Subject indexing for author name disambiguation: Opportunities and challenges

Abstract

Author name disambiguation is becoming increasingly important due to the prevalent availability of publications in digital libraries. Various approaches for author name disambiguation are available, utilising a variety of information, e.g., author name, affiliation, title, journal and conference name or venue, citation, co-author, and topic information (Ferreira et al., SIGMOD Rec 41(2), 2012). Topics can be obtained, e.g., using subject information captured in various controlled vocabularies, classifications and mappings between them used to index publications (Torvik et al., J Am Soc Inf Sci Technol 56(2):140–158, 2005). Research interests of authors, evident in topics, might change over time though (Ferreira et al., SIGMOD Rec 41(2), 2012), and thus limit their usefulness for author name disambiguation. Here we present a longitudinal analysis of topics with respect to their suitability for author name disambiguation. We analyse the distribution of subject headings and classification notations taken from the Thesaurus (TSS) and the Classification for the Social Sciences (CSS) (http://www.gesis.org/en/services/research/thesauri-undklassifikationen/) for research projects and literature (available in sowiport—http:// sowiport.gesis.org maintained by GESIS) and the changes in distribution over time. To assess the suitability of subject information for author name disambiguationmore closely, we then analyse the changes in the annotation over time for a selection of authors and author groups at different stages in their career, also taking into account the hierarchical organisation of the applied controlled vocabularies.