Improving documentation processes with data mining methods: A case study in the aviation industry

Many businesses have to operate within the framework of national and international standards to meet customers' expectations, security requirements, and environmental regulations. Compared to other industries, the aviation industry has high-security requirements. Following the fundamental quality principle of "Write what you do and do what you write" and standard obligations, it is essential to interpret the activities from a process perspective and document all processes. However, the continuous publication of documents has resulted in document inflation over time, document-process structure that deviates from a simple philosophy. Given the problem at hand, this study aims to develop a streamlined document processing system that aligns with the lean philosophy in the aviation company under investigation. The thesis consists of two chapters. In the first chapter, the cosine method, a distance-based approach, was used to determine the similarity rates among all published English documents. Documents with a similarity rate exceeding 50% were examined, and a proposal was made to either combine these documents or establish relationships between them. The second chapter focuses on predicting the category to which a document to be published belongs. Various machine-learning algorithms are used in this section. A hybrid model consisting of a Support Vector Machine (SVM) and Logistic Regression algorithms achieved an increase of 6%, reaching a 76% accuracy rate.The application was implemented using Python programming language.