bitREC introduces personalized TV playlists

Wednesday, July 5th, 2017
bitREC logo

VILNIUS — bitREC introduces personalized TV playlist feature in its video content personalization platform. The feature aims to merge linear and on-demand viewing by creating a daily personalized playlist for each user from catch-up, VOD and other available on-demand content sources. The playlists are based on past viewership data and serves as a “personal TV channel” for the platform user.

“Personalized TV channel is a great usability hack. The items in the daily generated playlist are aligned by how likely the user is to watch them. They are presented in the same fashion as EPG from the traditional linear channel. After watching one item, the next item plays automatically, creating the illusion of linear viewing. However, the user may change the order of the items playing, switch to the next item like in the traditional on demand realm” – noted CEO of bitREC Romas Juskevicius.

The playlist is generated daily for each user and usually consists of 30-50 content items. They are presented in EPG fashion just like content from traditional linear channel. The highest priority is placed on catch-up content the user might have missed the previous days. It also populates the playlist with other VOD content based on content similarity and user viewing patterns.

“Pay TV platform operators have huge catch-up and on-demand libraries. Multiply the 24/7 recording by the 100+ number of channels, add a substantial VOD library, and you have a content discovery problem on your hands, where even by employing a recommendation engine, traditional recommendation presentation methods become problematic. Daily personalized playlists allow to present content for the user in the “daily TV schedule” fashion he easily understands. The autoplay feature allows this playlist to be watched “in the background” in the same way most linear viewing occurs. It is also a great way to increase on-demand viewing, where users stop watching after one content item” – noted bitREC CEO Romas Juskevicius.

The feature was created as a way to boost on-demand viewership, increase time on platform and simplify content discovery. It proved especially useful in the non-prime-time viewing hours.

“The users return home before the linear prime-time. In this timeslot they can tune-in into their personalized TV playlist for the most recent content they might have missed or other contextually relevant content, leave the TV “in the background” and go on with their usual routine. Coupled with additional features like live TV notifications, users may switch to live TV channels when something relevant comes up on linear TV channels” – noted Romas Juskevicius.

In addition to collaborative filtering and contextual similarity, bitREC platform employs Recurrent Neural Networks to improve the quality of recommendations. It puts the greatest emphasis on recent user behavior and viewership patterns, thus enabling to recommend more relevant content in a timely fashion. Utilization of Deep Learning algorithms allows bitREC to improve recommendation precision by at least 30 percent as compared to standard approach from other recsys providers.

The feature is currently being trialed by a several of European pay-TV operators. It will become part of bitREC content recommendation suite available to all company’s clients by the end of 2017.