Inference actions. OpenAIREplus original entities can be subject to the following data inference actions, performed by data mining and inference services and by humans: • Merging of entities; • Adding new inferred entities (relationship between main entities); • Updating of properties of an original entity; • Removing original entities. Such actions may be caused by different data inference workflows, which may or may not involve humans. Examples in the first OpenAIRE service settings are: • End-user feedbacks, when approved by data curators; • Data curator edit activities; • Automatic inference algorithms: (i) similarity relationships between result entities, (ii) information extraction from full-text of publications, which may infer title, authors, organizations, and emails of publication entities, (iii) citation management, which may lead to the introduction of publication results not available to the information space, (iv) subject classification, which may lead to the introduction of new entities, etc. • End-user claim/vaidation actions through the OpenAIRE portal, e.g., project coordinators validating publication-project relationships. • Deduplication services identifying a set of entities of the same type which all correspond to the one and the same real-world entity. In the following we shall focus on how to modify the OpenAIRE data model in order to represent the outcome of data inference actions on the information space and to resolve data inference conflicts that may arise between original and inferred entities. For each of these aspects, we shall provide and explanation of the problem, an extension of the data model to handle the problem, and, where necessary, a solution to the problem using the updated model.
