Using Text Mining Techniques to Analyze Notes in Financial Statements

Firms meet legally established accounting rules by including notes in financial statements that essentially hide key information in plain sight. That is, they contain information necessary to understand the statements, but due to the volume of notes and arcane language may be uninterpretable by even lawyers. However, these notes may be interpretable by a software application. For example, the notes in the Enron case were found to be decipherable by a human being, but only after substantial effort. Ultimately, we wish to determine where the surface meaning of the financial statements misrepresents the reality of the organization where this can be discovered or at least suggested by the notes. As well, we aim to gather information about what notes are present, consider variations in wording, and support multiple languages. This project will benefit CaseWare by providing the company with new capabilities to analyze notes in financial statements.

Faculty Supervisor:

Frank Rudzicz

Student:

Abraham Escalante

Partner:

CaseWare International

Discipline:

Computer science

Sector:

Information and communications technologies

University:

University of Toronto