Data Leaves: Scenario-oriented Metadata for Data Federative Innovation

TL;DR

This paper introduces a scenario-oriented metadata method for datasets that emphasizes real-world events and actions to enhance data integration, usability, and practical knowledge acquisition for innovative business and AI applications.

Contribution

It presents a novel metadata representation connecting datasets through scenarios, improving relevance and applicability for real-world data use cases.

Findings

Enhances data integration via scenario-based metadata

Improves alignment of data with real-world applications

Facilitates practical knowledge extraction from datasets

Abstract

A method for representing the digest information of each dataset is proposed, oriented to the aid of innovative thoughts and the communication of data users who attempt to create valuable products, services, and business models using or combining datasets. Compared with methods for connecting datasets via shared attributes (i.e., variables), this method connects datasets via events, situations, or actions in a scenario that is supposed to be active in the real world. This method reflects the consideration of the fitness of each metadata to the feature concept, which is an abstract of the information or knowledge expected to be acquired from data; thus, the users of the data acquire practical knowledge that fits the requirements of real businesses and real life, as well as grounds for realistic application of AI technologies to data.