A New Similarity Measure for Subject Hierarchical Structures
Journal of Documentation
Subject hierarchical structure similarity measure, Subject hierarchical structure comparison method, Tree similarity
Library and Information Science
– The purpose of this paper is to introduce a new similarity method to gauge the differences between two subject hierarchical structures.
Design/methodology/approach– In the proposed similarity measure, nodes on two hierarchical structures are projected onto a two-dimensional space, respectively, and both structural similarity and subject similarity of nodes are considered in the similarity between the two hierarchical structures. The extent to which the structural similarity impacts on the similarity can be controlled by adjusting a parameter. An experiment was conducted to evaluate soundness of the measure. Eight experts whose research interests were information retrieval and information organization participated in the study. Results from the new measure were compared with results from the experts.
– The evaluation shows strong correlations between the results from the new method and the results from the experts. It suggests that the similarity method achieved satisfactory results.
– Hierarchical structures that are found in subject directories, taxonomies, classification systems, and other classificatory structures play an extremely important role in information organization and information representation. Measuring the similarity between two subject hierarchical structures allows an accurate overarching understanding of the degree to which the two hierarchical structures are similar.
– Both structural similarity and subject similarity of nodes were considered in the proposed similarity method, and the extent to which the structural similarity impacts on the similarity can be adjusted. In addition, a new evaluation method for a hierarchical structure similarity was presented.
Zhang, Jin and Lei Zeng, Marcia (2014). A New Similarity Measure for Subject Hierarchical Structures. Journal of Documentation 70(3), 364-391. doi: http://dx.doi.org/10.1108/JD-12-2012-0160 Retrieved from http://digitalcommons.kent.edu/slispubs/18