Preference learning with clustering
US10636073B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Apr 15, 2016 |
| Grant date | Apr 28, 2020 |
| Priority date | — |
| Expiry date | Nov 5, 2036 |
Classification
- Technology area (CPC H)Electricity
- CPC primaryH04N21/25891
- WIPO fieldAudio-visual technology
- WIPO sectorElectrical engineering
Abstract
In one aspect, there is provided a method for preference learning. The method may include: determining whether existing data associated with a user is available, wherein the existing data includes preference weights assigned to user clusters; when existing data associated with the user is unavailable: selecting, based on a ranking of the user clusters, a user cluster; providing, based on a selected user cluster, recommendations to the user; and assigning, based on a response of the user to the recommendations, a preference weight to each of the selected user cluster and/or another user cluster; and when existing data associated with the user is available, determining whether to exploit the existing data or to explore new preferences for the user, wherein exploiting comprises providing the recommendations based on the existing data, and wherein exploring comprises providing the recommendations by selecting a user cluster not assigned a preference weight.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.