Natural Language Processing and Chinese Computing

Name: Natural Language Processing and Chinese Computing | 6th CCF International Conference, NLPCC 2017, Dalian, China, November 8-12, 2017, Proceedings
Brand: Springer
Price: 53.49 EUR
Availability: OnlineOnly

6th CCF International Conference, NLPCC 2017, Dalian, China, November 8-12, 2017, Proceedings

Xuanjing Huang Jing Jiang Dongyan Zhao Yansong Feng Yu Hong(Herausgeber*in)

Springer (Verlag)

Erschienen am 3. Januar 2018

XXII, 966 Seiten

E-Book

PDF mit Wasserzeichen-DRM

Systemvoraussetzungen

978-3-319-73618-1 (ISBN)

53,49 €inkl. 7% MwSt.

Systemvoraussetzungen

für PDF mit Wasserzeichen-DRM

E-Book Einzellizenz

Als Download verfügbar

Beschreibung

Weitere Details

Weitere Ausgaben

Inhalt

Intro
Preface
Organization
Contents
IR/Search/Bot
Jointly Modeling Intent Identification and Slot Filling with Contextual and Hierarchical Information
1 Introduction
2 Models
2.1 LSTM
2.2 Contextual Hierarchical Joint (CHJ) Models
3 Experiments
3.1 Datasets and Settings
3.2 Comparisons with Recent Work
3.3 Tradeoff Between Multi-tasks
3.4 Case Study
4 Conclusion
References
Augmenting Neural Sentence Summarization Through Extractive Summarization
1 Introduction
2 Neural Sentence Summarization
2.1 Sequence-to-Sequence Model
2.2 Encoder
2.3 Decoder
3 Our Models
3.1 Extractive Summarization
3.2 Fusion of Extractive Summaries
4 Experiment
4.1 Dataset
4.2 Implementation
4.3 Training Details
4.4 Experimental Results
5 Related Work
6 Conclusion
References
Cascaded LSTMs Based Deep Reinforcement Learning for Goal-Driven Dialogue
1 Introduction
2 Model
2.1 Overview
2.2 Turn Embedding and Dialogue Embedding
2.3 Learning Method
3 Experiments
3.1 Experimental Settings
3.2 Model Comparison
3.3 Hyper-Parameter Analysis
3.4 Dialogue Embedding Analysis
4 Conclusion
References
Dialogue Intent Classification with Long Short-Term Memory Networks
1 Introduction
2 Related Work
2.1 Dialogue Act Classification
2.2 Memory Network
3 Proposed Model
3.1 Overview
3.2 Hierarchical LSTM
3.3 Memory Augmented Hierarchial LSTM
3.4 Model Training
4 Experiments
4.1 Data and Setup
4.2 Results
4.3 Error Analysis
5 Conclusion
References
An Ensemble Approach to Conversation Generation
1 Introduction
2 System Architecture
2.1 Overview
2.2 Data Preprocessing
2.3 Candidates Selecting
2.4 Embedding Pre-training
2.5 Learn-to-Rank Model
2.6 Generation-Based Method
2.7 Ensemble
3 Experiments
3.1 Experimental Setup
3.2 Results
3.3 Case Study of the Seq2seq Results
3.4 Analysis of the Ensemble Module
4 Conclusions
References
First Place Solution for NLPCC 2017 Shared Task Social Media User Modeling
Abstract
1 Introduction
1.1 Subtask One
1.2 Subtask Two
2 Method
2.1 Subtask One
2.2 Subtask Two
3 Experiments and Results Analysis
3.1 Subtask One
3.2 Subtask Two
4 Conclusion and Future Work
Acknowledgments
References
Knowledge Graph/IE/QA
Large-Scale Simple Question Generation by Template-Based Seq2seq Learning
1 Introduction
2 Task Definition
2.1 Knowledge Bases
2.2 Generating Questions from Triples
3 Pure Template-Based Method
4 Template-Based Neural Generation
4.1 Triple Encoder
4.2 Template Decoder
4.3 Template-Based Seq2seq
5 Experiments
5.1 Dataset and Evaluation Metrics
5.2 Experiment Setup
5.3 Quality Analysis
5.4 Study on Diversity
5.5 Proposed KBQA Corpus
6 Conclusion
References
A Dual Attentive Neural Network Framework with Community Metadata for Answer Selection
1 Introduction
2 Related Work
2.1 Deep Semantic Matching
2.2 Network Representation Learning
2.3 Attention Mechanism
3 Model
3.1 The Problem
3.2 User Community Metadata
3.3 Model Illustration
3.4 Attention Calculation
4 Experiment
4.1 Dataset
4.2 Baselines
4.3 Evaluation Criteria
4.4 Experimental Settings
4.5 Experimental Results
5 Conclusion and Future Work
References
Geography Gaokao-Oriented Knowledge Acquisition for Comparative Sentences Based on Logic Programming
1 Introduction
2 Related Work
3 System Architecture
4 Identifying Comparative Sentences from Geographical Texts
4.1 Identifying Candidate Comparative Sentences
4.2 Filtering Out Non-comparative Sentences
5 Extracting Comparative Elements from the Identified Sentences
6 Experiment
6.1 Datasets
6.2 Labeling
6.3 Identifying Comparative Sentences
6.4 Extracting Comparative Elements
7 Conclusion
References
Chinese Question Classification Based on Semantic Joint Features
Abstract
1 Introduction
2 Related Work
3 Chinese Question Classification with Semantic Joint Features
3.1 Surface Word Features
3.2 Syntactic Trunk Features
3.3 Weighed Word-Embedding Semantic Extension
3.4 Semantic Joint Features
4 Experimental Setup and Results
4.1 Data Description
4.2 Experimental Setup
4.3 Experimental Results
5 Conclusion
Acknowledgment
References
A Chinese Question Answering System for Single-Relation Factoid Questions
1 Introduction
2 Related Work
3 Architecture
3.1 Entity Linking
3.2 Candidate Predicates Generation
3.3 Deep CNNs Architecture
3.4 Ranking
4 Experiment
4.1 Dataset
4.2 Settings
4.3 Results
4.4 Upper Bound Analysis
4.5 Further Experiments
5 Conclusion
References
Enhancing Document-Based Question Answering via Interaction Between Question Words and POS Tags
1 Introduction
2 The Proposed Model
2.1 Word-Level Feature Extraction
2.2 Convolutional and Pooling Module
2.3 Matching Scores Between Questions and Candidate Sentences
2.4 Softmax Output Module
2.5 Cross-Entropy Loss Function
2.6 Dropout
3 Experiments
3.1 Performance on the Validation Dataset
3.2 Performance on the Test Dataset
4 Conclusion
References
Machine Learning
A Deep Learning Way for Disease Name Representation and Normalization
1 Introduction
2 Methods
2.1 Processing Pipeline
2.2 Word2vec and TreeLSTM for Distributed Representation
2.3 Perceptron for Similarity Score
2.4 PLTR for Concept Assignment
2.5 Training Details
3 Datasets and Results
3.1 Datasets
3.2 Results
4 Conclusion
References
Externally Controllable RNN for Implicit Discourse Relation Classification
1 Introduction
2 Related Work
3 Methods
3.1 Model Architecture
3.2 General Sequence Pairs Modeling
3.3 Mutually Guided Sequence Pairs Modeling
3.4 Model Training
4 Experiments
4.1 Settings
4.2 Results
4.3 Discussion
5 Conclusion
References
Random Projections with Bayesian Priors
1 Introduction
2 Related Work
3 Inherent Challenges with High Dimensional Word Vectors
4 Our Contributions
4.1 Bayes Rule and Information Updating
4.2 Our Proposed Algorithm
4.3 Computing the Integral 01 a p(a) L(a) da
4.4 Numerical Analysis of Algorithm and Time Taken
5 Our Experiments
6 Conclusion and Future Work
References
A Convolutional Attention Model for Text Classification
1 Introduction
2 A CNN-Based Attention Model
3 Convolutional-Recurrent Attention Neural Networks
3.1 RNN-Based Sequence Encoder
3.2 CNN-Based Attention Extraction
3.3 Text Classification
4 Experiments
4.1 Datasets
4.2 Model Training and Hyper-parameters
4.3 Baselines
4.4 Results and Analysis
4.5 Case Study
5 Conclusion
References
Shortcut Sequence Tagging
1 Introduction
2 Recurrent Neural Networks for Sequence Tagging
3 Exploration of Shortcuts
3.1 Shortcut Blocks
3.2 Gates Computation
4 Neural Architecture for Sequence Tagging
4.1 Network Inputs
4.2 Network Outputs
5 Experiments
5.1 Combinatory Category Grammar Supertagging
6 Related Work
7 Conclusions
References
Machine Translation
Look-Ahead Attention for Generation in Neural Machine Translation
1 Introduction
2 Neural Machine Translation
3 Model Description
3.1 Concatenation Pattern
3.2 Enc-Dec Pattern
3.3 Dec-Enc Pattern
4 Experiments
4.1 Dataset
4.2 Training Details
4.3 Results on Chinese-English Translation
4.4 Results on English-German Translation
5 Related Work
6 Conclusion
References
Modeling Indicative Context for Statistical Machine Translation
1 Introduction
2 Indicative Context Based Translation Disambiguation
2.1 Model Structure
2.2 Model Training
2.3 Integration into SMT Decoding
3 Experiments
3.1 Setup
3.2 Baselines
3.3 Evaluation on NIST Task
3.4 Analyses of Translation Disambiguation
4 Conclusion and Future Work
References
A Semantic Concept Based Unknown Words Processing Method in Neural Machine Translation
Abstract
1 NMT and the Problem of Unknown Words
1.1 Neural Machine Translation
1.2 The Problem of Unknown Words
2 Framework of Our Method
2.1 WordNet
2.2 Replacement of Unknown Words
2.3 Restore Translation for Unknown Words
3 Experiments
3.1 Settings
3.2 Training Details
3.3 Preliminary Experiments
3.4 Comparative Experiments and Main Results
3.5 Comparison of Translating Details
4 Conclusion and Future Work
Acknowledgments
References
Research on Mongolian Speech Recognition Based on FSMN
Abstract
1 Introduction
2 Acoustic Modeling Based on FSMN
3 i-Vector
4 MMI and sMBR
5 Experiments Setup
5.1 Dataset
5.2 ASR System
6 Experiments
6.1 Baseline Experiments
6.2 Experiments on Different Structure for FSMN
6.3 Experiments with i-Vector Features
6.4 Sequence-Discriminative Training
7 Conclusions
Acknowledgements
References
Using Bilingual Segments to Improve Interactive Machine Translation
Abstract
1 Introduction
2 Interactive Machine Translation
2.1 Prefix-Based IMT
2.2 Segment-Based IMT
2.3 Bilingual Segment Based IMT
3 User Interface
3.1 Overview
3.2 Segment Split-Merging
3.3 Translating Option Re-ranking
3.4 Suffix Predicting
4 Decoding
5 Experiments
5.1 Data Setup
5.2 Evaluation Metrics
5.3 Participants and Procedures
5.4 Results and Analysis
5.5 Comparison with Related Work
6 Conclusion and Future Work
Acknowledgements
References
Vietnamese Part of Speech Tagging Based on Multi-category Words Disambiguation Model
Abstract
1 Introduction
2 Linguistic Features of Vietnamese and Construction of Part of Speech Set
2.1 Linguistic Features of Vietnamese
3 Construction of Vietnamese Part of Speech Tagging Set
4 Construction of Vietnamese POS Tagging Model
4.1 Dictionary of Multi-category Words and Non-multi-category Words
4.2 Multi-categories Words Corpus
4.3 Building POS Tagging Model
5 Construction of the Part of Speech Disambiguation Model
5.1 Selecting the Features for Multi-category Words
5.2 Constructing of Disambiguation Model
5.3 Realizing of Part of Speech Tagging in Vietnamese
6 Experiment and Result Analysis
6.1 Data Sets
6.2 Experimental Evaluation
6.3 Experimental Design
7 Summary
Acknowledgment
References
NLP Applications
Unsupervised Automatic Text Style Transfer Using LSTM
1 Introduction
2 Seq2Seq Model
3 The Proposed Model
4 Experiments
5 Conclusions
References
Optimizing Topic Distributions of Descriptions for Image Description Translation
1 Introduction
2 Methodology
2.1 Framework
2.2 Image Search
2.3 Context Extraction
2.4 Topic Remodelling Based Translation Model Reinforcement
3 Experiments
3.1 Corpus
3.2 Settings
3.3 Results and Analysis
4 Related Work
5 Conclusion and Future Work
References
Automatic Document Metadata Extraction Based on Deep Networks
1 Introduction
2 Related Work
2.1 Document Metadata Extraction
2.2 Deep Neural Networks
3 Problem Definition
4 Deep Learning Model Design
4.1 Sentence Modeling
4.2 Image Modeling
4.3 Sequence Modeling
5 Information Extraction System Based on Deep Learning
5.1 System Overview
5.2 Word and Char Embedding
5.3 Training and Parameters
6 Experiment and Evaluation
6.1 Datasets
6.2 Fine-Tuning
6.3 Experiments on Image
6.4 Experiments on Text
6.5 Experiments on Information Union
7 Conclusions
References
A Semantic Representation Enhancement Method for Chinese News Headline Classification
1 Introduction
2 Related Work
3 Fasttext
4 Semantic Representation Enhancement
4.1 Feature Expansion
4.2 Pretreatment
4.3 Pre-train
4.4 Keyword Expansion
5 Experiments and Results Analysis
5.1 Dataset Sources
5.2 Performance Evaluation Indicators
5.3 Baseline
5.4 Results
5.5 Results Analysis
6 Conclusions
References
Abstractive Document Summarization via Neural Model with Joint Attention
Abstract
1 Introduction
2 Related Work
3 Our Method
3.1 Subword Part
3.2 RNN Encoder-Decoder
3.3 Joint Attention
4 Experiments
4.1 Dataset
4.2 Implementation
4.3 Evaluation
5 Conclusion
Acknowledgments
References
An Effective Approach for Chinese News Headline Classification Based on Multi-representation Mixed Model with Attention and Ensemble Learning
Abstract
1 Introduction
2 Multi-representation Mixed Model Based on Attention Mechanism
2.1 Look-up Layer
2.2 Mixed Encoding Layer
2.3 Attention Layer
2.4 Softmax Classifier
3 Combination Model
3.1 Model Selection
3.2 Strategy in Use
4 Experiments
4.1 Datasets
4.2 Experiment Settings
4.3 Results
4.4 Discussion
5 Conclusion
References
NLP Fundamentals
Domain-Specific Chinese Word Segmentation with Document-Level Optimization
Abstract
1 Introduction
2 Related Work
3 Character-Level and Sentence-Level CWS
3.1 Character-Level CWS with LSTM
3.2 Sentence-Level CWS with ILP
4 Document-Level CWS
5 Experimentation
5.1 Experimental Settings
5.2 Experimental Results on Character-Level and Sentence-Level CWS
5.3 Experimental Results on Domain-Specific Document-Level CWS
6 Conclusion
Acknowledgments
References
Will Repeated Reading Benefit Natural Language Understanding?
1 Introduction
2 Models
2.1 Notations
2.2 Standard BiLSTM Model and Application on Different Tasks
2.3 Repeated Reading Models
3 Experiments
3.1 Part-of-Speech Tagging
3.2 Sentiment Analysis
3.3 Semantic Relation Classification
3.4 Event Extraction
4 Discussion
5 Guidelines for NLPers
6 Conclusion
References
A Deep Convolutional Neural Model for Character-Based Chinese Word Segmentation
1 Introduction
2 The Deep Convolutional Neural Model for CWS
2.1 Position Embeddings
2.2 Deep Representation Module
2.3 Tag Scoring Module
2.4 Dropout
2.5 Tag Prediction and Word Segmentation
2.6 Model Training
3 Multi-task Learning
4 Experiments
4.1 Empirical Comparison with Other Models
4.2 Learning Curves of DCN Model
4.3 Contributions of Techniques
4.4 Shallow versus Deep Representations
5 Conclusion
References
Chinese Zero Pronoun Resolution: A Chain to Chain Approach
1 Introduction
2 Background Knowledge
2.1 ZP Resolution
2.2 Related Work
3 Motivation
4 ZP Resolution: Chain to Chain Approach
4.1 ZP Coreferential Chains Generation
4.2 ZP Coreferential Chains Linking
4.3 Incorporating Additional Chain-Level Features
5 Experimentation and Discussion
5.1 Experimental Setup
5.2 Experimental Results
6 Conclusion and Future Work
References
Towards Better Chinese Zero Pronoun Resolution from Discourse Perspective
1 Introduction
2 Background Knowledge
2.1 Chinese Zero Pronoun
2.2 Discourse Parsing
3 Related Work
4 Baseline
4.1 ZP Detection
4.2 ZP Resolution
5 Discourse-Based Approach
5.1 Motivation
5.2 A EDU-Level Approach to Chinese ZP Detection
5.3 A Discourse Rhetorical Structure-Based Approach to Chinese ZP Resolution
6 Experiments and Discussion
6.1 Experimental Setup
6.2 Experimental Results and Discussion
7 Conclusion
References
Neural Domain Adaptation with Contextualized Character Embedding for Chinese Word Segmentation
1 Introduction
2 Method
2.1 Contextualized Character Embedding
2.2 Character Sequence Auto-Encoder
2.3 Neural Segmenter
2.4 Train Strategy
3 Experiments
3.1 Dataset
3.2 Hyper-parameter Settings
3.3 Differences Between Different Domain
3.4 Main Results
4 Related Work
5 Conclusion
References
BiLSTM-Based Models for Metaphor Detection
1 Introduction
2 Related Work
3 Our Approach
3.1 Three Kinds of Sub-sequences
3.2 BiLSTM-Based Models for Metaphor Detection
3.3 Training Objective
4 Experimental Results
4.1 Dataset and Data Pre-processing
4.2 Model Implementation
4.3 Hyperparameter Settings
4.4 Results
5 Conclusions
References
Hyper-Gated Recurrent Neural Networks for Chinese Word Segmentation
1 Introduction
2 Recurrent Neural Networks for Chinese Word Segmentation
2.1 Embedding Layer
2.2 Recurrent Neural Network Layer
2.3 Inference Layer
3 Hyper-Gated Recurrent Neural Networks for Chinese Word Segmentation
3.1 Model-I: Gate Independent Model
3.2 Model-II: Gate Fusing Model
3.3 Hyper-Gated Recurrent Neural Networks with Enhanced Gates
4 Training
5 Experiments
5.1 Datasets
5.2 Experimental Configurations
5.3 Overall Results
5.4 Effects of Hyper Gates
5.5 Convergency
6 Related Work
7 Conclusion
References
Effective Semantic Relationship Classification of Context-Free Chinese Words with Simple Surface and Embedding Features
1 Introduction
2 System Description
2.1 Surface Features
2.2 Embedding Features
2.3 Learning Algorithm
3 Experiments
3.1 Datasets
3.2 Evaluation Metrics
3.3 Experiments on Training Data
3.4 Results on Test Data
4 Conclusion
References
Classification of Chinese Word Semantic Relations
Abstract
1 Introduction
2 Methodology
2.1 Classification System
2.2 Semantic Relation Classification Based on Dictionary
2.3 Semantic Relation Classification Based on Various Features
2.4 Semantic Relation Classification Based on Linguistic Knowledge
3 Experiment Settings
3.1 Data Set and Evaluation
4 Results and Analysis
5 Conclusion
References
Social Network
Identification of Influential Users Based on Topic-Behavior Influence Tree in Social Networks
1 Introduction
2 Related Work
3 Problem Formulation
4 Topic-Behavior Network Reconstruction
4.1 Relationships Between Users
4.2 Topic Relationship
4.3 Behavioral Relationship
4.4 Topic-Behavior Network Reconstruction
5 Influence Tree Model
5.1 Influence Tree Generation
5.2 Identifying Influential Users
6 Experimental Results
6.1 Experimental Setup
6.2 Performance Analysis
7 Conclusions
References
Hierarchical Dirichlet Processes with Social Influence
1 Introduction
2 Social Hierarchical Dirichlet Processes
2.1 The Preliminaries
2.2 The Proposed Method
2.3 Inference
2.4 Social Influence
3 Experiments
3.1 Data Sets and Settings
3.2 Evaluation Results
4 Conclusions
References
A Personality-Aware Followee Recommendation Model Based on Text Semantics and Sentiment Analysis
1 Introduction
2 Related Work
3 Our Method
3.1 User Micgroblogging Text Analysis
3.2 User Sentiment Analysis
3.3 Personality-Based Factors
4 Personality-Aware Followee Recommendation Model Based on Text Semantics and Sentiment Analysis
5 Experiment
5.1 Experimental Evaluation Criteria
5.2 Experimental Design
5.3 Results
6 Conclusions
References
A Novel Community Detection Method Based on Cluster Density Peaks
1 Introduction
2 Related Work
2.1 Social Network Definition
2.2 Classical Community Detection Methods
2.3 Density Cluster-Based Method
3 Community Detection by Cluster Density Peaks
3.1 Definitions of Local Density and Minimum Climb Distance
3.2 Procedure of the Community Detection by Cluster Density Peaks
4 Experiments
4.1 Datasets
4.2 Evaluation Metrics
4.3 Comparison Methods
4.4 Experiment Results and Analysis
5 Conclusions
References
Text Mining
Review Rating with Joint Classification and Regression Model
Abstract
1 Introduction
2 Related Work
3 Basic LSTM Models for Review Rating
3.1 Basic LSTM Network
3.2 Review Rating with LSTM Classification Model
3.3 Review Rating with LSTM Regression Model
4 Review Rating with Joint Classification and Regression Model
4.1 The Main Task
4.2 The Auxiliary Task
4.3 Joint Learning
5 Experimentation
5.1 Experimental Settings
5.2 Experimental Results
6 Conclusion
Acknowledgments
References
Boosting Collective Entity Linking via Type-Guided Semantic Embedding
1 Introduction
2 Model
2.1 Type-Guided Semantic Embedding
2.2 Collective Entity Linking
3 Experiments
3.1 Experimental Settings
3.2 Results on Chinese Corpora
3.3 Results on English Corpora
4 Related Work
5 Conclusion
References
Biomedical Domain-Oriented Word Embeddings via Small Background Texts for Biomedical Text Mining Tasks
Abstract
1 Introduction
2 Related Work
3 Method
3.1 Extracting Functional Units
3.2 Training Based on Maximum-Margin Integrating Functional Units
3.3 Complexity of the Model
4 Experiments
4.1 Bacteria Biotope Event Extraction
4.2 Biomedical Event Extraction
5 Discussion
6 Conclusion
Acknowledgment
References
Homographic Puns Recognition Based on Latent Semantic Structures
Abstract
1 Introduction
2 Related Work
3 Features of Latent Structures Behind Puns
3.1 Inconsistency Structure
3.2 Ambiguity Structure
3.3 Emotion Structure
3.4 Linguistics Structure
4 Experiments
4.1 Experimental Setting
4.2 Homographic Puns Recognition
4.3 The Effect of Latent Semantic Structures
5 Conclusion and Future Work
Acknowledgments
References
Short Papers
Constructing a Chinese Conversation Corpus for Sentiment Analysis
1 Introduction
2 Construction of Corpus
2.1 Data Collection
2.2 Annotation
3 Automatic Polarity Classification
3.1 Classification Methods
3.2 Evaluation Metrics
4 Evaluation
4.1 Datasets
4.2 Experimental Setup
4.3 Results and Discussion
5 Conclusion and Future Work
References
Improving Retrieval Quality Using PRF Mechanism from Event Perspective
1 Introduction
2 Related Work
3 Proposed EO-PRF Mechanism
3.1 Pre-building Event Language Model
3.2 The Generation of Pseudo-Feedback Set from Event Perspective
3.3 Fitting the Feedback Documents Using Event Mixture Model
3.4 Identifying the Target Event
4 Improving Retrieval Quality with EO-PRF
5 Evaluation
5.1 Experimental Set
5.2 The Effectiveness of EO-PRF to Identify the Target Event
5.3 The Analysis of Retrieval Quality
6 Conclusions and Future Work
References
An Information Retrieval-Based Approach to Table-Based Question Answering
1 Introduction
2 Related Work
3 Task Overview
3.1 Table
3.2 IR-Based TBQA
4 Approach
4.1 Anchor Cell Detection
4.2 Answer Cell Representation
4.3 Answer Cell Ranking
5 Experiment
5.1 Experiment Setting
5.2 Experiment Result
5.3 Error Analysis and Discussion
6 Conclusion and Future Work
References
Building Emotional Conversation Systems Using Multi-task Seq2Seq Learning
1 Introduction
2 Preprocessing
2.1 Data Cleaning
2.2 Response Expansion
3 System Description
3.1 Multi-emotional Conversation Generation
3.2 Ranking Policy
4 Experiments
4.1 Model Details
4.2 Experimental Results and Analysis
4.3 Emotion Interaction Analysis
4.4 Shared Task Evaluation Results
5 Conclusions and Future Works
References
NLPCC 2017 Shared Task Social Media User Modeling Method Summary by DUTIR_923
Abstract
1 Introduction
1.1 POI Recommendation
1.2 Gender Prediction
2 Methods
2.1 Subtask One
2.2 Subtask Two
3 Data Set and Experiments
3.1 Subtask One
3.2 Subtask Two
4 Conclusion and Future Work
Acknowledgments
References
Babbling - The HIT-SCIR System for Emotional Conversation Generation
1 Introduction
2 Rule Based Model
3 Seq2Seq Model
3.1 Multi-hop Attention
3.2 LTS Learning to Start
3.3 Emotion Embeddings
4 Experiments
4.1 Dataset
4.2 Model Implementations
4.3 Results
5 Conclusions and Future Work
References
Unsupervised Slot Filler Refinement via Entity Community Construction
1 Introduction
2 Related Work
3 Methodology
3.1 Query Expansion and Retrieval
3.2 Filler Community Construction
3.3 Query Community Construction
4 Experimental Results
4.1 Settings
4.2 Overall Performance
4.3 Discussion and Analysis
5 Conclusion
References
Relation Linking for Wikidata Using Bag of Distribution Representation
1 Introduction
2 Related Work
2.1 Relation Extraction
2.2 Sentence Similarity
2.3 Entity Linking
3 Problem Definition
4 Relation Linking Using Bag of Distribution Representation
4.1 Preprocessor
4.2 Mention Clustering
4.3 Bag of Distribution Pattern Representing
4.4 Relation Classification
5 Experiments and Evaluations
5.1 Data Sets
5.2 Experiments Setting
5.3 Performance Analysis
6 Conclusions
References
Neural Question Generation from Text: A Preliminary Study
1 Introduction
2 Approach
2.1 Feature-Rich Encoder
2.2 Attention-Based Decoder
2.3 Copy Mechanism
3 Experiments and Results
3.1 Implementation Details
3.2 Human Evaluation
3.3 Results and Analysis
4 Conclusion and Future Work
References
Answer Selection in Community Question Answering by Normalizing Support Answers
1 Introduction
2 Related Work
3 Approach
3.1 Overview
3.2 LSTM
3.3 Attention
3.4 CNN
3.5 Support Answers
4 Experiments
4.1 Setup
4.2 Results and Discussions
5 Conclusion
References
An Empirical Study on Incorporating Prior Knowledge into BLSTM Framework in Answer Selection
1 Introduction
2 Related Work
3 BLSTM Based Answer Selection Framework
4 Incorporating Prior Knowledge into BLSTM Framework in Answer Selection
4.1 Prior Knowledge in Different Levels
4.2 Strategies to Incorporate Prior Knowledge into BLSTM
5 Experiment
5.1 NLPCC DBQA Dataset
5.2 Experimental Setup
5.3 Results and Analysis
6 Conclusion
References
Enhanced Embedding Based Attentive Pooling Network for Answer Selection
1 Introduction
2 Model Architecture
2.1 Convolution Neural Network
2.2 Attentive Pooling Neural Network
2.3 Triplet Ranking Loss Function
2.4 Sampling Strategy
3 Experimental Evaluation
3.1 Dataset
3.2 Embedding
3.3 Result
4 Conclusion
References
A Retrieval-Based Matching Approach to Open Domain Knowledge-Based Question Answering
Abstract
1 Introduction
2 Related Work
3 Our Approach
3.1 Entity Linking
3.2 Relation Chain Inference
3.3 Ranking Answer Entities
4 Experiments
4.1 Dataset
4.2 Setup
4.3 Results
4.4 Error Analysis
5 Conclusion
References
Improved Compare-Aggregate Model for Chinese Document-Based Question Answering
1 Introduction
2 Related Work
3 Model
3.1 Compare-Aggregate Matching Model
3.2 Combined Matching Model
4 Experiment
4.1 Dataset and Evaluation Metric
4.2 Experiment Setup
4.3 Results and Analysis
5 Conclusion
References
Transfer Deep Learning for Low-Resource Chinese Word Segmentation with a Novel Neural Network
1 Introduction
2 Transfer Learning by Leveraging High-Resource Datasets
3 Unified Global-Local Neural Networks
4 Mini-Batch Asynchronous Parallel Learning
5 Experiments
5.1 Setup
5.2 Results and Discussions
6 Conclusions
References
AHNN: An Attention-Based Hybrid Neural Network for Sentence Modeling
1 Introduction
2 Background
2.1 Word Representation
2.2 Convolutional Neural Networks
2.3 Recurrent Neural Networks
3 The Model
3.1 Parallel CNN
3.2 Attention-Based RNN
3.3 Expert Layer
4 Experiment
4.1 NLPCC News Headline Categorization Task
4.2 Training Details
4.3 Experimental Results and Discussions
5 Conclusion
References
Improving Chinese-English Neural Machine Translation with Detected Usages of Function Words
1 Introduction
2 Chinese Function Word Usages and Automatic Recognition
2.1 Chinese Function Word Usages
2.2 Automatic Recognition of Chinese Function Word Usages
3 NMT with Chinese Function Word Usages
3.1 NMT Model
3.2 NMT with Function Word Usages
4 Experiments
4.1 Settings
4.2 Data and Results
5 Related Work
6 Conclusion
References
Using NMT with Grammar Information and Self-taught Mechanism in Translating Chinese Symptom and Disease Terminologies
1 Introduction
2 Related Work
3 Attention-Based NMT
4 Adding POS Features of Source-Side Terminologies
5 Adding Synthetic High-Quality Parallel Data by Self-taught Learning Algorithm
6 Experimental Settings
6.1 Dataset
6.2 Train and Evaluation Details
6.3 Models to be Compared
7 Experimental Results
8 Conclusion and Future Work
References
Learning Bilingual Lexicon for Low-Resource Language Pairs
Abstract
1 Introduction
2 Related Works
3 Methodology
3.1 Model
3.2 Inference
4 Experiments
4.1 Data
4.2 Setup and Evaluation
4.3 Results and Discussion
4.3.1 Overall Performance
4.3.2 Effect of Seed Lexicon Variation
4.3.3 Effect of Result Lexicon Size Variation
5 Conclusion and Outlook
Acknowledgments
References
Exploring the Impact of Linguistic Features for Chinese Readability Assessment
Abstract
1 Introduction
2 Related Work
3 Research Methodology
3.1 Features
3.2 Data Statistics
3.3 Experiments and Evaluation
4 Results and Analysis
5 Conclusions
Acknowledgements
References
A Semantic-Context Ranking Approach for Community-Oriented English Lexical Simplification
1 Introduction
2 Related Work
3 The Ranking Approach
4 Evaluation
4.1 Dataset
4.2 Metrics
4.3 Baselines
4.4 Parameter Tuning
4.5 Results
4.6 Discussions
5 Conclusions
References
A Multiple Learning Model Based Voting System for News Headline Classification
Abstract
1 Introduction
2 Related Work
3 Methodology
3.1 Multiple Classification Models Based Voting
3.2 Model of Convolutional Neural Networks (CNN)
3.3 Model of Gated Recurrent Units (GRU)
4 Experiments
4.1 Datasets
4.2 Experimental Settings
4.3 Experimental Results and Discussion
5 Conclusion
Acknowledgments
References
Extractive Single Document Summarization via Multi-feature Combination and Sentence Compression
Abstract
1 Introduction
2 Related Work
3 Our System
3.1 System Architecture
3.2 Keyword Selection
3.3 Multi-feature Combination
3.4 Sentence Extraction and Compression
4 Experiments
4.1 Multi-feature Combination
4.2 Sentence Compression
4.3 Official Evaluation Results
5 Conclusion
Acknowledgments
References
A News Headlines Classification Method Based on the Fusion of Related Words
Abstract
1 Introduction
2 NBOW Model of Fusion Related Words
2.1 Short Text Representation
2.2 Short Text Extensions
2.3 Related Word Filtering
2.4 Fusion Related Words
2.5 Classification
3 Experiments
3.1 Dataset
3.2 Settings
3.3 Model Variations
3.4 Results and Discussion
4 Conclusion
Acknowledgment
References
Resolving Chinese Zero Pronoun with Word Embedding
Abstract
1 Introduction
2 Related Work
3 Our Methods
3.1 AZP Resolution
3.2 SNN Method
3.3 LSTM Method
4 Experiments
4.1 Preparation Work
4.2 Results and Analysis
5 Conclusion and Future Work
References
Active Learning for Chinese Word Segmentation on Judgements
Abstract
1 Introduction
2 Related Work
3 Data Annotation
4 Active Learning CWS on Judgement
4.1 Framework Overview
4.2 Sample Selection Strategy
4.2.1 Informative Character Selection Strategy
4.2.2 Context Selection Strategy
5 Experimentation
5.1 Experimental Settings
5.2 Experimental Results
5.2.1 The Results of CWS Based on CRF and LSTM
5.2.2 The Results of Active Learning on CWS
6 Conclusion
Acknowledgments
References
Study on the Chinese Word Semantic Relation Classification with Word Embedding
1 Introduction
2 Related Work
3 Model Description
3.1 Pre-Trained Word Embedding
3.2 Linear Regression
3.3 Convolutional Neural Network with Word Embedding
3.4 Parameter Tuning
4 Experiments
4.1 Implementation
4.2 Evaluation Set
4.3 Results
4.4 Discussions
5 Conclusion
References
HDP-TUB Based Topic Mining Method for Chinese Micro-blogs
Abstract
1 Introduction
2 Related Research
3 The HDP-TUB Model
3.1 HDP Topic Model
3.2 HDP-TUB Model
4 Results and Analysis
4.1 Experimental Data
4.2 Evaluation Metric
4.3 Experimental Results
5 Summary
References
Detecting Deceptive Review Spam via Attention-Based Neural Networks
1 Introduction
2 Related Work
3 The Proposed Model
3.1 The Feature Extraction Module
3.2 The Feature Attention Module
4 Experiments
4.1 Datasets and Evaluation Metrics
4.2 Our Model v.s. The State-of-the-Arts Work
4.3 The Effectiveness of the Feature Attention Module
4.4 The Attention Spam Example in Datasets
5 Conclusion
References
A Tensor Factorization Based User Influence Analysis Method with Clustering and Temporal Constraint
1 Introduction
2 Problem Setup
3 User Influence Analysis Model
3.1 Neural Network Clustering Model
3.2 Construction of Tensor User Influence Model
3.3 Factorization of Tensor User Influence Model
3.4 Measurement of Users Influence
4 Experiments
4.1 Datasets
4.2 Baseline
4.3 Precision of User Influence Ranking
5 Conclusion
References
Cross-Lingual Entity Matching for Heterogeneous Online Wikis
1 Introduction
2 Related Work
2.1 Entity Matching
2.2 Cross-Lingual Links Discovery
3 Problem Formulation
4 Cross-lingual Entity Matching
4.1 Candidate Selection
4.2 Feature Extraction
4.3 Candidate Ranking
5 Experiments
5.1 Datasets
5.2 Comparison Methods
5.3 Results
5.4 Equivalence Judgement Evaluation
6 Conclusion
References
A Unified Probabilistic Model for Aspect-Level Sentiment Analysis
1 Introduction
2 Related Work
3 Our Model
3.1 Generative Process
3.2 Inference
3.3 Seeding
3.4 Post-Processing
4 Experiments and Results
4.1 Data Set
4.2 Results
5 Conclusions and Future Work
References
An Empirical Study on Learning Based Methods for User Consumption Intention Classification
Abstract
1 Introduction
2 Problem and Approach Description
2.1 Problem Description
2.2 Implementation Framework
2.3 Text Feature Presentation and Classifier Construction
3 Experiments
3.1 Experiment Setup
3.2 Experiment Results and Analysis
4 Related Work
5 Conclusion and Future Work
Acknowledgment
References
Overview of the NLPCC 2017 Shared Task: Chinese Word Semantic Relation Classification
Abstract
1 Introduction
2 Task Setup
2.1 Dataset Construction
2.2 Experiment Setting
3 Evaluation Results
4 Participating Systems
5 Conclusion
Acknowledgement
References
Overview of the NLPCC 2017 Shared Task: Emotion Generation Challenge
1 Introduction
2 Task Definition
3 Dataset Description
4 Annotation Schema
5 Submission Statistics
6 Evaluation Results
6.1 Overall Results
6.2 Emotion-Specific Results
7 Models from Submission Teams
8 Summary
References
Overview of the NLPCC-ICCPOL 2017 Shared Task: Social Media User Modeling
Abstract
1 Background
2 Data Description
3 Task Description
3.1 Interested Location Prediction Task
3.2 User Profiling Task
4 Evaluation Metrics
5 Evaluation Results
6 Conclusion
References
Overview of the NLPCC 2017 Shared Task: Single Document Summarization
1 Introduction
2 Task
3 Data
4 Participants
5 Evaluation
5.1 Evaluation Metric
5.2 Results
5.3 Some Representative Systems
6 Conclusion
References
Overview of the NLPCC 2017 Shared Task: Chinese News Headline Categorization
1 Task Definition
2 Data
3 Evaluation
4 Baseline Implementations
5 Participants Submitted Results
6 Some Representative Methods
7 Conclusion
References
Overview of the NLPCC 2017 Shared Task: Open Domain Chinese Question Answering
Abstract
1 Background
2 Task Description
2.1 KBQA Task
2.2 DBQA Task
3 Evaluation Metrics
4 Evaluation Results
5 Conclusion
References
Author Index

Systemvoraussetzungen

Als PDF speichern Als Link merken