Deductive databases – Specialised databases series – Part 1

It’s been forever since I blogged! I’ve been working hard and studying a lot (my final year) so it’s been pretty tough. Through this though I have been studying about specialised databases and their roles in today’s world.Rather than let all my hard research work go to waste I’ve decided to do a series of blog posts on summarising some of the specialised databases out there. I’ll be lightly covering Deductive, Parallel, Spatial and Temporal Database systems. I will go slightly more in-depth into the futures of both Parallel and spatial databases and finish off with “Cloud Computing, the Semantic Web and the Future of Database Technology”, I’m hoping to release the 5 part series over 5 weeks. This will all, of course be referenced so that you can go into further reading if it interests you. I hope you guys enjoy the series and at the very least it will be a reference point for me in the future!  So first up is Deductive databases.

Deductive Database System

A deductive database uses a declarative language, one that describes what you want from the data rather than how you want to get it, to outline rules.  A “deduction mechanism” (Elmasri & Navathe, 2003) can then infer new facts about the dataset using the outlined rules.  The deductive database uses Datalog (a variation of Prolog) to define rules to be applied to relations. The Datalog language is similar to SQL in that it forms an if-then structure however Datalog allows rules to be recursive.

According to (Ramakrishnan & Gehrke, 2003) deductive databases were a solution to the limitations in the query language and constraint definition capabilities of relational database systems, confirming this  Alexander P. Pons states that in the mid-1990s researchers proposed deductive database systems to overcome the obstacles found in relational database systems (Pons, 2003).

Due to its ability to handle large amounts of data and apply reasoning and inferences a deductive database is

“Attractive to Decision Support that require a business’s historical and current information to make predictions” (Pons, 2003)

This would imply that deductive databases could be useful in a number of spheres e.g. Customer Relational Management systems, where predictions based on historical information would prove to be highly valuable to future sales and customer relationships. In a 2011 article Andreas Behrend states that until recently the deductive database inference methods had rarely been used in commercial database systems due to a lack of

“Uniform approach well-suited for implementation in an SQL-based system”  (Behrend, 2011)

A company named HighFleet have been using deductive databases to provide their clients with

“Question-answering capabilities never before available”

And comments that

“Enterprises can understand what their data means without an excrescence of heavy weight, brittle and expensive applications or without weeks to months of special effort by IT staff” (HighFleet, n.d.)

In summary deductive databases are inference mechanisms that use rules and facts to give meaning to data. Although we do see deductive databases commercially they do not compare to the strong hold of relational databases.

Let me know if you’ve got any questions, can’t promise I can answer them but I’ll do must best to find out if I can’t!

Thanks,

Sara :)

<shameful plug>If you enjoy the post please retweet ;) </shameful plug>

If you can’t wait for the series here is the full text.

References

 

Filed under: Research, Technical, , , ,

1 Response

  1. Nick Says:

    Interesting series of posts I’ll hopefully be keeping up with.

    As for deductive database systems, could you tell me how it would differ from a relational database (or any kind of NoSql) that a fed a rules engine in any specific programming language?

    I’m trying to understand why I would want to move all of my business rules and formulas into a database – albeit a database specifically defined for this purpose.

    What advantages does it give me?

    When I read the comments about “enterprises understanding their data without any IT staff”, I was thinking it was akin to web forms data grid or LightSwitch etc – software development for spreadsheet jockeys.

    How’s that for a bit of heckling to get you back into blogging? Looking forward to your answers :)

    Posted on February 5th, 2012 at 21:54

Leave a Reply