A true virtual graph for Excel and CSV hosted remotely

I know one can do virtual graph import of CSV - which really does not produce a virtual graph but rather a materialized graph.

What if Excel and CSV files are maintained by domain experts and available on some file content repository? Instead of downloading and re-importing it every time:

  • Can Stardog leverage an ODBC driver for a given Excel or CSV and create a truly virtual graph representation of an Excel or CSV subset?

Thanks,
Radu

Hey Radu,

We have contemplated implementing something like this but have no immediate plans to do it. Perhaps you can setup a script to load the CSV into Stardog at a regular interval?

Jess

Jess,

Thanks for your reply. Yes such old approach - cron enabled pull script - can be used. It will require custom E(TL) code to be developed - which I tried to avoid by using virtual/material graph import from Excel and or CSV. I bet the generic E part is already done by some open source tools. That involves:

  1. Periodically checking if the file changed
  2. Finding delta
  3. Loading delta per SMS2 mapping

As much as we would like not to be the case - Business Users are in love with Excel.

Looking forward to when this feature will be on Stardog roadmap.

Regards,
Radu

Radu,
How do you suppose a "true" virtual graph would be any better here? It's required to scan the whole file to compute changes. It's easier to just reload the whole thing at this point.
Jess

Jess,

This should provide a good enough abstraction layer https://www.cdata.com/drivers/excel/jdbc/

I suppose you already can do virtual graph against MS SQL Server so it shouldn't be that different ... I hope. I haven't tried though...

Regards,
Radu

Sure, you can create a virtual graph using this JDBC driver. Let us know if you need any help customizing the settings.