Publikation
Explanation of Similarities in Temporal Case-Based Reasoning by Visualization
Roman Schander; Alexander Schultheis; Ralph Bergmann
In: Case-Based Reasoning Research and Development - 34th International Conference, ICCBR 2026, Bremen, Germany, August 13th - 16th, 2026, Proceedings. International Conference on Case-Based Reasoning (ICCBR-2026), 34th International Conference on Case-Based Reasoning, August 13-16, Bremen, Germany, Lecture Notes in Computer Science (LNCS), Springer, 2026.
Zusammenfassung
Temporal Case-Based Reasoning (TCBR) relies on time series similarity measures whose global similarity values emerge from multiple local contributions, making the development, validation, and maintenance of similarity models difficult for knowledge engineers. This paper investigates visualization-based explanations for mapping-based time series similarity measures that illustrate how such values are composed. To this end, a systematic literature study on visualization approaches for time series similarities is conducted, and a requirements catalog for similarity visualizations in TCBR is derived. Existing techniques are then assessed against these requirements. Based on the results, an extended point mapping visualization is presented that makes mappings between time series entries explicit and complements them with interactive tools for exploration and inspection. The concept is implemented in the ProCAKE framework and was evaluated in an exploratory user study. The results suggest that the approach is perceived as interpretable and practically useful for supporting the understanding of time series similarity values and their local composition in TCBR.