From zero to enterprise data hub: Building content recommendation system using open source tools (Spark, Hadoop, R, SciPy)

International Conference on Big Data Analysis and Data Mining

May 04-05, 2015 Kentucky, USA

Radosław Kita

Scientific Tracks Abstracts: J Comput Sci Syst Biol

Abstract :

TVN Group is the biggest Polish media and entertainment group (10 million unique users, 200 million page views and almost 100 million stream views per month). For several years, there is a clear trend in the transmission of advertising to the Internet. This also applies to television -the former leader in advertising revenue. TVN faced with the need to build a solution that enables the delivery of video content for Internet users through both the classic web pages or mobile applications or Connected TV. We had to build in the short time an internal audience reporting system and video streams recommendation system. We decided to rely on open source solutions. We use Spark to pre-data analysis and reporting what is happening in the last minutes; Hadoop as a basic data warehouse. We conduct more advanced analysis using R and Python (SciPy). I would like to tell you that solutions come true in our case, what problems we encountered and how they were able to solve them.

Biography :

Radosław Kita for over 12 years was responsible for internal reporting systems at Onet.pl ?the biggest Polish internet portal (3 billion page views and 75 million cookies per month). Afterwards he was employed by Alior Bank - one of the leading banks in the country ? on project of creating 360-degree customer view system based on internal bank, mobile operators and internet portals data resources. He is currently responsible for the implementation of the recommender system at TVN Group. From beginning 2011 he actively uses Hadoop. He promotes the use of open source tools in the course of numerous national conferences.