This video contains proprietary information and cannot be shared publicly at this time.
Figure 1
Team 20
Team Members |
Faculty Advisor |
WingSheng Sze |
Dongjin Song Sponsor Other |
sponsored by
Boehringer Ingelheim Natural Language Processing Software
A web application integrated with a BERT-based natural language processing model fine-tuned for question and answering. The natural language processing model answers questions to various pharmaceutical focused research questions. The application's user-interface is stream-lined to provide ease of usage for users that are experienced and inexperienced with the system. The design of the user interface ensures a lack of experience is not detrimental on a user's effectiveness with the system. The application aims to ensure ease of use by providing relevant and probable results while requiring minimal input from the user. The application utilizes existing data for question and answering to fine-tune the performance of the model to account for various phrasing and speech for how to analyze a question. To collect data and find ways to improve the quality of query results, the application also includes a user feedback form. The feedback form attempts to gauge user opinions on the relevance and quality of results along with the number of responses from the model that correct out of all results that are returned by the model.