information retrieval
PulseAugur coverage of information retrieval — every cluster mentioning information retrieval across labs, papers, and developer communities, ranked by signal.
9 day(s) with sentiment data
-
MoE models get smarter pruning, retrieval, and inference efficiency
Researchers are exploring advanced techniques for Mixture-of-Experts (MoE) language models to improve their efficiency and performance. One paper introduces HOPE (Higher-Order Pruning of Experts), a novel pruning object…
-
LLM framework refines dataset extraction for displacement and conflict documents
Researchers have developed a weakly supervised framework to extract dataset mentions from documents related to forced displacement and conflict. This approach uses a lightweight model trained on general literature to id…
-
New framework OpenTI aims to unify table insight extraction research
A new research paper introduces Open Tabular Insight Extraction (OpenTI), a framework designed to unify fragmented research efforts in extracting knowledge from large corpora of tables. The paper formalizes OpenTI by de…
-
New method traces query expansion effects using sparse autoencoder features
Researchers have developed a method to trace the effects of query expansion (QE) in information retrieval systems by analyzing sparse autoencoder (SAE) features. This approach decomposes layer-wise retriever representat…
-
TaxoConf system automates conference program organization using topic taxonomy
Researchers have developed TaxoConf, a novel system designed to automate the labor-intensive process of organizing conference programs. This system leverages a conference-specific topic taxonomy to group accepted papers…
-
ExpertLens framework visualizes MoE embedding spaces for better retrieval explainability
Researchers have introduced ExpertLens, a novel framework designed to enhance the explainability of Mixture-of-Experts (MoE) enhanced dense retrievers used in information retrieval. Unlike existing methods that focus on…
-
Workshop tackles online misinformation with LLMs and credible IR
The 6th Workshop on Reducing Online Misinformation through Credible Information Retrieval (ROMCIR 2026) aims to address the growing problem of information pollution, including fake news, deceptive reviews, and unverifie…
-
New VerTox framework enables verifiable corpus poisoning attacks on AI ranking models
Researchers have developed VerTox, a novel framework that uses verifiable reward-guided reinforcement learning to perform corpus poisoning attacks against neural ranking models. This method injects subtly crafted docume…
-
Research explores creator supply impact on short-video platforms
A research paper titled "Content Exploration Beyond the Feed: Creator Supply and the Shared Corpus" analyzes experiments on a short-video platform to understand the impact of content exploration on creator participation…
-
LLMs show promise for Portuguese legal IR relevance assessment, despite bias
Researchers have developed NormasTCU, a new dataset for Brazilian Portuguese Information Retrieval (IR) that includes 14,469 legal documents and human relevance judgments. The study evaluated the effectiveness of using …
-
GNNs scaled for friend recommendations with multi-hash embeddings and temporal sampling
Researchers have developed a scalable Graph Neural Network (GNN) system for friend recommendation on large social graphs, addressing challenges in production environments. The system utilizes multi-hash ID embeddings to…
-
New method improves multimodal CTR prediction for e-commerce
A new research paper proposes a "Mine-Then-Train" method to improve multimodal representation learning for e-commerce click-through rate (CTR) prediction. Current methods often pre-train multimodal encoders separately f…
-
New research examines IR model robustness to growing document collections
A new arXiv paper explores the robustness of Information Retrieval (IR) models when collections of documents grow. Researchers Emmanouil Georgios Lionis and colleagues introduce the concept that an IR system's effective…
-
New recommender system task explains item ranking differences
Researchers have introduced a new task for recommender systems focused on explaining why one item is ranked higher than another, moving beyond single-item explanations. This approach, termed pairwise interpretation, is …
-
New model offers transparency for recommender system providers
Researchers have developed a new approach to understand how recommender systems expose content to users, focusing on the needs of item providers rather than just recipients. This method uses surrogate modeling to approx…
-
AI agents improve document authoring with structural conditioning, paper finds
A new paper explores the effectiveness of multi-agent systems in authoring formal documents, specifically focusing on tender responses. The research found that while converting documents to structural markup like XML im…
-
Chunking strategies for dense retrieval evaluated for effectiveness and cost · 2 sources tracked
A new paper evaluates eight different chunking strategies for dense retrieval systems, considering not only retrieval effectiveness but also system-level costs like throughput and latency. The research indicates that co…
-
Domain-specific fine-tuning enhances AI models for entity resolution
A new research paper explores the adaptation of general-purpose text embedding models for entity resolution tasks. The study, published on arXiv, investigates domain-specific triplet fine-tuning to improve the models' a…
-
Vector Databases: A Deep Dive into LLM Integration and Applications
Vector databases are essential for Large Language Models (LLMs), particularly for Retrieval-Augmented Generation (RAG). These specialized databases efficiently store, index, and query high-dimensional vectors representi…
-
Vector Databases: The Engine Behind Modern AI Applications
This article provides an in-depth explanation of vector databases, highlighting their crucial role in powering many AI applications. It delves into concepts such as embeddings, nearest neighbor search, and their functio…