TEXTUAL METADATA INFRASTRUCTURE

TEXTUAL METADATA INFRASTRUCTURE

By W H Inmon

In a recent post about the metadata infrastructure required for textual data, it was stated that when text is formatted into a structured format that a different approach to structuring the data found in text is needed. This is due to the inherent flexibility and free form format required by text. Text is fundamentally different than structured data. Anyone can say anything or write anything they want. There is no prescribed format or constraints for speech and the written word. Text is inherently free form. As such a different structuring of data is needed when text is converted to a structured format.

It is true that text needs to be placed into a structured format in order to be analyzed using standard analytical processing. The need for a structured format is dictated by analytical software – Excel, Tableau, knowledge graph data bases, and so forth. But even so, there are some inherent differences between a classical structured structuring of data and a textual structuring of data.

In order to understand these differences consider a simple classical data base structure –

The simple structured data base shows that the column headers directly specify – constrain - what data is to be found in the column. In the column for school, only a school belongs there. In the column for city, only cities and towns belong there. The column header directly describes and constrains the contents of the column.

Now consider a data base approach for textual data –

In the textual data base, the column heading for word indirectly describes what is in the column. The column for word contains only words. But the words can be anything. The column heading for word places no constraints on what the content might be for the column. The direct description of the contents of the word column are described by another column – context. The data found in context in the same row describes the contents of the data found in word.

The column “context” allows for there to be a direct description of the contents of the word column.

In a structured data base there are only direct column headings. In the textual data base there are indirect and direct descriptors of data.

The ability to have many types of data in the word column allows the structured data base for text to accommodate undisciplined and unstructured text.

?

Bill Inmon lives in Denver with his wife and his two Scotty dogs – Jeb and Lena.? It is a nice cool winter day in Denver. It snowed last night but melted off by noon. Jeb and Lena went out to the back yard in the sun and played. Lena is faster than Jeb because she is younger than him. But Lena puts up with Jeb in any case.