BigDataGrapes D4.2 - Methods and Tools for Distributed Inference

The objective of this deliverable is to develop inference methods that support efficient information selection from heterogeneous data pools. There are many challenges in data reasoning and inference based on distributed data. The first one is addressing data security and access rights to both original data and inferred information. The second challenge is how the actual inference over distributed sources can be performed and implemented. We address the main principles applied to data inference and different types of inference – rule-based, query-based, model-based and fuzzy inference – and their application in BigDataGrapes project. The Final section is dedicated to state of the art with a standard theoretical approach to inference from descriptive logic standpoint, as well as related work in implementing those approaches.