Characterizing Regulatory Documents and Guidelines Based on Text Mining

Implementing rules, constraints, and requirements contained in regulatory documents such as standards or guidelines constitutes a mandatory task for organizations and institutions across several domains. Due to the amount of domain-specific information and actions encoded in these documents, organizations often need to establish cooperations between several departments and consulting experts to guide managers and employees in eliciting compliance requirements. Providing computer-based guidance and support for this often costly and tedious compliance task is the aim of this paper. The presented methodology utilizes well-known text mining techniques and clustering algorithms to classify (families) of documents according to topics and to derive significant sentences which support users in understanding and implementing compliance-related documents. Applying the approach to collections of documents from the security and the medical domain demonstrates that text mining is a promising domain-independent mean to provide support to the understanding, extraction, and analysis of regulatory documents.