Case Study: AI-Powered Recommendation Platform for DirectTV

Industry: Media & Entertainment

The Client

An American direct broadcast satellite service provider and broadcaster transmitting digital satellite television and audio to households in the United States, Latin America and the Caribbean.

The Business’s Needs

  • The customer wants to develop a recommendation platform that can serve multiple platforms including STB, mobile devices, tablets and web browsers through a cloud-based solution
  • The platform will recommend users based on specific scenarios and users’ and friends’ activities such as Top Pick, Watched, Friends Watched, Liked, Friends Liked and Trending.

The Solutions

  • We applied onsite/offshore model and Agile methodology and divided into sub teams: Data Ingestion, UI, and Analytic.
  • Data Ingestion & Analytic:
  • Finished the POC phase with clustering and analytic to evaluate recommendation result.
  • Collected and processed ~ 1TB data.
  • UI built on Liferay to show recommendation and visualization results.

The Benefits

  • The FPT Software team delivered a high quality product at a reasonable cost.
  • FPT provided the best developers in Java and Big Data as well as analysts and researchers.
  • The customer has trusted FPT to suggest, discuss and decide the design and development approaches on our own.

Technology Details

  • Data Ingestion (DI): Load, clean, merge, stitch and ingest the raw data from external sources (JSON, CSV, XML) to generate the suitable formats on HBASE or HDFS
  • Data Submission: Get data from DI to generate the JSON format and post to an external system to get the recommendation results.
  • Data Analytics (DA): Get data from DI to calculate some values such as last actions, the hottest, etc. and cluster them also. The output will be stored in HBASE.
  • User Data Service (UDS): Read the content of DA in HBASE to expose web services.
  • Recommendation Service (RS): Get data from UDS and filter them. Gets the results from the external recommendation modules and PGES and send them to UI to display.
  • User Interface (UI): A Liferay portal to display recommended movies, programs, shows for users with different rules and criteria.

Recommendation Platform for DirectTV

We are digitalizing the Media & Entertainment industry. Checking out our Service offering for Media & Entertainment here.