The Record Issue 13: Summer 2019

95  MED I A & ENT E R TA I NMENT “Artificial intelligence (AI) and machine learning is rapidly maturing. Over the past few years we have seen a phenomenal growth in areas where these technologies can replace, or even surpass, humans. The applications for media firms have so far been centred around mitigating labour- intensive mechanical tasks like identifying faces or turning spoken words in to close captions and translations. The next phase, which is already emerging, is where AI and machine learning take on creative roles, complex tasks like analysing patterns and recognising behaviours to act as camera man or technical director. The automation that this enables can be a huge leap in democratising media production. With AI and machine learning enabling content creation, where labour costs have historically been prohibitive, professional level production capabilities will be put in the hands of anyone.” “With social media networks encouraging everyone to share their own content, the number of reporters has increased tremendously as lay people become part of the information and entertainment chain. In this scenario, Audience News Gathering (ANG) for editors and reporters minimises the effort of finding the right person from the audience sharing outstanding content and go live with them during the show. The ANG helps bloggers gain more visibility, editors to easily reach out to excellent contributors, have better fact checking and the TV audience increases by interacting with their favourite shows. Available on the Microsoft Azure cloud, it uses Skype to easily call contributors and go live during an event.” We use AI to deliver deep personalisation in three steps: “First, we use computer vision, deep learning and information retrieval techniques to identify and analyse different pieces of information about people, objects, emotions and other contexts in videos and tag with scene level contextual metadata. Second, we have created a user experience layer (as pictured) that gives viewers real time interactive access to every little detail in the scene. Third, we combine the output of the above two – content metadata and engagement data – and use a combination of sophisticated neural network models, heuristic-driven models and content models to deliver deep personalisation. That could be personalisation of content, advertisements or commercial offers for the audience.” Pontus Eklöf Chief operating officer, LiveArena Sharique Husain CEO and co-founder, Dive Marco Montenovo CMS-DVS worldwide solution architect, HPE