Likaholix

Likaholixに関する記事を読んだので、記述的な背景をhttp://likaholix.com/aboutで確認してみた
引用：
To help users create likes, we built a database of various items across the web including books, movies, tv shows and products. For each of these items, we have a list of landing pages which describe the item, various significant topics about the item along with images and videos of the item. We built this database by crawling the web and using data from Freebase and Wikipedia. To supplement this database, we also use real time Google web, image and video search results when users create likes.

We use Python for our crawl, data analysis and some backend servers. Our recommendation system is built using Solr, and our website is developed using PHP and JQuery backed by a MySQL database. We use Facebook Thrift for communication between servers and serializing data. We run our crawl and data processing on Amazon EC2 and S3 and use Cloudfront for media serving. The website is hosted on dedicated Linux servers at M5 Hosting.