Ordered Trees: A Structure for the Mental Representation of Information

Abstract

This paper reviews the ordered tree technique for deriving a structured mental representation, and discusses potential applications for infonnation science. The ordered tree clustering algorithm that we are developing provides an alternative to more traditional approaches of clustering based on similarity matrices. This algorithm is based on patterns of linear strings, typically free-recall orders, but potentially any material organized in a linear fashion. The technique has proven useful in representing ordered classifications, such as an alphabet or a numerical sequence, imbedded within a hierarchical semantic structure. Furthermore, the ordered tree technique, its corresponding data structures, and the data analysis procedures, can be generalized to include cross-classification used by human subjects and the classification of subjects themselves into categories such as experts and novices. discusses potential applications for infonnation science. The ordered tree clustering algorithm that we are developing provides an alternative to more traditional approaches of clustering based on similarity matrices. This algorithm is based on patterns of linear strings, typically free-recall orders, but potentially any material organized in a linear fashion. The technique has proven useful in representing ordered classifications, such as an alphabet or a numerical sequence, imbedded within a hierarchical semantic structure. Furthermore, the ordered tree technique, its corresponding data structures, and the data analysis procedures, can be generalized to include cross-classification used by human subjects and the classification of subjects themselves into categories such as experts and novices.