Mining Data for Financial Applications: 4th ECML PKDD Workshop, MIDAS 2019, Würzburg, Germany, September 16, 2019, Revised Selected Papers (Lecture Notes in Computer Science)

Preface
	MIDAS 2019: The 4th Workshop on MIning DAta for financial applicationS
	Workshop Description
Organization
Contents
MQLV: Optimal Policy of Money Management in Retail Banking with Q-Learning
	1 Introduction
	2 Background
	3 Algorithm
	4 Experiments
	5 Related Work
	6 Conclusion
	References
Curriculum Learning in Deep Neural Networks for Financial Forecasting
	1 Introduction
	2 Related Work
	3 Data
		3.1 Data Structure
		3.2 Microsoft Baseline
	4 Methods
		4.1 RNN Model: Encoder-Decoder LSTM
		4.2 CNN Model: Dilated CNN
		4.3 Evaluation
	5 Results
		5.1 World-Wide Error Rates
		5.2 Segment-Level MAPEs
	6 Discussion
	References
Representation Learning in Graphs for Credit Card Fraud Detection
	1 Introduction
	2 Related Work
	3 Inductive Representation Learning-Based Fraud Detection
		3.1 Graph Structure
		3.2 Traditional Network Featurization
		3.3 Transductive Representation Learning
		3.4 From Transductive Towards Inductive Representation Learning
	4 Experimental Setup
		4.1 Dataset
		4.2 Tools
		4.3 Experimental Design
		4.4 Evaluation Metrics
	5 Results
	6 Conclusion
	References
Firms Default Prediction with Machine Learning
	1 Introduction
	2 Related Work
	3 Firm-Default–Prediction Problem
		3.1 Definition of Adjusted Default Status
	4 Data and Methods
		4.1 Dataset Description
		4.2 Machine-Learning Approaches
	5 Experimental Results
		5.1 Evaluation Measures
		5.2 Datasets
		5.3 Balanced Versus Imbalanced Classes
		5.4 Baselines
		5.5 Prediction of Adjusted Default
		5.6 A Practical Application: Probability of Default for Loan Subgroups
	6 Conclusion
	References
Convolutional Neural Networks, Image Recognition and Financial Time Series Forecasting
	1 Introduction
	2 Methods and Data
		2.1 Recurrence Plots
		2.2 CNN Model
		2.3 Datasets
	3 Experiments and Results
		3.1 CNN Specifications
		3.2 Experiment 1: Predicting Direction of SP500
		3.3 Experiment 2: Bankruptcy Detection
	4 Conclusions
	References
Mining Business Relationships from Stocks and News
	1 Introduction
	2 Related Work
	3 Relationships from News
	4 Relationships from Stocks
	5 Comparing Relationship Graphs from Stocks and News
	6 Conclusions
	References
Mining Financial Risk Events from News and Assessing their Impact on Stocks
	1 Introduction
	2 Related Work
	3 Risk Event Extraction
		3.1 Risk Sentence Classification with Word Localization
		3.2 Entity Recognition and Disambiguation
		3.3 Target Entity Identification
	4 Assessing Impact of Risk Events
	5 Stock Movement Prediction
	6 Experiments and Evaluations
		6.1 Data Description
		6.2 Results
	7 Conclusion and Future Work
	References
Monitoring the Business Cycle with Fine-Grained, Aspect-Based Sentiment Extraction from News
	1 Introduction
	2 Literature Review
	3 Data
	4 Fine-Grained, Aspect-Based Sentiment Analysis Approach
	5 Preliminary Findings and Conclusions
	References
Multi-step Prediction of Financial Asset Return Volatility Using Parsimonious Autoregressive Sequential Model
	1 Introduction
	2 Recursive Multi-step Prediction Using GARCH Family of Models
	3 The Model
		3.1 Conditional Density Estimation
		3.2 Single-Step Density Regression Using Gated Recurrent Unit Model
		3.3 Parsimonious Autoregressive Multi-step Density Regression (PA-MS-DR)
		3.4 Training of the PA-MS-DR Model
	4 Experiments
	5 Conclusion
	References
Big Data Financial Sentiment Analysis in the European Bond Markets
	Abstract
	1 Introduction
	2 Data and Methodology
	References
A Brand Scoring System for Cryptocurrencies Based on Social Media Data
	1 Introduction
	2 Web Crawling in ENEAGRID
		2.1 The Software Solution
		2.2 The Virtual Laboratory
	3 Semantic Brand Scoring in ENEAGRID
		3.1 The Metric
		3.2 The Virtual Laboratory
	4 Proposal of Current Development
		4.1 Social Networks Crawling
		4.2 Semantic Brand Score and Cryptocurrencies
	5 Conclusions
	References
Author Index