This project is a demonstrator application for a video recommendation engine that uses publically available video and user data to show what videos are recommended for a user based on certain characterstics.
Problem Description: The video content management product space is relatively mature in academia and nascent in the enterprise. One of the common threads across all institutions that use video, is the need to effectively disseminate knowledge with little friction. While products that actively share this knowledge are present and in use, Youtube or Netflix style discovery is still not in use. The key difference in the approach for passive discovery between consumer video platforms and non-consumer platforms like Panopto or Kaltura is in the intent. Youtube and Netflix aim to maximize video consumption whereas a VCMS’s goal is to maximize the knowledge and skills acquired by users using video. With that motivation in mind, building a recommendation engine that would help with a VCMS’s product goals will involve understanding specific scenarios of users in academia and enterprise. A VCMS user extracts value from it by obtaining knowledge through videos. Being able to discover existing videos within their institution’s video library allows for knowledge already captured to spread across the organization. Without intelligent discoverability of content, the pathways for users to watch videos are from other users directly sending it to them or by searching for specific terms that match with the video metadata. With the ability to easily discover existing videos that are relevant, users are not able to extract the full value of videos already created by their organization.
Purpose and Goals: The purpose of building a video recommendation engine demonstrator for a VCMS is to achieve the following, if the system is implemented within a VCMS:
- Improve knowledge sharing across an organization through intelligent discovery of video content. Measurable outcome: Higher overall engagement actions per user on videos recommended by the intelligent engine across an institution’s VCMS as compared to non-recommended videos
- Better user experience and value obtained by users who are recommended videos by the intelligent engine as compared to without recommendations. Measurable outcome: Higher net promoter score and better scores on user surveys on VCMS customers with the recommendation engine in place.
This demonstrator project can be used by VCMS companies to plug in their private video and user data to evaluate the user experience amongst stakeholders and also in usability studies.
The project was created directly from publically available data and source code. The references to the data and source code can be found here: