Question

I am parsing a text document and converting much of the data into a relational format. A variety of date formats are used throughout, but one of these is “Fall 1997”.

What would be the best way to go about storing this information, which is non-specific in one sense (not an absolute date), without losing the granularity that “fall” does provide?

As additional info: my specific database is MySQL.

Answer 1

There are various techniques I’ve seen, but without loading it in straight varchar, you’ll have to decide on the scope. You can use a regular date column containing 1997-09-01 but augment it with a column which indicates the specificity – ‘S’ for season (using the firsts of months 3, 6, 9, 12) ‘Q’ for quarter (using the firsts of months 1, 4, 7, 10) or whatever. Similarly for semesters and things like that.

When interpreting such dates – particularly in “did this event happen before or after a certain time”, you have to decide how to handle them. Like whether the whole period has to come before a date.

You can convert such a “date” to an entry in a period table with a description and a start and end date and then link to the period. Then Fall 1997 is a database entity in the sense that it is well defined as a row in a table and other rows have foreign keys to it. This does not remove the need for logic to decide things about these dates, but it does mean that such logic can be table driven.

It would depend a lot more on the usage scenarios and the variety of data before I would make a call.