Encoding Techniques for Digital Trace Data

This chapter discusses the role of encoding methods in effectively preparing trace data for a wide range of data mining tasks. By exploring the relationship between trace data and both shallow and deep learning methods, this chapter argues for a systematic exploration of the potential of encoding techniques in representing trace data. It highlights their impact on classification and prediction tasks, enabling the extraction of data-driven insights that trigger process optimization and automation. It provides an up-to-date overview of key coding techniques, their categorization, and a nuanced analysis of their specific advantages and disadvantages from different perspectives. These perspectives include effectiveness, efficiency, expressiveness, correlation power, scalability, and domain agnosticism. The selection of an appropriate approach depends on specific requirements and contextual constraints. To support this choice, we systematically compare different solutions using a memory-time projection space that categorizes coding techniques. Finally, we highlight some of the open challenges and anticipated future research directions in the field. This chapter serves as a comprehensive guide for researchers and practitioners seeking to harness the potential of trace data for improved decision-making and optimization in their respective domains.