Automating the process of assessing business entities by using new representation methods to assess the risk of bankruptcy and the reliability of financial statements
The subject of the proposed project is to develop methods of representation of information on companies (both financial and non-financial on companies) and solutions in the form of models:
- Ecovis Vector Space (development of methods for representation of multimodal data of sequential nature, where sequential data will be – financial information over time),
- Ecovis Classification (development of a method for classifying companies and financial reports),
- Ecovis Exploration (development of a method for the analysis of company data and its quantity and exploration using deep neural networks).
- Ecovis Nonfin (development of analysis methods using deep language models.
A tool will be built to rescan the general search results using representation based on deep language models. The final objective of the project is to create an innovative product – a tool combining the above solutions to provide a new advisory service (expertise): for entities other than audit firms – Ecovis as an external advisor providing due diligence and forensic audit services, including going concern assessment as well as identification of management override of internal controls, whose work in that field is used to assist the auditor in obtaining sufficient appropriate audit evidence, in particular as regards the identification of the risk of material misstatement.