A5086972170- Topic Modeling
- Advanced Graph Neural Networks
- Data Quality and Management
- Text and Document Classification Technologies
- Complex Network Analysis Techniques
- Sentiment Analysis and Opinion Mining
- Natural Language Processing Techniques
- Recommender Systems and Techniques
- Cardiac Valve Diseases and Treatments
- Cardiovascular Effects of Exercise
- Advanced Text Analysis Techniques
- Cardiovascular Function and Risk Factors
- Legal Education and Practice Innovations
- Cancer survivorship and care
- Plant Molecular Biology Research
- Lymphatic System and Diseases
- Plant Stress Responses and Tolerance
- Rough Sets and Fuzzy Logic
- Dispute Resolution and Class Actions
- Photosynthetic Processes and Mechanisms
- Marine and coastal plant biology
- Physical Activity and Health
- Human Mobility and Location-Based Analysis
- Algal biology and biofuel production
- E-Learning and Knowledge Management
Institute of Computing Technology
2020-2025
Chinese Academy of Sciences
2020-2025
Ningbo University
2024
Ministry of Agriculture and Rural Affairs
2024
University of Chinese Academy of Sciences
2020-2022
Abstract Background and Aims Physical activity has proven effective in preventing atherosclerotic cardiovascular disease, but its role degenerative valvular heart disease (VHD) remains uncertain. This study aimed to explore the dose–response association between moderate vigorous physical (MVPA) volume risk of VHD among middle-aged adults. Methods A full week accelerometer-derived MVPA data from 87 248 UK Biobank participants (median age 63.3, female: 56.9%) 2013 2015 were used for primary...
Knowledge editing has become a promising approach for efficiently and precisely updating knowledge embedded in large language models (LLMs). In this work, we focus on Same-Subject Editing, which involves modifying multiple attributes of single entity to ensure comprehensive consistent updates entity-centric knowledge. Through preliminary observation, identify significant challenge: Current state-of-the-art methods struggle when tasked with related pieces the same subject. To address lack...
Objective To explore the association of wearable device-measured moderate-to-vigorous intensity physical activity (MVPA) with cardiovascular disease (CVD) risk in long-term cancer survivors. Methods This retrospective analysis involved a prospective cohort 6109 survivors without CVD from UK Biobank accelerometry subsample. The MVPA volume is categorised into four groups based on guideline recommendations (0–75 min/week, 75–150 150–300 ≥300 min/week). Cox proportional hazard models are used...
Knowledge graphs (KGs) are structured representations of diversified knowledge. They widely used in various intelligent applications. In this article, we provide a comprehensive survey on the evolution types knowledge (i.e., static KGs, dynamic temporal and event KGs) techniques for extraction reasoning. Furthermore, introduce practical applications different including case study financial analysis. Finally, propose our perspective future directions engineering, potential combining power...
The flourishing of knowledge graph (KG) applications has driven the need for entity alignment (EA) across KGs. However, heterogeneity practical KGs, characterized by differing scales, structures, and limited overlapping entities, greatly surpasses that existing EA datasets. This discrepancy highlights an oversimplified in current datasets, which obstructs exploration application. In this paper, we study performance methods on highly heterogeneous KGs (HHKGs). Firstly, address settings...
Accurate and consistent evaluation is crucial for decision-making across numerous fields, yet it remains a challenging task due to inherent subjectivity, variability, scale. Large Language Models (LLMs) have achieved remarkable success diverse domains, leading the emergence of "LLM-as-a-Judge," where LLMs are employed as evaluators complex tasks. With their ability process data types provide scalable, cost-effective, assessments, present compelling alternative traditional expert-driven...
The need to judge the relations between two entities at a specific time arises in many natural language understanding and knowledge graph related tasks, where traditional relation extraction (RE) task without considering is not feasible. Therefore, it an important extract dynamic from sentences containing entities. However, existing studies focus on extracting static while ignoring temporal information or encode as sequence infer relation. Considering these limitations of studies, we propose...
Short text classification is an important task in the area of natural language processing. Recent studies attempt to employ external knowledge improve performance, but they ignore correlation between and have poor interpretability. This paper proposes a novel Background Knowledge Graph based method for Text Classification called BaKGraSTeC short, which can not only from graph enrich information, also utilize its structural information through neural network promote understanding texts....
The flourishing of knowledge graph applications has driven the need for entity alignment (EA) across KGs. However, heterogeneity practical KGs, characterized by differing scales, structures, and limited overlapping entities, greatly surpasses that existing EA datasets. This discrepancy highlights an oversimplified in current datasets, which obstructs a full understanding advancements achieved recent methods. In this paper, we study performance methods settings, specifically focusing on...
Entity Alignment (EA) is vital for integrating diverse knowledge graph (KG) data, playing a crucial role in data-driven AI applications. Traditional EA methods primarily rely on comparing entity embeddings, but their effectiveness constrained by the limited input KG data and capabilities of representation learning techniques. Against this backdrop, we introduce ChatEA, an innovative framework that incorporates large language models (LLMs) to improve EA. To address constraints ChatEA...
Hierarchical Text Classification (HTC) is a useful tool for document categorization based on the taxonomic hierarchy. However, current HTC methods treat labels under each category separately, which makes it difficult to model multiple inheritance labels. To solve this problem, we propose Bidirectional Consistency Constraint (BCC) method. BCC aims better handle and class imbalance by ensuring hierarchy-compliant text-to-label mapping through relation consistency constraints balancing loss...
The human-oriented applications aim to exploit behaviors of people, which impose challenges on user modeling integrating social network (SN) with knowledge graph (KG), and jointly analyzing two types data. However, existing representation learning methods merely represent one graphs alone, hence are unable comprehensively consider features both SN KG profiling the correlation between them, resulting in unsatisfied performance downstream tasks. Considering diverse gap difficulty associating...
Abstract Background Despite the escalating incidence of degenerative valvular heart disease (VHD), recommended preventive interventions are conspicuously absent. Physical activity has proven effective in preventing atherosclerotic cardiovascular disease, but its role VHD remains uncertain. This study aimed to explore association between moderate-to-vigorous intensity physical (MVPA) and incident left-sided middle-aged adults from UK biobank. Methods Data wrist-worn accelerometer...