Disclosed herein is an improved mechanism for filtering videos based on protected topics. The mechanism can begin by identifying a particular portion of search queries for videos (e.g., the top X percent of search queries for videos). The mechanism can then determine a group of topics to be protected based on the identified search queries. The mechanism can, for each protected topic in the group of topics, identify a subset of channels that provide videos that are associated with the protected topic (e.g., the top N channels that provide videos associated with the protected topic). The mechanism can, for each channel in the identified subset of channels, calculate a channel trust score (e.g., based on channel ranking among channel results for search queries for a topic and based on a number of policy violations associated with the channel). The mechanism can then identify channels associated with a channel trust score greater than a channel trust score threshold. Videos from these channels having a channel trust score greater than a channel trust score threshold can be provided for recommendation (e.g., in response to search queries). For example, a video content sharing service can provide videos associated with a trusted topic to be obtained from one or more trusted channels.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.