Work

Semantic time series analytics leveraging incremental knowledge graph enrichment

Year: 2026
Type: dissertation
Abstract: The development and utilization of modern, digital technologies, such as wearable devices or automated, intelligent machines, have led to a substantial increase in time series data across many domains including healthcare, finance and industry. Although this data holds significant potential for deriving actionable insights regarding strategic decision-making, a common problem of organizations is to systematically interpret and reuse the knowledge generated in analytical processes. While different task-specific analyses are carried out on collaboratively used storage systems, produced insights commonly stem from isolated workflows and remain implicit, fragmented, or tied to particular models and projects. This independent processing limits accessibility and reusability. Furthermore, generated insights are often produced without incorporating scenario-specific context and expert knowledge, which reduces their interpretability and degree of validation. In fact, no common solution for classifying and linking prior knowledge and derived insights across the full scope of time series analytics exists, which prevents synergies such as knowledge transfer and propagation. This thesis addresses the challenges of systematic knowledge integration from and for time series analytics in combination with related contextual facts. It proposes a comprehensive, semantically enriched knowledge graph, which serves as a central, machine-interpretable repository that enables the structured storage, contextualization, and exploitation of knowledge associated with time series data. The core idea is a continuous enrichment of facts, making the graph a dynamically growing repository of exploitable insights.
Language: English
Open Access status: green
OpenAlex ID: W7225296695
Cites:
Cited by:
Related to: 10