A Data-Centric Methodology and Task Typology for Time-Stamped Event Sequences

Abstract

Figure 1: Our methodology for task abstraction and taxonomy-building for dataset-specific tasks comprises five phases. A variety of methods can be applied in the three early phases of data collection, coding, and task categorization, for moving forward with two study sources. Phase four includes a task synthesis, followed by the fine-grained elaboration on action-target-(criterion) crosscuts.

IVESA - Visual Analysis of Time-Stamped Event Sequences

“…Some tasks, such as the identification of trends, are special and cannot be easily transferred to TSEQs. However, with a certain degree of abstraction, some user goals and tasks for time series analysis can also be adopted for the analysis of TSEQs [82]. Examples include preprocessing [13,14,21] (segmentation and alignment in the TSEQs case), the representation of data with features through descriptors [38,55,120] (through metrics in the TSEQs case), the discovery of patterns/motifs/subsequences [64], content-based similarity, search, and retrieval [76,86], segmentation [12], prediction [74], exploratory analysis [118], and clustering [35,62].…”

“…For the task abstraction, we used the typology of analysis tasks proposed by Peiris et al [82]. The typology guided our abstraction process by offering a long list of 23 tasks at a comparable abstraction level, that the authors derived from 65 interviews with non-experts and 16 design studies related to TSEQs.…”

“…In Figure 3, we present task abstraction results using the targetcriteria crosscut method [82]. With the triples consisting of actions, targets, and criteria, we arrived at a fine-grained level of task description, to be leveraged in the iterative design phase of IVESA.…”

“…Fig.3. Task abstraction using the target-criteria crosscut method[82]. Overall, the domain experts are mainly interested in 12 task actions, applied to different data targets and based on different data criteria.…”