How Related is Author Topical Similarity to Other Author Relatedness Measures?

Using a dataset of 26,228 Psychology document surrogates from Elsevier databases, we compare author relatedness measure outcomes for 125 authors based on topic modelling to more traditional approaches that rely on direct citation, co-citation and collaboration. Outcomes for the author topical similarity measure are compared to existing co-authorships in the dataset using UCINET/NetDraw. We demonstrate how author topical similarity outcomes provide a similar, but more complete, picture of author relationships than the co-authorship network. Nonparametric correlation analysis results of author topical similarity, co-authorship, citation, and co-citation were also compared for thirty author pairs of differing author topical similarity values. There is a significant correlation between author topical similarity and co-authorship and direct citation-based measures for high similarity author pairs, but not with co-citation measures. The author topical similarity measure, therefore, may serve as a reasonable predictor of collaboration or direct citation for authors with high topical similarity. The measure may also identify potential collaborators based on high author pair similarity values, where there is a lack of existing collaboration, and serve as a complement to author relatedness based on co-citation analysis. Conference Topic Methods and techniques