Semantic Data Pre-Processing for Machine Learning Based Bankruptcy Prediction Computational Model

Abstract

This paper studies a Bankruptcy Prediction Computational Model (BPCM model) -a comprehensive methodology of evaluating a company's bankruptcy level, which combines storing, structuring and pre-processing of raw financial data using semantic methods with machine learning analysis techniques. Raw financial data are interconnected, diverse, often potentially inconsistent, and open to duplication. The main goal of our research is to develop data pre-processing techniques where ontologies play a central role. We show how ontologies are used to extract and integrate information from different sources, prepare data for further processing, and enable communication in natural language. Our Ontology of Bankruptcy Prediction (OBP Ontology) which provides a conceptual framework for a company's financial analysis, is built in the widely established Protégé environment. An OBP Ontology can be effectively described with a Graph database (DB). A Graph DB expands the capabilities of traditional databases by tackling the interconnected nature of economic data and providing graph-based structures to store information, allowing the effective selection of the most relevant input features for the machine learning algorithm. To create and manage the BPCM Graph DB, we use the Neo4j environment and Neo4j query language, Cypher, to perform feature selection of the structured data. Selected key features are used for the supervised Neural Network with a Sigmoid activation function. The programming of this component is performed in Python. We illustrate the approach and advantages of semantic data preprocessing, applying it to a representative use case.

Developing a Generic Predictive Computational Model using Semantic data Pre-Processing with Machine Learning Techniques and its application for Stock Market Prediction Purposes

“…This system significantly improves the previous construction of a predictive model [9], [10]: the Graph Database component now has a three-layered structure and the model is enriched with the Feedback Loop. However, the use case presented in the above papers can still be considered as a use case for the enriched model.…”

“…In Ontology template graph (Definition IV.1 in [10]) nodes are labelled by "abstract empty containers" while in Ontology full graph (Definition IV.2 in [10]) these "abstract containers" are filled with the values gained from the concrete data (i.e., the specific use case metadata).…”

“…The MLE type choice depends on the particular input dataset characteristics and the specific nature of the analytical objectives (forecasting or classification). For example, for one of the variations of GPCM -Bankruptcy Prediction Computational Model (BPCM) (see [10]), where input features are a company's financial ratios, we use Classical Neural Network to identify which category a company relates to (bankrupt or non-bankrupt).…”

“…A formal physical representation of this Ontology is created in the Protégé environment. It is far more complex than OBP Ontology (developed for the Use Case of companies' bankruptcy prediction in [10]).…”

“…Then hierarchical relationships between entities are added, as well as the relationships that are hidden in the annotations of the Ontology (non-hierarchical). Thus, an Ontology template graph (an empty graph without values and weights) is formed (Definition IV.2 in [10]).…”