Data Timeliness Evaluation Based on Attribute Value Dependencies

Abstract

Data timeliness, a critical determinant of data quality, can distort data analysis outcomes and lead decision-makers astray when utilizing outdated data for mining. Current methodologies for exploring data timeliness encompass the recovery of temporal data sequences and the determination of timeliness relying on factors such as timestamps and uncertain data relationships. Timestamp-centric techniques are constrained by their dependence on precise temporal values. In contrast, methods grounded in uncertain relationships among data entities excel in managing multiple tuples of same entity. However, limited research addresses data timeliness while considering dependencies among data attribute values. This paper addresses this gap by investigating data timeliness within the context of attribute value dependencies. The paper commences by introducing pertinent concepts and regulations related to attribute value dependencies. It then defines the prevailing issue and proposes regulations alongside a data timeliness assessment model. Furthermore, the paper presents algorithms for extracting and consolidating data attribute value dependency regulations, in conjunction with an algorithm for timeliness assessment. The efficacy of these algorithms and the accuracy of the data timeliness assessment model are validated using real-world datasets. The experimental results affirm that the proposed algorithms demonstrate high execution efficiency and yield precise data timeliness assessments.