Relational machine learning author disambiguation

Author disambiguation is an open issue in the world of academic digital libraries. As many problems arise when trying to identify if two different signatures are from the same author and than group them, this issue has become more relevant inside the scientific community. This paper illustrates a workflow that aims to solve this issue. By using the best of a relational database engine and data mining techniques implemented in R, we have implemented a workflow that correctly disambiguates different instances of an author's name present in academic publications retrieved from the Internet. To evaluate he results we perform a two=step-validation process inside the workflow, validating if two articles were written by the same author, and if so, validating the authors grouped together as unique disambiguated author. With the validations performed, the workflow implemented allows the process of identifying and disambiguating any new author.