Abstract
Techniques are described for targeted exploration in content recommendation using per-topic user knowledge state and curiosity-zone detection. For each user-topic pair, a system maintains an engagement-derived knowledge depth profile including exposure count and a depth distribution across difficulty levels. A curiosity intensity is computed as an inverted-U function of exposure count with parameters learned per user or cohort from engagement-versus-exposure data, and a curiosity state is classified based on exposure and an engagement trend. Candidate items are scored for exploration using the topic curiosity intensity, a depth-gap fit to a target information gap of approximately one difficulty level above the user’s current depth, and a content quality score. A per-user exploration rate is set as a function of a count of topics in a curious state and is used to blend exploration scoring with base relevance ranking. Exploration outcomes update profiles and may trigger difficulty adjustment or topic suppression.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Anonymous, "Systems and Methods for Curiosity-Zone Detection and Information-Gap-Driven Exploration in Content Recommendation", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10754