The NLP-TIPS project exploits advances in natural language processing (NLP) to detect the rich ideas that students develop during everyday experience. Using the ideas detected in student responses, NLP-TIPS designs adaptive, interactive guidance that prompts students to reconsider their own ideas and pursue deeper understanding.
This work continues a successful partnership between the University of California, Berkeley and ETS, as well as science teachers and paraprofessionals from a variety of middle schools enrolling students from diverse racial, ethnic, and linguistic groups whose cultural experiences may be neglected in science instruction.
The partnership leverages the powerful, open-source Web-based Inquiry Science Environment (WISE) to implement adaptive guidance designs endorsed by teachers that feature student-component dialog and peer interaction.
NLP-TIPS guidance designs encourage students to distinguish among their ideas rather than accumulating fragmented ideas, as often happens in science.
Further, we are developing Teacher Tools that can detect when an adaptive guidance interaction is stagnating and alert the teacher. Teachers’ responses to these alerts help shape the NLP technology to detect ideas that are valued but may be infrequent.