An Update for Taxonomy Designers

Abstract

Taxonomies are classification systems that help researchers conceptualize phenomena based on their dimensions and characteristics. To address the problem of ‘ad-hoc’ taxonomy building, Nickerson et al. (2013) proposed a rigorous taxonomy development method for information systems researchers. Eight years on, however, the status quo of taxonomy research shows that the application of this method lacks consistency and transparency and that further guidance on taxonomy evaluation is needed. To fill these gaps, this study (1) advances existing methodological guidance and (2) extends this guidance with regards to taxonomy evaluation. Informed by insights gained from an analysis of 164 taxonomy articles published in information systems outlets, this study presents an extended taxonomy design process together with 26 operational taxonomy design recommendations. Representing an update for taxonomy designers, it contributes to the prescriptive knowledge on taxonomy design and seeks to augment both rigorous taxonomy building and evaluation.

A Live Extensible Ontology of Quality Factors for Textual Requirements

“…We translate the rigorous taxonomy development guideline proposed by Nickerson et al [8] and extended by Kundisch et al [18] to the larger scale of our ontology. None of aspects in which our ontology design extends the design of a taxonomy contradicts the abstract guidelines, since (1) the ontology simply consists of four individual taxonomies and (2) the extraction guideline is applicable to the additionally included index and scope note attribute as well.…”

“…The three first authors distributed the set of 105 eligible studies among each other and split the resulting subset into four iterations. During each iteration, they extracted all relevant objects based on an extraction guideline, which was initiated during the first iteration and maintained according to ontology development protocol [8], [18]. Publications had to at least contain one eligible quality factor based on the definition established in Sec.…”

“…a) Transparent Ontology Design Process: A major challenge in developing any subject-based classification is the lack of transparency of the process [18], where the process obscures the rationale behind design decisions. Since our ontology is both meant to facilitate collaboration and a communitydriven maintenance and evolution, we mitigate this threat by disclosing all process documentations 1 .…”

Decision support for efficient XAI services - A morphological analysis, business model archetypes, and a decision tree

“…view_ only= 2a5e1 9822e b34b6 18a1e e2199 36576 a7). According to Kundisch et al (2021), cluster analysis can be conducted as an evaluation of the MBox with the goal of "better describing, identifying, classifying, analyzing, and clustering objects that represent a certain phenomenon compared to doing so without a taxonomy or other classification schemes" (p. 9). Similar XAI services with similar classified characteristics according to our MBox are grouped into one cluster (Kaufman & Rousseeuw, 1990).…”

“…Cluster analysis (Kaufman & Rousseeuw, 1990) via RStudio. Evaluation (Kundisch et al, 2021) Decision tree development (Pedregosa et al, 2011) These methods allowed us to determine how close the data is with others within a cluster and how far away one cluster is from the others (Saputra et al, 2020). Based on the clustering results, we derived our archetypical patterns of XAI business models, which involves identifying the similarities among the focuses of our archetypes.…”