XKOS for classification levels. The XKOS data vocabulary is an extension of SKOS, specifically geared towards meeting the needs of the statistical community for knowledge organization systems.46 The specific requirements of this community include increased specificity for both hierarchies (distinguishing partonymy from hyponymy) and associations (distinguishing causal, sequential, and temporal relations). Additionally, the statistical community recognises levels in their hierarchical structures. The XKOS specification states that “levels are used as a means to identify concepts within a classification [that are] used to classify instances at the same specificity” — similar to the purpose of category types in historical language thesauri.47 Although the purpose of the classification levels in XKOS matches that of category types in historical language thesauri, there is a difference between the two notions. This difference makes XKOS levels unfit for expressing category types. Levels in XKOS “correspond to all those concepts that are same distance from the top of the hierarchy”.48 All top categories belong to the first level, all categories directly subordinate belong to the second level, and so on. This approach is valuable for those thesauri in which categories at a given depth of the taxonomy all share the type of category. However, this condition does not always hold for historical language thesauri. As demonstrated in Chapter 1, TOE, HTE, and LSM, distinguish different category types. Some of their categories have subordinates of different category types. In other words, an equal depth in the tree for categories does not imply the same level in the hierarchy of category types. The three thesauri use these types in a manner more flexible than XKOS is able to express: conceptual levels, which reflect the hierarchy of category types, rather than tree levels. This distinction is portrayed in Figure
