Examples of using semi-structured data

Question

Data processing

In English there is such a term - semi-structured data - semi-structured data. This is data for which the exact structure is unknown, or known, but often changing. The bottom line is that for these reasons, it can be inconvenient to store such data in relational tables.

I am writing a paper and I need specific examples of working with such data. Share your experience when you chose xml / json / document-oriented storage, etc. for storage. Tell us what the specifics of the task were.

Thank you.

Answer 1

well, an example lying on the surface is configuration files.
By the way, the reason for choosing such a repository can be not only weak structuredness, but also a frequently changing scheme.
For example, user profiles. Today they want icq, and tomorrow some identifier in My Circle. You need to move all these fields to an auxiliary table or edit the schema frequently.
The idea of ​​document-oriented repositories is very well suited for creating universal cms. There, when developing a specific site, you have to supplement the standard schemes with additional fields that are stored in an auxiliary table. One field, one line. In a real project, I had up to a hundred fields.
In cms with document-oriented stores, this problem is solved much more elegantly: you simply describe the schema, and the document is still saved as a single document.

Answer 2

I use json & MongoDB
, many product parameters are presented in a semi-structured form,
we don’t know the name of the parameters before, it’s not profitable to block three-level relational queries, see the post habrahabr.ru/qa/3248/ about the proposed data structures, in particular my comments.
I went through this rake in another project

Answer 3

You can use a universal table of the form:
record-id (int)
record-type (int)
record-value (text)
You can also add a fourth field and store a numeric value for the record in it:
Rank-records (float)