The Improvement of TV show and film Recommendation through Artificial Intelligence and Machine Learning

点击翻译按钮获取中文摘要

"The way in which we watch TV and Movie content is changing. Over the air, scheduled programmes are being replaced by high quality Video on Demand (VOD) services from providers such as Netflix, Amazon, Sky, Now TV and many more globally.Due to this shift from scheduled program guides, we are now expected to know what we would like to watch or be presented with a VoD providers list of available content driven by factors outside of our control.There are two main obstacles with this new approach to content offerings:1. If we know what we want to watch, it is difficult to know which provider, if any, has that content available. We have to login and search and if its not available (more often than not) move onto our next provider.2. If we don't know what we want to watch we are presented with an overwhelming set of available content, some of which we may have already seen and some irrelevant.This leads to a frustration that is common amongst viewers and there is a shift to recommendations from friends / family via social media.https://www.ericsson.com/en/networked-society/trends-and-insights/consumerlab/consumer-insights/reports/tv-and-media-2016, https://cstonline.net/why-cant-i-find-anything-to-watch-on-tv-by-john-ellis/With more reported VoD suppliers on the horizon (Google - techradar.com/news and Apple - [macworld.co.uk/news][0]) this problem will only worsen.Our vision is to develop an app that at its core has an Artificially Intelligent recommendation engine that with minimal user interaction will recommend relevant content for viewers based on personal viewing habits.By utilising whole of market data sources ([TheMovieDB.org][1], IMDB.com) as a central database, mapped to VoD providers content availability we can provide a solution to problem 1\. By creating a recommendation engine and smart interactive app that learns via interaction (seen, like, don't like, not interested) and Machine Learning on recommendations we solve problem 2\. Our aim is to further utilise pre-defined lists from both influencers (critics, celebrities etc) and popular current affairs sources to further entice interaction and provide more intelligent AI data.[0]: http://macworld.co.uk/news[1]: http://TheMovieDB.org"