Recommendation engine company Gravity R&D from Hungary started off back in 2006 when the key R&D team behind it participated in a movie recommendation contest for Netflix, improving results by 10 percent. In monetary terms it meant $80-$90 million in additional revenue for Netflix.
After the six-year journey, it has finally broken into the TV space and today Gravity R&D announced that its client, Canadian telecom provider SaskTel, will introduce Gravity R&D ñ powered personalized recommendations to the residential subscribers of its IPTV service Maxô Entertainment Services.
Personalized TV recommendations are slowly being incorporated by the TV providers and SaskTel is apparently one of the pioneers. Earlier this year, U.S. TV behemoth Comcast was testing its digital video service X1, which will offer personalized TV recommendations. Its roll-out is planned for 2012.
The problem Gravity R&D and some other recommendation startups are trying to solve is the information overload. In the U.S. alone, 200 billion hours are spent watching TV on an annual basis, according to Clay Shirkyís book†Cognitive Surplus (hat tip to Daniel Nye Griffiths). The way Gravity R&D is solving the problem is by analyzing subscribersí viewing habits, which are recorded and stored by the IPTV provider.
Other companies are working on the problem of solving the live TV and video on demand recommendations as well, and they tend to approach the problem differently.
Filmaster, from Poland, allows users to rate a list of suggested films with simple ìlikedî and ìdid not likeî answers. These answers are combined with other user ratings, editorial curation of its movie and TV series database, and so-called ìmoodsî classification, and Filmaster then provides personalized movie recommendations. When I tested their beta version a few months back, their recommendations were spot-on, with the minor exception of a Polish-language sci-fi film from the 80s.
According to the founder and CEO of Filmaster, Borys Musielak, the Gravity R&D recommendation approach has a problem with a so-called ìcold startî, an inability to make recommendations immediately. Filmaster solves this problem, having reviewed and classified its existing movie database to quickly offer recommendations to its new users after they rated a few films. At present, Filmaster is testing a white-label solution with one of the major European cinema chains. At SXSW, the startup backed by HackFwd also announced its Google TV app to offer recommendations on Google TV, which is yet to be released.
Other companies are relying on social graph recommendations. For example, German startup Tweek, which launched its iPad app at London Web Summit this week, is trying to solve the TV content selection problem by aggregating ìlikesî and analyzing sentiments of a social graph related to particular video content.
In the other camp is Synopsi.tv, a Slovakian startup still in beta, which does not buy into the idea of social recommendations. When I spoke to company founder and CEO Rastislav Turek a few months back in Bratislava, he was pretty sure that selecting a movie has nothing to do what other people think of it. Given how often couples disagree about which movie to watch, Turek has made a decision to make movie recommendation based solely on the individualís own taste, which can even change over time.
The challenge Gravity R&D might face is telling which family member is currently in front of the screen. If the whole family shares the TV, it is likely that the viewing preferences of the family members will confuse the algorithms. At least in our family these preferences are very different.
To deal with the problem Gravity R&D pays attention to the time of the day the programs are viewed. Alternatively users can log in to the TV to watch personally relevant content. Built-in face recognition will solve this problem eventually.
This post is written by our regular contributor Natasha Starkell, the CEO of GoalEurope, the outsourcing advisory firm and a publication about outsourcing, innovation and startups in Central and Eastern Europe. Twitter @NatashaStarkell. Gplus.to/natashastarkell.
