(130b) Text Analytics Applications for Material Manufacturing

Global manufacturing companies generate vast amounts of unstructured data from within their operations and enterprise resource planning (ERP) systems. These text-based data sources can be difficult to integrate into traditional reporting and analysis workflows that require querying and numerical calculations. With the advancement and ease of entry into developing text-based models and features, Dow is employing several types of text modeling methods to improve margins, reliability, safety, and our customer’s experience.

Dow has developed sustainable event recording processes that provide free-form text fields to precisely record the events and actions taken, but are difficult to then aggregate into reports or understand trends. In this talk, we will share examples of how Dow has developed machine learning models to categorize text for efficient trend analysis, as well use deep learning to relate documents to one another. We will then show how leveraging these text analytics methods can deliver value through new insights and streamlined work processes.