Data Mining: 19th Australasian Conference on Data Mining, AusDM 2021, Brisbane, QLD, Australia, December 14-15, 2021, Proceedings (Communications in Computer and Information Science)

Preface
Organization
Contents
Research Track
Parallel Nonlinear Dimensionality Reduction Using GPU Acceleration
1 Introduction
2 GPU-RAPIDS Platform for UMAP
2.1 GPU-RAPIDS Platform
2.2 Constructing k-NN Graph: UMAP-Learn vs CuML-UMAP
2.3 Get the Nearest Neighbours
2.4 k-NN Graph Optimisation on GPU
2.5 Embedding Optimisation
3 Benchmark Experimental Comparison
3.1 F-MNIST Data Set
3.2 MNIST Digits Data Set
4 A Case Study on Leukemia Diagnostic Data Analysis
5 Conclusion
References
Taking the Confusion Out of Multinomial Confusion Matrices and Imbalanced Classes
1 Introduction
2 Prior Work on Making Sense of Confusion Matrices
3 Factoring the Confusion Matrix Using Class Odds
4 Application and Demonstration
4.1 Eddy's Probabilistic Reasoning Challenge (2 Classes)
4.2 Cancer of Unknown Primary—CUP (17 Classes)
4.3 HAndwritten SYmbols—HASY (379 Classes)
5 Discussion and Conclusions
References
Sharpshooting Most Beneficial Part of AUC for Detecting Malicious Logs
1 Introduction
2 Definition of AUC and pAUC
3 Problem
4 PUMA
4.1 Exact Definition of pAUC
4.2 Formulation
4.3 Training
5 Datasets
6 Evaluation
6.1 Experimental Conditions
6.2 Feature Extraction
6.3 Methods Evaluated
6.4 Experimental Results
7 Related Work
8 Conclusion
References
A Drift Aware Hierarchical Test Based Approach for Combating Social Spammers in Online Social Networks
1 Introduction
2 Background and Related Work
3 An Approach of Spam Drift Detection Based on User Behavior Changes
3.1 Hierarchical Test-Based Approach for Drift Detection
3.2 Drift Detection Based on Kullback-Leibler Divergence
3.3 Drift Detection Based on Feature Similarity Differences.
3.4 Validation of Drifted Users Based on Peer Acceptance
3.5 Drift Adaptation with Newly Predicted Labels
4 Experimental Setup
4.1 Datasets and Criteria for Evaluation
4.2 Experiment Results and Evaluation
5 Discussion
6 Conclusion
References
Hospital Readmission Prediction Using Semantic Relations Between Medical Codes
1 Introduction
2 Related Works
3 Proposed Model
3.1 Basic Notation
3.2 Generate Description Features
3.3 Learning Representations with description Features
3.4 Prediction Model
4 Experiments
4.1 Experimental Setup
4.2 Experimental Results
5 Conclusion
References
HFM++: An Enhanced Holographic Factorization Machine for Recommendation
1 Introduction
2 Preliminary
3 Our HFM++ Model
3.1 Model Architecture
3.2 Model Learning
4 Empirical Evaluation
4.1 Experimental Setup
4.2 Experimental Results
5 Related Work
6 Conclusion
References
Deep Learning for Bias Detection: From Inception to Deployment
1 Introduction
2 Literature Review
3 Language Based Deep Learning Model for Bias Detection
3.1 Neural Network Language Model
3.2 Regularising Classifier with Language Model
4 Empirical Analysis
4.1 Data Collection
4.2 Baseline Models
4.3 Experimental Results: SSW Dataset
4.4 Experimental Results: the iShield.ai Dataset
5 Deployment and Architecture of Integrated System
6 Conclusion
References
Exploring Fusion Strategies in Deep Learning Models for Multi-Modal Classification
1 Introduction
2 Related Work
3 Methodology: Multi-modality Deep Learning Classification
3.1 Feature Extraction
3.2 Fusion Strategy
3.3 Classification
3.4 Complexity Analysis of a Fusion Strategy
4 Experiments
4.1 Datasets
4.2 Baseline Models
4.3 Experimental Set-Up
4.4 Evaluation Criteria
5 Results and discussion
5.1 Sentiment Analysis
5.2 Hate Speech Detection
5.3 Crisis Event Detection
5.4 Fake News Detection
5.5 Final Remarks
6 Conclusion
References
Application Track
Chameleon: A Python Workflow Toolkit for Feature Selection
1 Introduction
1.1 Feature Selection for Classification
1.2 Python Toolkit for Feature Selection and Classification
1.3 Contribution
2 Methods
2.1 Chameleon Toolkit Structure
2.2 Feature Selection Methods
2.3 Feature Selection with Common Features
2.4 Classification Methods
2.5 Performance Metrics
2.6 Data Sets
2.7 Feature Selection Evaluation
2.8 Software
3 Results and Discussion
3.1 Comparing Toolkit Feature Selection Methods
3.2 Evaluating Common Features Selection Method
4 Conclusion
References
PostMatch: A Framework for Efficient Address Matching
1 Introduction
2 Related Work
3 The PostMatch Framework
3.1 `Site' and `Locality' Fields in Addresses
3.2 Address Parsing
3.3 Normalisation
3.4 The `Postparse' Post-Processing Method
3.5 Address Pair Matching
4 Experimental Evaluation
4.1 Data and Experimental Environment
4.2 Experiments on Address Parsing
4.3 Experiments on Address Matching
5 Conclusions
References
Detection of Classical Cipher Types with Feature-Learning Approaches
1 Introduction
2 Related Work
3 Neural Cipher Identifier
3.1 Long Short-Term Memory
3.2 Transformer
3.3 Learning-Rate Schedulers
3.4 Transfer Learning
3.5 Ensemble Learning
4 Conclusion
References
SOMPS-Net: Attention Based Social Graph Framework for Early Detection of Fake Health News
1 Introduction
2 Related Work
3 Dataset
4 Problem Definition
5 Architecture
5.1 Social Interaction Graph
5.2 Publisher and News Statistics
6 Experiments and Results
6.1 Experimental Setup
6.2 Comparison Systems
6.3 Results Analysis
6.4 Early Detection
7 Conclusion
References
The Impact of Sentiment in the News Media on Daily and Monthly Stock Market Returns
1 Introduction
2 Related Literature
2.1 Feedback from Text Media to Financial Markets
2.2 Computational Methods for Sentiment Analysis
3 Methodology
3.1 The Data
3.2 Topic Modelling of the News Media Corpus
3.3 BERT for Sentiment Classification
3.4 Times Series Analysis of Sentiment Scores
4 Results
5 Conclusion
References
Investigation of Topic Modelling Methods for Understanding the Reports of the Mining Projects in Queensland
1 Introduction
2 Related Works
2.1 Document Representation Model
2.2 Recent Studies of Topic Modelling
3 Topic Modelling Approaches
3.1 Two-Dimensional Topic Modelling: LDA and NMF
3.2 Three-Dimensional (3D) Topic Modelling: Tensor Clustering
3.3 Dataset
3.4 Data Preprocessing
3.5 Evaluation Measures
3.6 Topic Keyword Matching
4 Results and Discussion
4.1 Document Group by LDA
4.2 Document Group by Tensor
4.3 Indecisiveness of Tensor Clustering
5 Conclusion
References
A Semi-automatic Data Extraction System for Heterogeneous Data Sources: a Case Study from Cotton Industry
1 Introduction
2 Background and Related Work
3 Automated Discovery of Relevant Information
3.1 Data Acquisition
3.2 Semantic Enrichment Using Chunking and NER
3.3 Query Formulation
4 Empirical Analysis
4.1 Data Sources
4.2 Implementation
4.3 Results: Information Retrieval Performance
5 Conclusion
References
Nonnegative Matrix Factorization to Understand Spatio-Temporal Traffic Pattern Variations During COVID-19: A Case Study
1 Introduction
2 Related Work
3 Nonnegative Matrix Factorization for Traffic Pattern Elicitation
3.1 Nonnegative Matrix Factorization
3.2 NMF for Traffic Pattern Elicitation
3.3 Understanding Traffic Pattern Variations
4 A Case Study
4.1 Dataset
4.2 Evaluation Measures
4.3 Traffic Pattern Results for 2019
4.4 Traffic Pattern Results for 2020
4.5 Traffic Pattern Variation Analysis
5 Conclusion
References
Correction to: Taking the Confusion Out of Multinomial Confusion Matrices and Imbalanced Classes
Correction to: Chapter “Taking the Confusion Out of Multinomial Confusion Matrices and Imbalanced Classes” in: Y. Xu et al. (Eds.): Data Mining, CCIS 1504, https://doi.org/10.1007/978-981-16-8531-6_2
Author Index