TITLE:
A Single-Logic and Minimal-Data-Unit Framework for Human-Computer Interaction in Data Analysis and Modeling
AUTHORS:
Mingyang Wei
KEYWORDS:
Human-Computer Interaction, Visual Analytics, Visualization Provenance, Explainable AI, Human-in-the-Loop Machine Learning, Design Science Research, Interaction Traceability
JOURNAL NAME:
Journal of Computer and Communications,
Vol.14 No.11,
November
5,
2026
ABSTRACT: Contemporary data analysis systems combine large datasets, visual dashboards, machine learning models, and automated decision-support tools, but their analytical processes can be difficult to follow when operations are distributed across complex interfaces, hidden model structures, and multistage workflows. This paper proposes and partially instantiates a Single-Logic and Minimal-Data-Unit (SL-MDU) framework for human-computer interaction in data analysis and modeling. A Single Logic (SL) is a registered analytical operation that satisfies four testable atomicity criteria—declared input, one rule with exposed parameters, one inspectable output, and no intermediate result the task requires the user to check; a Minimal Data Unit (MDU) is the smallest data entity that is meaningful for the task at a declared granularity level. Relative to subset-flow, modular-composition, and provenance systems, SL-MDU contributes a prescriptive task-level contract: each result is accountable to an explicit (MDU, SL, parameters, output) record, rule outputs are kept distinct from human overrides, and a granularity policy decides which units are exposed. Following a design science research approach, the framework was abstracted from a working keyword-driven web implementation. We report code-level evidence of the implemented dispatch mechanism and a technical evaluation in a browser: on an illustrative 142-student × 4-indicator workload, the unmodified dispatcher executed 710 per-unit operations in a median of 30.5 ms, or 0.27 ms when the same rules were applied at cohort granularity, and per-unit execution scaled linearly to 2.7 s for 14,200 students. A heuristic evaluation by three evaluators (two developers and one professional tester) recorded 41 usability problems (mean severity 2.3 on a 0 - 4 scale), most severely the lack of undo and weak feedback on which unit an operation targets, together with 18 positive findings. Effects on user understanding, cognitive load, and task performance are stated as a falsifiable hypothesis for a subsequent controlled study rather than claimed.