Large Language Model (LLM)-based agents have achieved promising performance in web task planning. However, they primarily focus on task completion rather than preference modeling, leading to imprecise personalization. This can be attributed to two facts: (1) Contextual scarcity. Vague instructions and insufficient background of users’ inputs make it difficult to accurately identify users’ intents. (2) Preference unpredictability. It’s hard for a single response path to handle users’ sudden intent changes. To address these issues, we propose SAPA, a Situation-Augmented Personalized web Agent. Specifically, a Situational Intent Grounding module is proposed to enrich the contexts of users’ inputs by incorporating historical records and candidate constraints, effectively alleviating the contextual scarcity. Then, we propose a Task-Driven Semantic Expansion module to explore diverse response paths and reduce preference unpredictability in turn, which is achieved by employing a test-time expansion strategy. We construct SAPA with representative LLMs and conduct extensive experiments on PersonalWAB and Amazon Reviews datasets. On PersonalWAB, SAPA achieves new state-of-the-art performance with improvements of +14.0%, +8.8%, and +1.3% in Overall score on personalized recommendation, search, and question answering, respectively. We further evaluate on three single-category scenarios from Amazon Reviews (Office Products, Beauty and Personal Care, Electronics), where SAPA consistently improves Overall scores by 16%–18% with gains across all task types. Ablation studies and case analyses further confirm that the gains stem from jointly addressing contextual scarcity and preference unpredictability.

Open-Domain Timeline Summarization (Open-TLS) seeks to construct coherent evolutionary narratives for a topic by distilling information from large, unstructured web streams. Existing retrieve-and-generate pipelines often fail to capture logically necessary but semantically distant evidence, resulting in macro-level causal gaps (missing intermediate stages) and micro-level element gaps (underspecified milestones). We propose Active Timeline Completion (ATC), an iterative framework that converts implicit narrative gaps into targeted retrieval questions, enabling reflection-driven evidence acquisition. ATC leverages three complementary mechanisms—parametric causal gap detection, historical trajectory analogy, and dynamic ontology-guided completion—while a Logical Utility Verification module ensures only evidence directly addressing these gaps updates the timeline. Experiments on Open-TLS demonstrate that ATC consistently outperforms strong baselines in both content coverage and temporal alignment, showing that active, question-driven retrieval improves the completeness and coherence of open-domain timelines.

Concept induction, which infers appropriate abstract categories from concrete instances, is both a core ontology engineering task and a testbed for LLM semantic abstraction. Although LLMs offer broad knowledge and reasoning that may ease expert-driven ontology construction, prior work shows that knowledge deficiencies still limit them in specialized domains, while systematic evaluation of their induction failures remains lacking. To address this gap, we propose a familiarity-controlled evaluation framework that uses knowledge probing to separate item familiarity from concept induction ability. The evaluation reveals a systematic tendency to induce overly general concepts: LLMs often produce semantically correct but overly broad labels, and direct property augmentation improves performance without resolving this granularity mismatch. We further propose PPACI, an ontology-prior-guided approach that builds concept prototypes by integrating concept definitions with member item properties. For each query group, PPACI uses item properties to retrieve local item-level and global group-level candidate concepts as induction anchors. Experiments across multiple LLMs show that PPACI reduces overly general predictions by 17–28 percentage points while maintaining competitive induction quality. Our code and data are publicly available at https://github.com/Orchideous77/PPACI.

Low-Rank Adaptation (LoRA) is widely used for fine-tuning large language models, but suffers from catastrophic forgetting. Existing continual learning methods mitigate this by estimating per-parameter importance independently. However, this assumption breaks down under LoRA’s factorization 𝛥 W = BA: infinitely many pairs (B, A) yield the same weight update, making importance scores on A and B separately unreliable. We propose HLoRA, which computes path-integral importance in the combined weight update space, bypassing the unreliable separate estimation on A and B. HLoRA applies hierarchical regularization at two levels: element-wise constraints preserve parameters critical for general knowledge, and layer-wise coefficients adapt regularization strength based on each layer’s aggregate importance. Experiments on scientific, medical, and physical tasks using GPT-J and LLaMA-3 show that HLoRA mitigates catastrophic forgetting while enhancing adaptability, running approximately 20x faster and requiring only 10%–15% of the storage compared to prior methods. Code is released on https://anonymous.4open.science/r/HLoRA-915F.

Small Language Models (SLMs) have been widely deployed in edge computing and mobile devices. Generally, these models need to be continuously updated to adapt to dynamic environments and evolving user needs. Therefore, the performance of SLMs in continual learning has become a critical research focus. Compared to LLMs with rich redundant representation spaces, SLMs are more likely to overwrite previous representations during new task learning due to their fewer attention heads and layers. To address this issue, we propose LightLoRA, a sparse fine-tuning method specifically designed for SLMs. This method constructs a conflict-aware importance mechanism by combining raw importance from SVD decomposition with task conflict degrees based on gradient direction consistency. By selecting and updating heads that provide high contributions and low conflicts, LightLoRA successfully maintains sufficient space for knowledge preservation. Experimental results show that the proposed method reaches state-of-the-art performance across multiple benchmarks without adding extra parameters. Code is released on https://anonymous.4open.science/r/LightLoRA-077C.

Despite extensive safety alignment, Large Language Models (LLMs) often fail against jailbreak attacks. While machine unlearning has emerged as a promising defense by erasing specific harmful parameters, current methods remain vulnerable to diverse jailbreaks. We first conduct an empirical study and discover that this failure mechanism is caused by jailbreaks primarily activating non-erased parameters in the intermediate layers. Further, by probing the underlying mechanism through which these circumvented parameters reassemble into the prohibited output, we verify the persistent existence of dynamic jailbreak paths and show that the inability to rectify them constitutes the fundamental gap in existing unlearning defenses. To bridge this gap, we propose Jailbreak Path Unlearning (JPU), which is the first to rectify dynamic jailbreak paths towards safety anchors by dynamically mining on-policy adversarial samples to expose vulnerabilities and identify jailbreak paths. Extensive experiments demonstrate that JPU significantly enhances jailbreak resistance against dynamic attacks while preserving the model’s utility.

Open-domain timeline summarization (TLS) faces challenges from information overload and data sparsity when processing large-scale textual streams. Existing methods struggle to capture coherent event narratives due to fragmented descriptions and often accumulate noise through iterative retrieval strategies that lack effective relevance evaluation. This paper proposes: Reflective Retrieval-Augmented Timeline Summarization with Causal-Semantic Intergration, which offers a novel perspective for open-domain TLS by time point completion and event element completion. R2A-TLS establishes an initial retrieval, reflection, and deep retrieval system that reduces noise through a double filtering mechanism that iteratively generates a timeline for each text which passes the filtering. Then, the system reflects on the initial timeline with the aim of identifying information gaps through causal chain analysis and FrameNet based element validation. These gaps are reformulated into targeted queries to trigger deep retrieval for refining timeline coherence and density. Empirical evaluation on Open-TLS dataset reveals that our approach outperforms the best prior published approaches.

Large language models (LLMs) generate human-aligned content under certain safety constraints. However, the current known technique “jailbreak prompt” can circumvent safety-aligned measures and induce LLMs to output malicious content. Research on Jailbreaking can help identify vulnerabilities in LLMs and guide the development of robust security frameworks. To circumvent the issue of attack templates becoming obsolete as models evolve, existing methods adopt iterative mutation and dynamic optimization to facilitate more automated jailbreak attacks. However, these methods face two challenges: inefficiency and repetitive optimization, as they overlook the value of past attack experiences. To better integrate past attack experiences to assist current jailbreak attempts, we propose the JailExpert, an automated jailbreak framework, which is the first to achieve a formal representation of experience structure, group experiences based on semantic drift, and support the dynamic updating of the experience pool. Extensive experiments demonstrate that JailExpert significantly improves both attack effectiveness and efficiency. Compared to the current state-of-the-art black-box jailbreak method, JailExpert achieves an average increase of 24% in attack success rate and 2.7 times improvement in attack efficiency.

Citing comprehensively and appropriately has become a challenging task with the explosive growth of scientific publications. Current citation recommendation systems aim to recommend a list of scientific papers for a given text context or a draft paper. However, none of the existing work focuses on already included citations of full papers, which are imperfect and still have much room for improvement. In the scenario of peer reviewing, it is a common phenomenon that submissions are identified as missing vital citations by reviewers. This may lead to a negative impact on the credibility and validity of the research presented. To help improve citations of full papers, we first define a novel task of Recommending Missed Citations Identified by Reviewers (RMC) and construct a corresponding expert-labeled dataset called CitationR. We conduct an extensive evaluation of several state-of-the-art methods on CitationR. Furthermore, we propose a new framework RMCNet with an Attentive Reference Encoder module mining the relevance between papers, already-made citations, and missed citations. Empirical results prove that RMC is challenging, with the proposed architecture outperforming previous methods in all metrics. We release our dataset and benchmark models to motivate future research on this challenging new task.

Multi-Document Scientific Summarization (MDSS) aims to produce coherent and concise summaries for clusters of topic-relevant scientific papers. This task requires precise understanding of paper content and accurate modeling of cross-paper relationships. Knowledge graphs convey compact and interpretable structured information for documents, which makes them ideal for content modeling and relationship modeling. In this paper, we present KGSum, an MDSS model centred on knowledge graphs during both the encoding and decoding process. Specifically, in the encoding process, two graph-based modules are proposed to incorporate knowledge graph information into paper encoding, while in the decoding process, we propose a two-stage decoder by first generating knowledge graph information of summary in the form of descriptive sentences, followed by generating the final summary. Empirical results show that the proposed architecture brings substantial improvements over baselines on the Multi-Xscience dataset.

Few-shot named entity recognition (NER) enables us to build a NER system for a new domain using very few labeled examples. However, existing prototypical networks for this task suffer from roughly estimated label dependency and closely distributed prototypes, thus often causing misclassifications. To address the above issues, we propose EP-Net, an Entity-level Prototypical Network enhanced by dispersedly distributed prototypes. EP-Net builds entity-level prototypes and considers text spans to be candidate entities, so it no longer requires the label dependency. In addition, EP-Net trains the prototypes from scratch to distribute them dispersedly and aligns spans to prototypes in the embedding space using a space projection. Experimental results on two evaluation tasks and the Few-NERD settings demonstrate that EP-Net consistently outperforms the previous strong models in terms of overall performance. Extensive analyses further validate the effectiveness of EP-Net.

Span-based joint extraction models have shown their efficiency on entity recognition and relation extraction. These models regard text spans as candidate entities and span tuples as candidate relation tuples. Span semantic representations are shared in both entity recognition and relation extraction, while existing models cannot well capture semantics of these candidate entities and relations. To address these problems, we introduce a span-based joint extraction framework with attention-based semantic representations. Specially, attentions are utilized to calculate semantic representations, including span-specific and contextual ones. We further investigate effects of four attention variants in generating contextual semantic representations. Experiments show that our model outperforms previous systems and achieves state-of-the-art results on ACE2005, CoNLL2004 and ADE.