Users who are learning to read or learning about a topic by viewing content on a device can benefit from conversational activities, such as question-answer turns for the viewed content. This disclosure describes techniques to perform natural language assessments of content that is being consumed on a user device. A chatbot is implemented using suitable technology, such as a large language model. With user permission, the model is used to generate questions that evaluate the user’s understanding of the content viewed. User provided answers are evaluated and suitable responses are provided to the user. The techniques enable automated assessment and feedback. The described features for assessment via chatbot (or virtual assistant) can be built into any application. Assessment is performed on-device and in a confidential manner.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Wong, Alan; Lacey, Vincent; Gharpure, Chaitanya; Hao, Rebecca; Venkatraman, Priya; Elidan, Gal; Engelberg, Roee; Hackmon, Lidan; Rabin, Roni; Fink, Michael; Yu, Paxon; Yang, Tiffany; Dikker, Vitaliy; and Levin, Adva, "Reading Comprehension Assessment Using LLM-based Chatbot", Technical Disclosure Commons, (June 30, 2023)