With a glut of competing priorities, the financial industry faces major challenges in extracting timely, relevant, and specifically-focused information from text. Without clear-cut business cases, making the investment in text analysis methods does not justify the return on investment. Furthermore, the business landscape continues to become increasingly complex, and at a rapid pace of change. Thus, difficulties in managing and extracting domain knowledge through heavily people-based systems leads to increased risk, the impacts of which range in billions of dollars of waste from lost business opportunity or even regulatory fines. The increased complexity provides an unprecedented opportunity to use a focused and content-rich ontology around financial markets concepts as context for meaningfully connecting ideas found in text. Ontologies encompass a formal naming of domain concepts and relations in a manner that should improve problem solving within the domain. By infusing ontologies into text analytics, the outcome is more likely to yield better capture of semantic content. The result is a declared vocabulary that provides quick and actionable insights for financial industry practitioners to stay on top of changes in their fast moving domain. This work presents a novel means for augmenting a financial markets ontology through the use of probabilistic topic modeling. In particular, this work has used the Latent Dirichlet Allocation (LDA) method for topic modeling which is able to ingest large volumes of text data and quickly extract themes across documents. LDA, which yields topic models, was applied to a large corpus of 11,693 text articles and speeches in the financial industry domain. The outcome is an informed ontology that demonstrates the evolution of concepts at deeper levels of granularity, and over distinct time periods such as changes in market policy under the Obama Administration as compared to the Trump Administration.

Last modified

• 12/17/2019

Creator

• Troy Danka

DOI

• https://doi.org/10.21985/n2-p5hk-fp63

Subject

• Data Science

Keyword

• knowledge graphs
• LDA
• ontologies
• topic modeling
• text analytics
• machine learning

Date created

• 2019

Resource type

• Masters Thesis

Rights statement

• In Copyright