You are invited to a OnPoint presentation by Dr. Russ Rhinehart on "Linguistic Modeling; an Application to Student Success Prediction.”
When: May 4, 2023 11:30 AM Central Time (US and Canada).
Hosted by the Education and Research Division (ERD).
To register in advance for this meeting, click on the link: http://us02web.zoom.us/meeting/register/tZEvce2qpz8tH9Pkz8Ny5EDbJSXtzalaL0bu#/registration
After registering, you will receive a confirmation email containing information about how to join the meeting.
Linguistic modeling uses human language to express cause-and-effect relations. The structure for the rules is “IF [antecedent} THEN [consequent]”. An example would be “IF the student does poorly in STEM courses as a Freshman and Sophomore, THEN they will do poorly in the upper-level ChE courses.” Another example would be “IF the raw material comes from supply A, THEN dilution Is needed to maintain product quality.” The human concept of the mechanism may be correct, wrong, or partially right. The objective of linguistic modeling is to express such understandable, natural language rules in a manner that they can be quantified, and that rule errors or deficiencies can be detected and corrected. Once rules are validated, you can have a high confidence in action based on the rule.
Linguistic models are simple and understandable by human decision makers. Linguistic models are explainable. These aspects are substantial contrast to neural network models, and much of the tools of Big Data and Artificial Intelligence, in which the models are complicated, and mechanistically unexplainable.
The applications for validating linguistic models are vast.
In this presentation, the model for predicting upper-level student performance based on lower-level performance will be developed and tested, as a case study to reveal how to use data to validate and adjust the human rule statements.