Anastasilia Noguier
Research Mentor(s): Stefan Larson
Research Mentor School/College/Department: DryvIQ
Presentation Date: 08/03/2022
Presentation Type: Poster
Poster Number: 47
Session: Session I: 12:30 – 1:20pm
Room: League Ballroom
Authors: Anastasiia Noguier, Stefan Larson
Abstract
Unstructured Data is common in today’s business landscape, and it often contains sensitive Personally Identifiable Information or PII. To protect personal information corporations and businesses in America must comply with the PII regulations. In this work, we explore Machine Learning based Natural Language Processing (NLP) models to test how well the NLP model can perform on non-English documents. Non-English documents can potentially be a source of unprotected, not easily recognized data, that may jeopardize a social right to remain forgotten and can be a foul of PII regulations. We are creating a dataset of Russian documents available on the Internet that includes the categories like resumes, press releases, agreements, floorplans, syllabi, etc. The dataset contains over 1200 documents classified into 25 categories. The documents will be tested on the text-based ML model to understand if the model can handle the non-English documents with the same accuracy as English ones. This experiment will help us understand whether we need to create a custom Russian-language ML model or we can rely on a simple translation of the documents. At this time, we do not have complete results of these experiments, but we hypothesize that the model will perform well, yielding an efficient outcome, and will be a step towards training the DryvIq ML model for many more languages. This project will be useful to many Artificial Intelligence and Data Governing companies, as it will significantly improve the precision of the Machine Learning algorithms for other non-English projects.The main benefit of this research is efficiently protecting data for enterprises in languages other than English. This research will benefit many individuals and corporations in the world of personal data protection.
