Fig. 1
From: Semi-automated title-abstract screening using natural language processing and machine learning
Workflow of presented methodology. First, the title and abstracts are successively screened by two human reviewers who decide if a citation should be included or excluded. Next, the titles and abstracts are processed by natural language processing (NLP) techniques to receive a numeric dataset, that is subsequently used to train machine learning (ML) models with respect to binary classification of in- or exclusion