Information Retrieval : 25th China Conference, CCIR 2019, Fuzhou, China, September 20–22, 2019, Proceedings

This book constitutes the refereed proceedings of the 25th China Conference on Information Retrieval, CCIR 2019, held in Fuzhou, China, in September 2019. The 13 full papers presented were carefully reviewed and selected from 45 submissions. Academic research is aimed at the precise acquisition of information and knowledge by human beings. The research results will support national strategic decision-making, promote the development of the Internet and IT fields, enhance the production efficiency of the industry, and have a major impact on various fields of social life. -- Provided by publisher Preface 6 Organization 7 Contents 11 Query Processing and Retrieval 13 Mining User Profiles from Query Log 14 1 Introduction 14 2 Related Work 15 3 Approach 16 3.1 Query Representation 17 3.2 User Representation 18 4 Model Initialization 19 4.1 Embedding Initialization 19 4.2 Topic Initialization 19 5 Model Learning 20 5.1 Addressing Label Noise 20 5.2 Multi-task Learning 20 6 Experiment Setup 21 6.1 Dataset 21 6.2 Compared Methods 21 6.3 Implementation Details 22 6.4 Evaluation Metric 23 6.5 Results and Discussion 24 7 Conclusions 24 References 25 A User Effort Measurement for Query Selection 27 1 Introduction 27 2 Related Works 28 3 User Effort Measurement 29 4 Statistical Results 30 5 Descriptive Examples 32 6 Simulation Results 33 7 Conclusion 34 8 Discussion 35 References 35 Temporal Smoothing: Discriminatively Incorporating Various Temporal Profiles of Queries 37 1 Introduction 37 2 Related Work 39 2.1 Document-Dependent Smoothing Methods 40 2.2 Time-Sensitive Smoothing Methods 41 3 Temporal Smoothing 41 3.1 Temporal Profile 42 3.2 Temporal Query Model 42 3.3 Temporal Smoothing by Using Background Temporal Model 43 3.4 An Unified Solution 43 4 Experiment and Evaluation 44 4.1 Corpus 44 4.2 Queries 44 4.3 Temporal Smoothing Result 45 4.4 Retrieval Result 47 4.5 Analysing and Evaluating 47 5 Conclusions and Future Work 48 References 48 Investigating Query Reformulation Behavior of Search Users 50 1 Introduction 50 2 Related Work 52 3 Dataset 53 4 Session-Level User Reformulation Behavior Analysis 54 4.1 Analysis of Session-Level User Reformulation Pattern 54 4.2 User Reformulation Behavior 58 4.3 Conclusions 60 5 Discussions and Future Work 61 References 61 LTRRS: A Learning to Rank Based Algorithm for Resource Selection in Distributed Information Retrieval 63 Abstract 63 1 Introduction 63 2 Related Works 64 2.1 Large-Document Methods 64 2.2 Small-Document Methods 65 2.3 Supervised Methods 65 3 Framework 65 3.1 Definitions 65 3.2 Architecture 66 3.3 Preprocess Module 66 3.4 Resource Description Module 66 3.5 Learning Module 67 4 Multi-scale Features 67 4.1 Term Matching Features 67 4.2 CSI-Based Features 68 4.3 Topical Relevance Features 69 5 The Proposed Algorithm 69 6 Experiments 71 6.1 Dataset 71 6.2 CSI Setup 71 6.3 Result Analysis 71 6.3.1 Performance Comparison 72 6.3.2 Feature Analysis 72 7 Conclusion 73 Acknowledgement 73 References 73 Knowledge and Entities 75 Simplified Representation Learning Model Based on Parameter-Sharing for Knowledge Graph Completion 76 1 Introduction 76 2 Notation and Definition 77 3 Related Work 78 4 Simplified Representation Learning Model Based on Parameter-Sharing for Knowledge Graph Completion 79 4.1 Overview 80 4.2 Optimization 81 5 Experiments 82 5.1 Experiment Settings 82 5.2 Link Prediction 83 5.3 Triple Classification 84 6 Conclusions 85 References 85 Document-Level Named Entity Recognition by Incorporating Global and Neighbor Features 88 1 Introduction 88 2 Related Work 90 2.1 Sentence-Level NER 90 2.2 Document-Level NER 90 3 Our Document-Level NER Model: GNG 91 3.1 Sentence-Level Bi-directional LSTM 91 3.2 Document-Level Module 92 3.3 CRF Layer 94 4 Experiments 95 4.1 Dataset 95 4.2 Baselines 95 4.3 Parameter Setting 96 4.4 Results and Discussion 96 5 Conclusion 98 References 98 NLP for IR 101 Ensemble System for Identification of Cited Text Spans: Based on Two Steps of Feature Selection 102 Abstract 102 1 Introduction 102 2 Related Work 103 2.1 Feature Selection in CTS Identification 103 2.2 Solutions to Class Imbalance Data Problem 104 3 Methodology 104 3.1 Feature Selection: Correlation Analysis of Features 105 3.2 Data Negative Sampling: The Imbalance of Dataset 107 3.3 Classifier-Feature Selection 108 3.4 Ensemble of Results of Basic Classifiers 109 4 Experiments and Results Analysis 109 4.1 Dataset and Preprocessing 109 4.2 Results Analysis 110 5 Discussion 112 5.1 Correlation Analysis-Based Feature Selection 112 5.2 Classifier-Feature Selection 112 5.3 Ensemble of Results of Basic Classifiers 112 6 Conclusion and Future Work 112 Acknowledgements 113 References 113 Joint Learning for Aspect Category Detection and Sentiment Analysis in Chinese Reviews 115 Abstract 115 1 Introduction 115 2 Related Work 117 3 Our Model 118 3.1 Architecture of Our Model 118 3.2 Character-Level Representation and Word-Level Representation 119 3.3 Highway Network and Sentence-Level Representation 120 3.4 Joint Learning 120 4 Experiment Results and Analysis 121 4.1 Dataset 121 4.2 Experiment Setup 122 4.3 Evaluation Index 122 4.4 Baseline Models 123 4.5 Experimental Results and Analysis 124 4.5.1 Experimental Results of Different Models on the Dataset 124 4.5.2 Validity of Character Level Representation 124 4.5.3 Word Embedding Impact Analysis 125 5 Conclusion 125 Acknowledgments 126 References 126 Selecting Paragraphs to Answer Questions for Multi-passage Machine Reading Comprehension 128 1 Introduction 128 2 Related Work 130 2.1 Datasets 130 2.2 Neural Network Models for Machine Reading Comprehension 130 3 The Proposed Approach 131 3.1 Task Definition 131 3.2 Question and Paragraph Modeling 131 3.3 Objective Function 133 4 Experiments 134 4.1 Dataset and Evaluation Metrics 134 4.2 Experimental Settings 134 4.3 Baselines 135 4.4 Results and Analyses 135 4.5 Case Study 137 5 Conclusion and Future Work 137 References 138 Social Computing 140 Temporal Convolutional Networks for Popularity Prediction of Messages on Social Medias 141 1 Introduction 141 2 Related Works 143 2.1 Feature-Driven Method 143 2.2 Point Process Method 143 3 Models 144 3.1 Problem Formulation 144 3.2 Temporal Convolutional Networks 145 4 Experiment Setup 146 4.1 Datasets 146 4.2 Baselines 147 4.3 Experiment Setup 147 4.4 Evaluation Metrics 148 5 Experimental Results 148 5.1 Prediction Performance 148 5.2 Effecitive Memory Length 149 5.3 Automatic Interval-Selection Characteristic of TCN 150 5.4 Efficiency Comparison 151 6 Conclusion 151 References 152 A Method for User Avatar Authenticity Based on Multi-feature Fusion 154 1 Intruction 154 2 Related Work 155 3 Method 156 3.1 An Overview of Proposed Method 156 3.2 User-Based Feature 156 3.3 Text-Based Feature 157 3.4 Avatar-Based Feature (Gender(u)) 159 3.5 Data Set Rebalance 159 3.6 Normalized 159 3.7 Summary 160 4 Experiment 160 4.1 Data Set 160 4.2 Evaluation Preparation 161 4.3 Evaluation Result 162 5 Conclusions and Future Work 164 References 165 Multi-granularity Convolutional Neural Network with Feature Fusion and Refinement for User Profiling 167 Abstract 167 1 Introduction 167 2 Related Work 168 3 Methodology 169 3.1 Multi-granularity User Feature Extraction 170 3.2 Feature Fusion Layer 171 3.3 Integrated Output Layer 174 4 Experiments 174 4.1 Experimental Setup 174 4.2 Experimental Results 175 5 Conclusion and Future Work 176 Acknowledgements 177 References 177 Author Index 179 Front Matter ....Pages i-xii Front Matter ....Pages 1-1 Mining User Profiles from Query Log (Minlong Peng, Jun Zhao, Qi Zhang, Tao Gui, Xuanjing Huang, Jinlan Fu)....Pages 3-15 A User Effort Measurement for Query Selection (Shusi Yu, Ting Jin, Zhefu Shi, Jing Li, Jin Pan)....Pages 16-25 Temporal Smoothing: Discriminatively Incorporating Various Temporal Profiles of Queries (Wang Pengming, Chen Qing, Wang Bin)....Pages 26-38 Investigating Query Reformulation Behavior of Search Users (Jia Chen, Jiaxin Mao, Yiqun Liu, Min Zhang, Shaoping Ma)....Pages 39-51 LTRRS: A Learning to Rank Based Algorithm for Resource Selection in Distributed Information Retrieval (Tianfeng Wu, Xiaofeng Liu, Shoubin Dong)....Pages 52-63 Front Matter ....Pages 65-65 Simplified Representation Learning Model Based on Parameter-Sharing for Knowledge Graph Completion (Yashen Wang, Huanhuan Zhang, Yifeng Li, Haiyong Xie)....Pages 67-78 Document-Level Named Entity Recognition by Incorporating Global and Neighbor Features (Anwen Hu, Zhicheng Dou, Ji-rong Wen)....Pages 79-91 Front Matter ....Pages 93-93 Ensemble System for Identification of Cited Text Spans: Based on Two Steps of Feature Selection (Jin Xu, Chengzhi Zhang, Shutian Ma)....Pages 95-107 Joint Learning for Aspect Category Detection and Sentiment Analysis in Chinese Reviews (Zihang Zeng, Junteng Ma, Minping Chen, Xia Li)....Pages 108-120 Selecting Paragraphs to Answer Questions for Multi-passage Machine Reading Comprehension (Dengwen Lin, Jintao Tang, Kunyuan Pang, Shasha Li, Ting Wang)....Pages 121-132 Front Matter ....Pages 133-133 Temporal Convolutional Networks for Popularity Prediction of Messages on Social Medias (Jiangli Shao, Huawei Shen, Qi Cao, Xueqi Cheng)....Pages 135-147 A Method for User Avatar Authenticity Based on Multi-feature Fusion (Weinan Zhang, Lianhai Wang, Yongli Zan)....Pages 148-160 Multi-granularity Convolutional Neural Network with Feature Fusion and Refinement for User Profiling (Bo Xu, Michael M. Tadesse, Peng Fei, Hongfei Lin)....Pages 161-172 Back Matter ....Pages 173-173