Yingli Zhou

A temporal graph is an undirected graph where each edge is associated with a timestamp representing the interaction time. As a fundamental problem in graph analysis, community search (CS) in temporal graphs has received tremendous research attention. Existing CS works on temporal graphs typically focus on identifying cohesive subgraphs (e.g., k -cores) within a specific time window. However, they overlook the specific roles of vertices (e.g., core, hub) and fail to capture their dynamic evolution over time. In this paper, we introduce a novel community model, called temporal structurally similar community (TSC), which models the communities in temporal graphs by explicitly considering the evolution of vertex roles across different time windows. Based on the TSC, we develop fast online and index-based algorithms that support efficient TSC queries over arbitrary time windows. Extensive experiments on real-world datasets demonstrate the effectiveness of the TSC model in revealing meaningful communities and the high efficiency of our proposed algorithms.

Retrieval-Augmented Generation (RAG) effectively integrates external knowledge into large language models (LLMs), enhancing accuracy, adaptability, interpretability, and trustworthiness. The modern RAG systems contain multiple modules with many tunable parameters, each of which can greatly influence the overall performance of downstream applications. To identify the optimal configuration for a given RAG pipeline, existing studies have relied on hyperparameter optimization (HPO) methods. However, these methods overlook parameter dependencies, lack experience-based learning, and incur substantial computational costs. To tackle those issues, we propose HAMMER, a hierarchical memory–guided Monte Carlo Tree Search (MCTS) system. Inspired by human learning, HAMMER employs a hierarchical graph memory to organize experimental insights and integrates it with MCTS for more reliable tuning. We further design a theoretically guaranteed query selection technique to reduce cost while preserving effectiveness. Extensive experiments on eight real-world datasets show that HAMMER improves exact match by up to 20.0% and F1-score by 15.2%, while reducing both tuning time and token consumption by up to 9×.

Densest subgraph discovery (DSD) is a fundamental research topic in network science and graph databases. As a typical variant of DSD, the anchored densest subgraph (ADS) problem aims to detect a densest subgraph that is anchored around user-defined seed vertices. The ADS problem has been shown to be very useful in some personalized applications, such as community search and recommendation. While the ADS problem is very useful, existing algorithms suffer from weak theoretical guarantees and perform poorly in practice. To address these issues, we first propose a novel approximation algorithm with improved time complexity, achieving the same accuracy as existing methods while requiring significantly fewer iterations. To further enhance practical performance, we introduce a graph reduction technique that localizes the ADS search to a much smaller subgraph, while still providing non-trivial theoretical guarantees. Building on this approximation, we also develop an efficient exact algorithm. We have performed an extensive empirical evaluation of our approaches on 12 real large datasets. The results show that our proposed algorithms are up to four orders of magnitude faster than the state-of-the-art.

Large language model (LLM)-based agents that reason, plan, and act through tools, memory, and structured interaction are emerging as a promising paradigm for automating complex workflows. Recent systems such as OpenClaw and Claude Code exemplify a broader shift from passive response generation to action-oriented task execution. Yet as agents move toward open-ended, real-world deployment, relying on fromscratch reasoning and low-level tool calls for every task become increasingly inefficient, error-prone, and hard to maintain. This survey examines this challenge through the lens of agent skills, which we define as reusable procedural artifacts that coordinate tools, memory, and runtime context under taskspecific constraints. Under this view, agents and skills play complementary roles: agents handle high-level reasoning and planning, while skills form the operational layer that enables reliable, reusable, and composable execution. Skills are therefore central to the scalability, robustness, and maintainability of modern agent systems. We organize the literature around four stages of the agent skill lifecycle-representation, acquisition, retrieval, and evolution-and review representative methods, ecosystem resources, and application settings across each stage. We conclude by discussing open challenges in quality control, interoperability, safe updating, and long-term capability management. All related resources, including research papers, opensource data, and projects, are collected for the community in https://github.com/JayLZhou/Awesome-Agent-Skills.

Modern databases often use the LIKE predicate to search text data. However, when the search condition is interrupted by wildcards, the existing search structure can degrade to a worst-case complexity linear to full table scale, resulting in poor performance. Traditional methods, such as B+-trees, fail to handle wildcards at both ends efficiently. Recent advances in language models offer a promising solution. These models can decode complex LIKE patterns into a small set of candidate values, which are then verified in dataset-size-invariant time via hash table lookups, greatly improving efficiency. However, integrating LLMs into databases faces challenges such as high latency, large storage requirements, and sensitivity to data distribution drifts. To address these issues, we propose SMILE, a S mall language M odel I ntegrated L IKE E ngine that learns column-local character distributions through small but exquisite parameters. Our SMILE acts as a neural translator that converts complex LIKE patterns into their corresponding result sets. Our approach achieves asymptotic complexity improvements while preserving SQL LIKE logic. We conduct comprehensive evaluation across diverse datasets to validate the efficacy of our approach. Our compact SMILE, with a parameter size 5 orders of magnitude smaller than large language models, achieves strong LIKE decoding efficiency and quality. Specifically, SMILE obtains high recall ability while accelerating LIKE by 3 orders of magnitude compared to large language models and sequential scans, 1.8-41.6 times faster than trigram indexes, and 2 orders of magnitude faster than B+-trees. Moreover, our model demonstrates robustness against potential data and query distribution drifts.

Influential community (IC) search has gained much attention with applications in many areas, such as event organization, recommendation systems, and biological analysis. The dynamic nature of graphs, with frequent insertions and deletions of vertices and edges, makes searching ICs over dynamic graphs computationally costly. To enable efficient IC search on large graphs, existing works often develop index-based approaches, which first compute all the ICs, a.k.a. IC decomposition, and then organize them into some index structures compactly. However, designed for static graphs, these index structures cannot be maintained efficiently for dynamic graphs, and little attention has been paid to the theoretical analysis of the index maintenance algorithms. To tackle the above issues, in this paper, we study the IC search problem on large dynamic graphs, and maintain the index structures efficiently by developing novel IC maintenance algorithms. We first theoretically show that all existing index maintenance algorithms, for the scenarios of both edge insertion and deletion, are relatively unbounded. We then propose a novel concept, called IC decomposition order (ICD-order), based on which we further develop novel efficient IC maintenance algorithms for the scenarios of edge insertions and deletions, respectively. These algorithms not only effectively reduce the scope of affected vertices for improving efficiency, but also offer more favorable time complexities. The comprehensive experiments on eight real datasets show that our algorithms are up to six and three orders of magnitude faster than state-of-the-art index maintenance algorithms under the edge insertion and deletion scenarios, respectively.

Given a directed graph G , the directed densest subgraph (DDS) problem refers to finding a subgraph from G , whose density is the highest among all subgraphs of G . The DDS problem is fundamental to a wide range of applications, such as fake follower detection and community mining. However, existing DDS solutions often incur significant redundant computations and have weaker theoretical guarantees. To tackle these issues, we present both practically and theoretically efficient DDS discovery algorithms (including both approximation and exact methods) in this paper. Specifically, we first introduce a novel graph reduction technique that locates the DDS into a highly smaller subgraph, with non-trivial theoretical guarantees. We further develop an efficient approximation algorithm by employing gradient projection and theoretically prove that it requires fewer iterations to achieve the same solution accuracy compared to state-of-the-art approximation DDS algorithms. Finally, we propose an efficient exact algorithm based on this novel approximation algorithm. We have performed an extensive empirical evaluation of our approaches on 15 real and 8 synthetic large datasets. The results show that our proposed algorithms are up to two orders of magnitude faster than the state-of-the-art.