Dimensions: Building Context for Search and Evaluation

Dimensions is a new database that focussed on a different set of problems to other scholarly search systems. Specifically, by the including not only data about publications and their natural associated citation graph, but by also including awarded grant data, patent data and clinical data and altmetric attention data, Dimensions holds a representation of an n-partite graph associated with the heterogeneous research objects. These objects have been treated as heterogeneous, but need to be brought onto the same footing (and homogenised) in order that they should make sense to a user. The links (or edges) in the expanded network of objects is created through extensive use of text mining and machine learning to extract and to normalise data. This article gives an overview of the techniques used to create the Dimensions dataset.