Big Data & Recommendation Systems
Big Data improves a company’s response to people’s needs. How do Google, Amazon or Facebook know our personal interests with amazing accuracy? This is no coincidence. There are complex algorithms behind such systems. These algorithms, similar to the Coca-Cola formula, are important trade secrets.
Travel Apps
Today a travel app provides much more information to its users than their desired destinations. The app also offers information related to the users’ interests and preferences related to sightseeing, food, accommodation etc. These interests are derived through the their smartphone activity, including the use of Bookmarks.
Big Data provides two ways to make recommendations in such apps. One is the classic (graph theory) approach of online shops: users who purchased the product A also bought product B; or travellers who were on site A were also in place B. While this approach correlates products and locations to each other, it provides little information about the users themselves.
The second approach is spectral or principal component analysis. Here specific behaviour patterns of all users are extracted. This approach is also referred to as feature extraction or dimension reduction.