Data preparation

Before using some of the link prediction / recommendation algorithms and metrics, it is sometimes necessary to prepare some data. For instance, content-based approaches take as input inverted indexes containing the contents generated by each user in the network, and supervised methods need a suitable set of feature vectors for each target-candidate user pairs.

In this section, we provide information about how we can generate these data.

Content index

• Content index generator

Feature vectors

• Feature vector generation
• Feature vector samplers