Sequential recommendation (SR) aims to predict a user's next action by learning from their historical interaction sequences. In real-world applications, these models require periodic updates to adapt to new interactions and evolving user preferences. While incremental learning methods facilitate these updates, they face significant challenges. Replay-based approaches incur high memory and computational costs, and regularization-based methods often struggle to discard outdated or conflicting knowledge. To overcome these challenges, we propose SA-CAISR, a Stage-Adaptive and Conflict-Aware Incremental Sequential Recommendation framework. As a buffer-free framework, SA-CAISR operates using only the old model and new data, directly addressing the high costs of replay-based techniques. SA-CAISR introduces a novel Fisher-weighted knowledge-screening mechanism that dynamically identifies outdated knowledge by estimating parameter-level conflicts between the old model and new data, allowing our approach to selectively remove obsolete knowledge while preserving compatible historical patterns. This dynamic balance between stability and adaptability allows our method to achieve a new state-of-the-art performance in incremental SR. Specifically, SA-CAISR improves Recall@20 by 2.0

Generative recommendation provides a novel paradigm in which each item is represented by a discrete semantic ID (SID) learned from rich content. Most existing methods treat SIDs as predefined and train recommenders under static indexing. In practice, SIDs are typically optimized only for content reconstruction rather than recommendation accuracy. This leads to an objective mismatch: the system optimizes an indexing loss to learn the SID and a recommendation loss for interaction prediction, but because the tokenizer is trained independently, the recommendation loss cannot update it. A natural approach is to make semantic indexing differentiable so that recommendation gradients can directly influence SID learning, but this often causes codebook collapse, where only a few codes are used. We attribute this issue to early deterministic assignments that limit codebook exploration, resulting in imbalance and unstable optimization. In this paper, we propose DIGER (Differentiable Semantic ID for Generative Recommendation), a first step toward effective differentiable semantic IDs for generative recommendation. DIGER introduces Gumbel noise to explicitly encourage early-stage exploration over codes, mitigating codebook collapse and improving code utilization. To balance exploration and convergence, we further design two uncertainty decay strategies that gradually reduce the Gumbel noise, enabling a smooth transition from early exploration to exploitation of learned SIDs. Extensive experiments on multiple public datasets demonstrate consistent improvements from differentiable semantic IDs. These results confirm the effectiveness of aligning indexing and recommendation objectives through differentiable SIDs and highlight differentiable semantic indexing as a promising research direction.

Cold-start recommendation remains a central challenge in dynamic, open-world platforms, requiring models to recommend for newly registered users (user cold-start) and to recommend newly introduced items to existing users (item cold-start) under sparse or missing interaction signals. Recent generative recommenders built on pre-trained language models (PLMs) are often expected to mitigate cold-start by using item semantic information (e.g., titles and descriptions) and test-time conditioning on limited user context. However, cold-start is rarely treated as a primary evaluation setting in existing studies, and reported gains are difficult to interpret because key design choices, such as model scale, identifier design, and training strategy, are frequently changed together. In this work, we present a systematic reproducibility study of generative recommendation under a unified suite of cold-start protocols.

Retrieval-Augmented Generation (RAG) enhances language models with external knowledge, but the lengthy retrieved context inflates the input and degrades inference efficiency. Soft context compression encodes each document into a substantially shorter embedding sequence. However, most existing approaches are trained by distilling outputs from uncompressed RAG systems, inherently limiting their performance relative to the original model. To address this limitation, we propose DEX-Comp, a two-stage training recipe: Pure Distillation warm-starts the compression model on the uncompressed RAG's correct responses only, and Hard Exploration then runs reinforcement learning solely on queries the uncompressed RAG fails, forcing the model to explore computation patterns better suited to compressed representations. On five open-domain QA benchmarks at retrieval depths from top-5 to top-30, DEX-Comp compresses retrieved contexts by 16× and accelerates inference by 4×–24×, while achieving performance comparable to or exceeding the uncompressed RAG baseline across retrieval depths. Ablations and evaluations across diverse datasets and backbones further confirm the contribution of each stage and the generalization of our approach.

Web search inherently involves multiple facets of user satisfaction, relevance, quality, authority, recency and user behavior, yet traditional learning-to-rank (LTR) models often overlook these dimensions, producing suboptimal results. We tackle this “satisfaction-diverse” challenge by unifying content-based signals (e.g., relevance, quality, authority, and recency) and behavioral signals (e.g., click-through and browsing time) within a pre-trained language model (PLM). Specifically, we address three key hurdles: (1) Signal Diversity: Standard methods often combine diverse signals into a single loss function with multiple terms, thereby neglecting the unique distinctions each signal brings to representation learning. (2) Unpaired Fusion: Only an extremely small fraction of query-document pairs include user behavioral signals, together with content-based ones, resulting in sparse and imbalanced supervision. (3) Mixture of Experiences: While a Mixture-of-Experts (MoE) architecture can handle diverse, sparse, and imbalanced supervision, naive aggregation (e.g., average pooling or concatenation) of multiple expert heads might undermine performance, as these signals reflect orthogonal user experiences. To overcome these challenges, we propose CBRank, a Content-and-Behavior fusion learning to Rank framework. The proposed solution employs a pyramid PLM for query–document representations on top of two MoE-based modules–one integrating core ranking features and another incorporating user behavioral feedback–each with multiple expert heads to handle diverse signals under sparse and imbalanced supervision. A dynamic fusion module is then fine-tuned using high-quality satisfaction-annotated query-document pairs to adaptively balance the learned representations and generate satisfaction-oriented ranking scores. Extensive experiments on a large-scale offline dataset demonstrate that our approach consistently outperforms strong baselines. We further conduct rigorous online evaluations using real-world web traffic from a large-scale commercial search engine.

Generative retrieval models perform document retrieval by autoregressively generating document identifiers (DocIDs). This process naturally forms a sequential decision problem, where each decoding step selects a DocID token and the complete token sequence determines the retrieved document. However, retrieval effectiveness is typically evaluated only after the full DocID is generated, creating a mismatch between token-level generation and document-level relevance supervision. As a result, existing reinforcement learning methods for generative retrieval mostly rely on sequence-level rewards, where the same document-level feedback is propagated to all decoding steps. Such coarse-grained feedback makes it difficult to identify which token decisions are responsible for successful or failed retrieval. In this work, we propose a fine-grained reinforcement learning framework for generative retrieval with token-level relevance rewards. Instead of assigning a single reward to the entire generated DocID, we estimate step-wise rewards by measuring how each token decision changes the expected retrieval quality of the corresponding generation trajectory. This enables more precise credit assignment and encourages the policy to favor token decisions that contribute more directly to document-level relevance. We further develop practical reward estimation strategies tailored to the DocID generation process and incorporate them into a policy optimization framework. Experiments on retrieval benchmarks show that our method consistently outperforms sequence-level reward baselines, demonstrating the effectiveness of fine-grained supervision for aligning autoregressive DocID generation with retrieval objectives.

Knowledge Editing has emerged as a promising solution for efficiently updating embedded knowledge in large language models (LLMs). While existing approaches demonstrate effectiveness in integrating new knowledge and preserving the original capabilities of LLMs, they fail to maintain fine-grained irrelevant knowledge facts that share the same subject as edited knowledge but differ in relation and object. This challenge arises because subject representations inherently encode multiple attributes, causing the target and fine-grained irrelevant knowledge to become entangled in the representation space, and thus vulnerable to unintended alterations during editing. To address this, we propose DiKE, a novel approach that Disentangles Knowledge representations for LLM Editing (DiKE). DiKE consists of two key components: a Knowledge Representation Disentanglement (KRD) module that decomposes the subject representation into target-knowledgerelated and -unrelated components, and a Disentanglement-based Knowledge Edit (DKE) module that updates only the target-related component while explicitly preserving the unrelated one. We further derive a closed-form, rank-one parameter update based on matrix theory to enable efficient and minimally invasive edits. To rigorously evaluate fine-grained irrelevant knowledge preservation, we construct FINE-KED, a new benchmark comprising fine-grained irrelevant knowledge at different levels of relational similarity to the edited knowledge. Extensive experiments across multiple LLMs demonstrate that DiKE substantially improves fine-grained irrelevant knowledge preservation while maintaining competitive general editing performance.

Recommender systems and personalized product search are critical components of modern online platforms, where recommender systems proactively surface relevant items while product search allows users to express their needs through queries. Unifying both tasks in a shared model is promising since it can enhance user modeling and item understanding. Previous approaches mainly follow a discriminative paradigm, utilizing shared encoders to process input features and task-specific heads to perform each task. However, this paradigm faces two key challenges: gradient conflict and manual architecture design overhead. From the information theory perspective, these challenges potentially both stem from the same issue — low mutual information between input features and task-specific output during optimization. To address this, we propose GenSR, a novel generative paradigm for unifying search and recommendation (S&R), which leverages task-specific prompts to induce task-conditioned optimization subspaces in the shared model, thereby enhancing mutual information. To construct effective subspaces for each task, GenSR first prepares informative representations for each subspace and then optimizes both subspaces in one unified model. Specifically, GenSR consists of two main modules:(1) Dual Representation Learning, which independently models collaborative and semantic historical information to derive expressive item representations; and (2) S&R Task Unifying, which utilizes contrastive learning together with instruction tuning to generate task-specific outputs effectively. Extensive experiments on three benchmarks show GenSR outperforms state-of-the-art methods across S&R tasks.

The 48th European Conference on Information Retrieval (ECIR 2026) was held in Delft, the Netherlands, during 29 March – 2 April 2026. The conference took place at Lijm & Cultuur, a former glue factory centrally located between the historic city centre and the TU Delft campus; the tutorial day was hosted on the TU Delft campus. ECIR 2026 was an exclusively in-person event, with authors unable to travel supported by proxy presenters, and attracted a record number of 476 participants from academia and industry making it the most attended ECIR till date. Building on recent editions, ECIR 2026 introduced several novelties, most notably the promotion of the IR-for-Good track to a parallel core track of the main conference anchored around the theme What is IR-for-Good? , the addition of a resource paper track, and the integration of IRRJ paper presentations into the main programme. Date: 29 March - 2 April 2026. Website: https://ecir2026.eu.

As advances in Artificial Intelligence (AI), primarily Generative AI, continue to push research and practical boundaries in the research, development, and evaluation of Information Retrieval (IR) systems, continuing to envision how the field could and should develop is a critical question. In the third edition of the Search Futures workshop, we continue the trend of prompting critical discussions of what IR currently is but what it can and should be. As part of this edition, we look back at previous visions of search futures and look at how the community has attained those futures. Moreover, the workshop continues to be a venue for the IR community to work collaboratively to build new visions of the future, address concerns with the trajectory of research, and to continue to further strengthen the field going forward.

LLM-based agents have moved automated program repair (APR) from fixed-context patch generation to interactive repository-level repair. However, existing agentic APR systems still struggle to use execution evidence to guide localization, patch generation, and validation. We propose EviACT (Evidence-to-Action), an agentic APR framework that coordinates three evidence-driven guardrails across repair stages. The retrieval scaffold grounds repair context, the compile gate filters invalid edits, and the test-driven gate checks target-test recovery before full regression. Across four benchmarks, EviACT improves resolve rate over the strongest reported comparable baselines by 1.6-6.0 percentage points and shows 70.1-88.6

Generative retrieval (GR) reformulates the Information Retrieval (IR) task as the generation of document identifiers (docIDs). Despite its promise, existing GR models exhibit poor generalization to newly added documents, often failing to generate the correct docIDs. While incremental training offers a straightforward remedy, it is computationally expensive, resource-intensive, and prone to catastrophic forgetting, thereby limiting the scalability and practicality of GR. In this paper, we identify the core bottleneck as the decoder's ability to map hidden states to the correct docIDs of newly added documents. Model editing, which enables targeted parameter modifications for docID mapping, represents a promising solution. However, applying model editing to current GR models is not trivial, which is severely hindered by indistinguishable edit vectors across queries, due to the high overlap of shared docIDs in retrieval results. To address this, we propose DOME (docID-oriented model editing), a novel method that effectively and efficiently adapts GR models to unseen documents. DOME comprises three stages: (1) identification of critical layers, (2) optimization of edit vectors, and (3) construction and application of updates. At its core, DOME employs a hybrid-label adaptive training strategy that learns discriminative edit vectors by combining soft labels, which preserve query-specific semantics for distinguishable updates, with hard labels that enforce precise mapping modifications. Experiments on widely used benchmarks, including NQ and MS MARCO, show that our method significantly improves retrieval performance on new documents while maintaining effectiveness on the original collection. Moreover, DOME achieves this with only about 60% of the training time required by incremental training, considerably reducing computational cost and enabling efficient, frequent model updates.

The Third Search Futures Workshop [Azzopardi et al., 2026], in conjunction with the Forty-eight European Conference on Information Retrieval (ECIR) 2026, looked into the future of search to ask questions such as: • How can we navigate data privacy in large language model (LLM)-based information retrieval (IR)? • How can we implement agentic IR for proactive knowledge synthesis? • How do we ensure trustworthy information access beyond citations in the age of language models? • How does deep search transition from matching to reasoning? • What is meant by information semantics, knowledge representation, and natural language in a world of LLM-powered search? • What are serendipity engines, and how do they explore proactive web search via LLM agents, retrieval augmented generation (RAG), and simulated user feedback? The third edition of the workshop opened with ten lightning talks from a diverse group of speakers. Rather than traditional paper presentations, these short talks offered concise overviews of emerging ideas and critical insights, enabling a rapid exchange across various topics. The format was designed to spark discussion and expose participants to a broad spectrum of future-facing research directions in a compact timeframe. This report, co-authored by the workshop organizers, presenters, and participants, summarizes the talks and key discussions. Our aim is to share these insights with the broader IR community and help seed further dialogue around the themes raised. Date: 2 April 2026. Website: https://searchfutures.github.io/.

Conversational search needs an understanding of the interaction history between users and systems to retrieve information that aligns with the user’s current query. A prominent challenge in this task lies in addressing irrelevant context, which can obscure the user’s true intent and hinder retrieval accuracy. Existing methods for conversational search rely on the dense retrieval architecture, which encodes the context into fixed-length vectors. However, these approaches are constrained by the inherent limitations of the bi-encoder architecture, making it difficult to mitigate interference from irrelevant contexts effectively. In this paper, we propose Conversational Generative Retrieval for Context Denoising (CGR4CD), a novel framework that leverages a sequence-to-sequence generative retrieval architecture with cross-attention layers to dynamically capture key information during decoding. Our method introduces innovative proposition-based docids and a two-stage training strategy, combining generation and rank losses for enhanced denoising and ranking. Extensive evaluations on three conversational search benchmarks (TopiOCQA, QReCC, and OR-QuAC) demonstrate that CGR4CD achieves state-of-the-art performance, outperforming baselines in 8 out of 9 evaluation metrics. On TopiOCQA, which is characterized by topic switching and high noise, CGR4CD achieves an average improvement of 8.2% over baselines of similar size and surpasses the LLM-based Chatretriever by 4.4%, highlighting its superior context denoising capability. Our approach also maintains parameter efficiency, achieving these results with a T5-large backbone compared to larger LLM-based retrievers. Detailed analysis further validates its superior capability in contextual denoising, particularly in noisy scenarios.

Search and recommendation (S R) are core to online platforms, addressing explicit intent through queries and modeling implicit intent from behaviors, respectively. Their complementary roles motivate a unified modeling paradigm. Early studies to unify S R adopt shared encoders with task-specific heads, while recent efforts reframe item ranking in both S R as conditional generation. The latter holds particular promise, enabling end-to-end optimization and leveraging the semantic understanding of LLMs. However, existing methods rely on full fine-tuning, which is computationally expensive and limits scalability. Parameter-efficient fine-tuning (PEFT) offers a more practical alternative but faces two critical challenges in unifying S R: (1) gradient conflicts across tasks due to divergent optimization objectives, and (2) shifts in user intent understanding caused by overfitting to fine-tuning data, which distort general-domain knowledge and weaken LLM reasoning. To address the above issues, we propose Gradient Multi-Subspace Tuning (GEMS), a novel framework that unifies S R with LLMs while alleviating gradient conflicts and preserving general-domain knowledge. GEMS introduces (1) Multi-Subspace Decomposition, which disentangles shared and task-specific optimization signals into complementary low-rank subspaces, thereby reducing destructive gradient interference, and (2) Null-Space Projection, which constrains parameter updates to a subspace orthogonal to the general-domain knowledge space, mitigating shifts in user intent understanding. Extensive experiments on benchmark datasets show that GEMS consistently outperforms the state-of-the-art baselines across both search and recommendation tasks, achieving superior effectiveness.

Existing measures of how much a text is about a concept read the surface of the text: dictionary word shares, topic proportions, embedding similarities. They score the words a text uses, not the judgment a reader forms about it. Recent work has shown that a gap exists in what Large Language Models (LLMs) know internally versus what they express in their response. This paper asks whether that internal knowledge, read by monitoring the activations of frozen, out-of-the-box LLMs, can stand in for task-specific fine-tuning when measuring concept content, and which extraction method reads it best. We extract such measures via the Recursive Feature Machine (RFM) algorithm and via linear probing, and compare these against an embedding baseline, surface baselines, and the same model's own answer to the question. We demonstrate the approach on financial text, a domain studied extensively and served by established annotated resources, using a human-annotated Environmental, Social and Governance (ESG) dataset. The best linear probe comes within 0.6 percentage points of a fine-tuned domain classifier's accuracy without any task-specific fine-tuning, and outscores the same model's own answer to the question in eleven of twelve comparisons, so the activations carry concept content the response does not report. The simple probe consistently beats the RFM concept vectors, which in turn provide what classification alone does not: a continuous score intended to reflect how strongly a concept is present in a text, whose validation awaits graded labels.

The rapid scaling of large language models (LLMs) has made full fine-tuning increasingly impractical due to its prohibitive computational, storage, and memory costs. Although parameter-efficient fine-tuning (PEFT) methods such as Low-Rank Adaptation (LoRA) significantly reduce the number of trainable parameters, most existing approaches treat layers independently and overlook potential cross-layer redundancy. To address this limitation, this paper aims to develop a more parameter-efficient adaptation strategy that can effectively exploit cross-layer parameter sharing while maintaining model expressiveness. Motivated by our preliminary analysis showing that the optimal sharing granularity varies across tasks, we propose ASLoRA, a cross-layer parameter-sharing framework that combines global sharing with partial adaptive sharing. Specifically, ASLoRA introduces a globally shared low-rank input projection (A) together with selectively merged layer-wise output projections (B), enabling the model to balance global parameter reuse with task-specific adaptation. Extensive experiments on diverse NLP benchmarks demonstrate that ASLoRA consistently outperforms standard LoRA while using only about 20% of the trainable parameters. Further analyses show that our proposed adaptive sharing strategy enhances both adaptation capacity and task-specific adaptability.