concept cloud in category recommender system

appears as: cloud, The cloud
Practical Recommender Systems

This is an excerpt from Manning's book Practical Recommender Systems.

Think about the data as a cloud of points, and you want to project the data into a lower-dimensional space where the distance between the objects is the same. Then the points that were further from each other before the reduction are also further away after the reduction, and nearby items are also closer after the reduction. You want to reduce the data so that only the directions that provide more information are retained. If the algorithm succeeds, then the vectors pointing in those directions are said to be latent factors.

  • Data needs to be in the cloud to be usable. If you have an on-premise solution (meaning that you have your servers locally), then you’ll have to somehow move all your data into the cloud for your system to do the calculations necessary to create the recommendations. Bandwidth, storage, and so on can quickly turn an economical solution into something far too costly.