白丝美女被狂躁免费视频网站,500av导航大全精品,yw.193.cnc爆乳尤物未满,97se亚洲综合色区,аⅴ天堂中文在线网官网

Personalized content recommendations based on consumption periodicity

專利號
US10897651B2
公開日期
2021-01-19
申請人
Comcast Cable Communications, LLC(US PA Philadelphia)
發(fā)明人
Hans Sayyadi
IPC分類
H04N21/466; H04N21/458; H04N21/442; G06N5/02; G06N7/00
技術(shù)領(lǐng)域
content,similarity,may,score,consumed,timediff,video,be,hours,candidate
地域: PA PA Philadelphia

摘要

Aspects described herein describe providing content recommendations such as, for example, recommendations for video content. A content recommendation may be based on when content was previously consumed.

說明書

The proposed approach for making content recommendations is described in further detail below with respect to video content including video content scheduled for broadcast at a particular broadcast time, on-demand video content, and digitally recorded video content. With the benefit of this disclosure, it will be appreciated that the approach to providing video content recommendations described below may be adapted for providing recommendations of other types of content. Various techniques may be selectively employed for making content recommendations such as, for example, video content recommendations. In some example implementations, content recommendations may be based on a determined similarity between candidate content and previously consumed content. In these example implementations, a similarity score for a pairing of the candidate content and previously viewed content is obtained, and the recommendation score for the candidate content is based, at least in part, on that similarity score. In other example implementations, the recommendation score may be obtained for a content menu rather than specific content. Content menus may correspond to content format (e.g., audio, video, text, etc.), content type (e.g., movie, television series, sports, etc.), content genre (e.g., drama, comedy, horror, etc.), and so forth. In these other example implementations, the recommendation score may be obtained for a content menu. In each example, the recommendation scores are further based on one or more contribution factors obtained for previously viewed content. The contribution factor is adjusted based on a time difference between a consumption time of the previously viewed content and a reference time. Generally stated, the techniques for adjusting the contribution factor of previously consumed content described herein may be employed in various techniques to making recommendations based on previously consumed content.

權(quán)利要求

1
微信群二維碼
意見反饋