Clustering for taxonomy-based archetype identification has become an established method in Information Systems (IS) research, aiding strategic decision-making across diverse research and business domains. However, the effectiveness of the approach depends critically on the compatibility of clustering methods and algorithms with the specific data characteristics. This study, based on a comprehensive review of 87 articles employing taxonomy-based clustering in IS research, reveals a notable mismatch between the chosen clustering algorithms and the nature of the data, particularly in the context of archetype development from taxonomy-based data. To address these methodological inconsistencies, we introduce a set of clustering guidelines tailored to the unique requirements of archetype development from taxonomy-based data. These guidelines are informed by a computational study involving seven identified datasets from the taxonomy-building literature, ensuring their practical applicability and scientific relevance. Our guidelines are designed to enhance the robustness and scientific validity of insights and decisions derived from taxonomy-based clustering. By improving the methodological rigor of clustering methods, our research addresses a critical mismatch in current practices and contributes to enhancing the quality of decision-making informed by taxonomy-based analysis in IS research.

中文翻译：

---

重新评估基于分类的数据聚类：揭示应用见解和指南


基于分类法的原型识别的聚类已成为信息系统 （IS） 研究中的一种既定方法，有助于跨不同研究和业务领域的战略决策。但是，该方法的有效性在很大程度上取决于聚类方法和算法与特定数据特征的兼容性。这项研究基于对 IS 研究中采用基于分类法的聚类的 87 篇文章的全面回顾，揭示了所选聚类算法与数据性质之间的明显不匹配，尤其是在从基于分类法的数据开发原型的背景下。为了解决这些方法上的不一致，我们引入了一套聚类指南，这些指南是根据基于分类的数据进行原型开发的独特要求而量身定制的。这些指南以一项计算研究为依据，该研究涉及从分类学构建文献中确定的七个数据集，确保其实际适用性和科学相关性。我们的指南旨在增强从基于分类法的聚类中得出的见解和决策的稳健性和科学有效性。通过提高聚类方法的方法严谨性，我们的研究解决了当前实践中的严重不匹配，并有助于提高 IS 研究中基于分类法的分析所告知的决策质量。