Singapore, September 2015: The Rakuten Institute of Technology has announced the winners of the Rakuten-Viki Global TV Recommender Challenge at the grand presentation and award ceremony.
Masaya Mori, Global Head, Rakuten Institute of Technology, Executive Officer, Rakuten, Inc presented the winning title, along with the grand cash prize of S$8,000, to Team Merlion. The first runner-up and second runner-up titles were awarded to Team Haipt and Team GM respectively.
Earlier this quarter, the Rakuten Institute of Technology partnered with Viki, a global TV site, to launch the Rakuten-Viki Global TV Recommender Challenge. Hosted on Dextra’s challenge platform from the 22nd July to the closing date of 31st August this year, participants had access to over 7 million lines of rich, anonymised data to gain a sense of viewers’ preferences, popular content features, and TV fan demographics. The goal was to build a personalised recommender system for Viki fans world-wide, following a set of user and business considerations.
Six teams were shortlisted by Rakuten Institute of Technology and Dextra to present their models to a panel of judges – including Mr Masaya Mori, Mr Rohit Dewan, Chief Technology Officer at Viki, Ms Ewa Szymanska, Head of Research at Rakuten Institute of Technology Singapore, Mr Takashi Umeda, Data Scientist at Rakuten Institute of Technology Tokyo and Mr Adam Lyle, Chairman of Newton Circus – at the final presentation and award ceremony, held at the IDA Labs @ National Design Centre.
Unlike many other online data challenges, accuracy was only one of the criteria used to select the winners, participants were required to submit their documented codes and detailed reports. The judges took into consideration engineering qualities including ease of implementation and risk management, business and user insights including demonstrating depth of understanding of the data and Viki business and use of innovative approaches alongside further ideas for improvement.
Winning team Merlion employed an elegant algorithm that performs outstandingly in both public and private leaderboards, identified and engineered very useful features, as well as discovered and visualised some key insights from the data provided.