CCKS 2021 - Evaluation Track: 6th China Conference on Knowledge Graph and Semantic Computing, CCKS 2021, Guangzhou, China, December 25-26, 2021, ... in Computer and Information Science, 1553)

This volume constitutes papers presented at the Evaluation Track of the 6th China Conference on Knowledge Graph and Semantic Computing, CCKS 2021, held in Guangzhou, China, in December 2021. The 17 competition papers went through a rigorious peer review and were accepted for publication. CCKS 2021 technology evaluation track aims to provide researchers with platforms and resources for testing knowledge and semantic computing technologies, algorithms and systems, promote the technical development in the field of domestic knowledge, and the integration of academic achievements and industrial needs. Preface Organization Contents A Biaffine Attention-Based Approach for Event Factor Extraction 1 Introduction 2 Related Works 3 Approach 3.1 Overview 3.2 Design of Stages 3.3 Our Strategies 4 Experiments 4.1 Data Set 4.2 Implementation 4.3 Competition Result 4.4 Ablation Study 5 Conclusion References Method Description for CCKS 2021 Task 3: A Classification Approach of Scholar Structured Information Extraction from HTML Web Pages 1 Introduction 2 Task Description 3 Relevant Work 4 Structured Information Extraction and Matching 4.1 Task Decomposition and Filter Strategy for Data 4.2 Three Classifications Tasks for Title, Language an Gender 4.3 Emails Tasks 4.4 Homepages Tasks 4.5 Others 5 Conclusions References A Dual-Classifier Model for General Fine-Grained Event Detection Task 1 Introduction 2 Related Work 3 Our Approach 3.1 Overview 3.2 Data Pre-processing 3.3 Dual-Classifier Event Detection Model 3.4 Model Ensemble 4 Experiment 4.1 Dataset 4.2 Hyper-parameters 4.3 Main Result 4.4 Ablation Study 5 Conclusion References A Joint Training Framework Based on Adversarial Perturbation for Video Semantic Tags Classification 1 Introduction 1.1 Dataset 1.2 Evaluation Metric 2 Methodology 2.1 Framework 2.2 Multimodal Feature Representation 2.3 Multimodal Feature Aggregate and Fusion 2.4 Classification Module Design 3 Experiments 4 Conclusion Reference A Multi-modal System for Video Semantic Understanding 1 Introduction 2 Related Work 3 System Description 3.1 VCT Model 3.2 VST Model 3.3 R-Drop Regularization 3.4 Model Integration 4 Experiment 4.1 Data Introduction 4.2 VCT Model Experiment 4.3 VST Model Experiment 5 Conclusion References An Integrated Method of Semantic Parsing and Information Retrieval for Knowledge Base Question Answering 1 Introduction 2 Method 2.1 Information Retrieval 2.2 Semantic Parsing 2.3 Answer Integration 3 Experiments and Results 3.1 Dataset and Experimental Settings 3.2 Mention Recognition Evaluation 3.3 Path Extraction Evaluation 3.4 SPARQL Generation Evaluation 4 Conclusion References Basic Profiling Extraction Based on XGBoost 1 XGBoost 1.1 Introduction 1.2 Basic Theory 2 Homepage 2.1 Introduction 2.2 Feature Design 2.3 Semantic Representation: 3 Email 3.1 Introduction 3.2 Email Feature Selection 4 Language 4.1 Introduction 4.2 Language Selection 4.3 Adaption of Name-Ethnicity Classification 5 Gender 5.1 Introduction 5.2 Gender Feature Selection 5.3 Selection of Classification Models 6 Title 6.1 Introduction 6.2 Title Prediction 7 Evaluation 8 Conclusion Appendix References Data Augmentation Based on Pre-trained Language Model for Event Detection 1 Introduction 2 Related Work 3 Method 3.1 Task Definition 3.2 Event Detection Model 3.3 Data Augmentation 3.4 Pseudo Labelling 3.5 Model Ensemble 4 Experiments 4.1 Dataset 4.2 Experimental Setup 4.3 Experimental Results 5 Conclusion References Does BERT Know Which Answer Beyond the Question? 1 Introduction 2 Method 2.1 Baseline Model 2.2 Analysis of Bad Cases 3 Innovation Strategies 3.1 Adversarial Training for Instances Augmentation 3.2 Multiple Training for Error-Prone and Difficult Instances 3.3 Minimum Edit Distance Search to Increase Target Instances 3.4 All Data Matter 3.5 Medical Word Vector for Data Augmentation 4 Experiments 4.1 Dataset 4.2 Evaluation 4.3 Results 5 Conclusion References End-to-End Pre-trained Dialogue System for Automatic Diagnosis 1 Introduction 2 Related Work 3 Model 3.1 Task Definition 3.2 T5-Medicine Model 3.3 Word Segmentation Based on “Words and Expressions” 3.4 Model Training 4 Experiments 4.1 Dataset 4.2 Data Preprocessing 4.3 Evaluation 4.4 Configuration 4.5 Result 5 Summary References Enhance Both Text and Label: Combination Strategies for Improving the Generalization Ability of Medical Entity Extraction 1 Introduction 1.1 Task Definition 1.2 Overview of Main Challenges and Solutions 2 Our Method 2.1 Basic Model Structure 2.2 Unsupervised Text Mode Enhancement Strategy 2.3 Semi-supervised Label Mode Enhancement Strategy 2.4 Model Training 3 Experiments 3.1 Dataset 3.2 Evaluation 3.3 Pre-processing 3.4 Implementation Details 3.5 Results 4 Conclusions References Knowledge-Enhanced Retrieval: A Scheme for Question Answering 1 Introduction 2 Methodology 2.1 Question Classification 2.2 Named Entity Recognition 2.3 Entity Linking 2.4 Path Generation 2.5 Path Ranking 3 Experiment 3.1 Dataset 3.2 Processing of Training Data 3.3 Question Classification 3.4 Named Entity Recognition and Entity Linking 3.5 Path Generation 3.6 Path Ranking 3.7 Overall Result 4 Conclusion References Multi-label Fine-Grained Entity Typing for Baidu Wikipedia Based on Pre-trained Model 1 Introduction 2 Related Work 2.1 Machine Learning Based Methods 2.2 Neural Based Methods 2.3 Pre-trained Based Methods 3 Approach 3.1 Data Preparation 3.2 Input Preprocess 3.3 Overall Model Structure 3.4 Multi-label Classification 3.5 Multi-classification 3.6 Adversarial Training 3.7 Model Voting 4 Experiment 4.1 Dataset 4.2 Implementation 4.3 Main Result 4.4 Ablation Study 5 Conclusion References Multi-strategies Integrated Information Extraction for Scholar Profiling Task 1 Introduction 2 Task 3 Methods 3.1 Extracting Information from Semi-structured Data 3.2 Recognize Text in Pictures 3.3 Extracting Information from Unstructured Data 4 Experiments 4.1 Data Distribution 4.2 Experiment Result 5 Conclusion References Named Entity Recognition and Event Extraction in Chinese Electronic Medical Records 1 Introduction 2 Task Formalism 2.1 Clinical Named Entity Recognition (CNER) 2.2 Clinical Event Extraction (CEE) 3 Methods 3.1 BERT Encoder 3.2 Conditional Random Fields (CRF) 3.3 Transform of Event Extraction 4 Evalution Metrics 4.1 Clinical Named Entity Recognition 4.2 Clinical Event Extraction 5 Experiments 5.1 Datasets 5.2 Settings 5.3 Results 6 Conclusion References Strategies for Enhancing Generalization Ability of Communication Event Co-reference Resolution 1 Introduction 2 Related Work 3 Model Introduction 3.1 Event Co-reference Framework 3.2 Generalization Ability Enhancement Strategy 4 Experiment 4.1 Data Analysis and Processing 4.2 Analysis of Experimental Results 5 Summary References Unmanned Aerial Vehicle Knowledge Graph Construction with SpERT 1 Introduction 2 Related Work 3 Knowledge Graph Construction 3.1 Schema Construction 3.2 Information Extraction 3.3 Knowledge Graph Visualization 4 Evaluation of Knowledge Graph Construction 4.1 Knowledge Graph Quality Evaluation 4.2 Knowledge Graph Usage Evaluation 4.3 Our Strengths 5 Conclusion References Author Index