extra training.
Open → 2610.13314v1

Generated query expansion improves strong sparse search systems performance

Generated Query Expansion Still Helps Strong Sparse Retrieval: A Controlled Study with SPLADE-v3

Abstract: Scientific queries are often brief, while relevant papers use specialized vocabulary. Generated query expansion can bridge this mismatch, but earlier work suggests that its value shrinks as the underlying retriever becomes stronger. We test the four generated formats of term lists, a pseudo-document, multiple pseudo-references, and corpus-steered text all together with SPLADE-v3 on NFCorpus, TREC-COVID, and SciDocs. Every condition searches the same frozen document index and follows the same query-side integration rule and 256-dimension budget, isolating the effect of the added content. All twelve method-collection comparisons improve aggregate nDCG@10, with best relative gains of 4.81%, 8.92%, and 9.47%. Eleven remain significant after Holm correction. The gain persists in 103 of 114 interpolation settings, including every setting that assigns at least 30% of the mixture weight to the original query. Shuffled-text and non-contextual lexical-bag controls also remain above baseline in all 24 aggregate comparisons, showing that the added vocabulary carries most of the benefit. A corpus-induced typed concept graph, by contrast, produces no consistent gain, and its relation, depth, validation, random, and gating controls do not rescue it. Generated vocabulary can therefore complement a strong learned sparse retriever, provided that the original query remains strongly represented.

Tue 29 SeptInformation RetrievalArtificial Intelligence
The gist
Finding scientific papers can be hard because short search queries don’t match the specialized words used in those papers. The authors tested different ways to add more related words to these short queries to help. They found that even with a strong search system called SPLADE-v3, adding generated expansion words helps find better results. The added words need to stay connected to the original query for the benefit to hold. Some complicated methods didn’t work as well as simply adding relevant vocabulary.
Open → 2609.37911v1

Efficient dense vector search boosts multimodal querying in knowledge graphs

Efficient Dense Vector Search within Knowledge Graph Content Embeddings

Abstract: Knowledge graphs are a core component of today's knowledge infrastructure, supporting reasoning and anchoring knowledge systems to verifiable facts. RDF stores and SPARQL engines fulfill this function, enabling a range of retrieval and inference tasks on structured knowledge. Coupling them with Language Models (LMs) extends RAG toward neurosymbolic reasoning, where structured queries gate or re-rank generative outputs. This line of reasoning requires that SPARQL evaluation natively support tensor operations on dense embeddings, enabling multimodal querying and learned similarity-based ranking to be expressed together with graph-structural constraints. This approach is feasible only if the engine can efficiently perform dense vector search. We present QLever-Unified Indexed Vector Embedding Retrieval (QUIVER), an extension to QLever that adds native support for dense vector retrieval within RDF knowledge graphs. It implements three optimizations: engine-level registration of tensor functions, vocabulary-time parsing of JSON-encoded vectors, and a virtual SERVICE that exposes a vector index inside the query. We propose two new benchmarks: an extension of Berlin Sparql Benchmark (BSBM) with text embeddings and an extension of DBpedia with image embeddings. Against the baselines, vocabulary-time parsing alone yields median speedups of up to 41.9x on BSBM and 20x on DBpedia for single-type ranking; adding an approximate nearest-neighbor index yields speedups of 355x on BSBM and 97.8x on DBpedia. The index further makes cross-modal vector joins on DBpedia feasible in seconds, whereas all non-indexed configurations time out.

Mon 21 SeptDatabases
The gist
Searching large knowledge graphs is important for finding facts and reasoning, but it can be slow with complex data like vectors from language or images. The authors improved an existing system by adding fast methods to search dense vectors inside these graphs, making it possible to quickly combine text, images, and structured data in queries. They also created new tests showing their approach can be tens to hundreds of times faster than before. Their work helps connect traditional knowledge bases with modern AI data for better and faster answers.
Open → 2609.24236v1

Mixture-of-experts language models improve search speed and accuracy

Mixture-of-Experts Language Models Can Be Strong and Efficient Retrievers

Abstract: Recent work has shown that fine-tuning decoder-only large language models (LLMs) for retrieval yields strong first-stage retrievers, with effectiveness improving as backbones grow in size. However, every query and document must pass through the full model, so encoding cost increases with model size. Mixture-of-Experts (MoE) LLMs activate only a subset of parameters per token and are widely used to scale generative models, yet remain underexplored as retrievers. We systematically study MoE backbones for retrieval by training MoE and dense LLMs from several families using the same procedure, evaluating them across diverse datasets, and measuring query encoding time under the same serving configuration. We show that MoE retrievers outperform dense retrievers with comparable active parameter counts by up to 3.0 nDCG@10 points on BEIR. One of our strongest MoE retrievers matches an 8B dense retriever with 59% fewer active parameters and 18% lower query encoding time. We further show that the number of experts used for query encoding can be reduced without retraining or re-indexing, retaining more than 99% of retrieval effectiveness while reducing query encoding time by up to 26%. Recent rerankers provide only modest additional gains over strong MoE first stages, which often match or exceed the reranked configurations we evaluate. Together, these results show that MoE LLMs can be strong and efficient first-stage retrievers.

Fri 11 SeptInformation RetrievalComputation and Language
The gist
Searching through large amounts of text quickly and accurately is hard because big language models take a lot of computing power. The authors studied a special kind of model called mixture-of-experts (MoE) that only uses part of its brain when answering, making it faster. They found these MoE models can find relevant information better and more efficiently than regular models of similar size. Also, you can reduce how much of the model is active during search without losing much accuracy. This means MoE models can make searching smarter and quicker at the same time.
Open → 2609.13486v1

Visual document retrieval improves by picking tokens after query arrives

Query-Aware Token Budgeting for Efficient Late-Interaction Visual Document Retrieval

Abstract: Late-interaction visual document retrievers preserve fine-grained page evidence by storing many token embeddings per page, but the resulting storage and query-time interaction costs make large-scale deployment expensive. Pooling document tokens before indexing offers a natural remedy, yet static pooling must decide which visual evidence to preserve before the query is known. We study an alternative: a heavily compressed hot-path index generates candidates, after which query-aware token budgeting operates on the original token sets of the shortlisted pages. We formulate this stage-two selection as a budgeted MaxSim coverage problem, show that a clipped version is monotone submodular, and compare coverage-only, cluster-guided, token-wise, and marginal-gain policies. On ten ViDoRe tasks with ColModernVBERT, direct static pooling reduces macro normalized discounted cumulative gain at rank five from 0.6309 without compression to 0.4738 at a thirty-two-fold pool factor. Under the same candidate-generation regime and a pool-factor-eight-equivalent reranking budget, token top-k recovers 93.93 percent of the full-token score, while greedy marginal-gain selection recovers 98.39 percent. Held-out and leave-one-dataset-out evaluations yield positive greedy improvements over token top-k on every dataset. The latency analysis reveals two useful operating points: token top-k for interactive retrieval and the naive greedy implementation as a quality upper envelope. Together, these results show that late-interaction visual retrieval benefits from query-aware allocation rather than query-agnostic pooling alone.

Mon 7 SeptInformation RetrievalArtificial Intelligence
The gist
Finding relevant images of documents quickly is hard because pages store lots of detailed information, which slows things down and uses lots of storage space. The authors study a two-step method: first, they use a small index to pick some promising pages, then they carefully choose which pieces of those pages to look at based on the actual search query. This method chooses tokens (small units of information) smartly, which keeps the search accurate but much faster. Their tests show this adaptive approach works better than just compressing data without considering the query.
Open → 2609.07262v1