Dataset for learning to predict the suitability of experts for statements

Dataset consisting of 1,700 (statement, expert) pairs for learning and evaluating whether a researcher can be cited as an expert for a statement.<br> A total of 170 statements were extracted from the dataset "De Argumentenfabriek" [1].<br> For each statement, two experts were manually selected from Google-Scholar and their data extracted.<br> In addition, for each of the 170 statements, 8 experts were randomly picked from the pool of 340 scientists and then manually assessed on a scale from 0 to 2 by two to three annotators whether that scientist can be cited as an expert for the statement. [1]: https://zenodo.org/record/4813727<br>