回复：Re: Patent application on converting SPARQL to SQL

I happened to read this patent when I worked for IBM research. Indeed, there is a "sister" patent named "RDF Store Database Design for Faster Triplet Access" by the same inventors. These two patents describe methods to store and query RDF data in relational databases. One patent focuses on storage work and claims a de-normalized schema to store RDF data, and the other one focuses on query work and claims a method to translate SPARQL queries to SQL queries over the de-normalized relational schema. To my understanding, such patents do not affect the WG's work too much.
 
FYI, IBM China Research (I used to be a staff member of it) has done some work on both RDF store [1,3] and mapping based SPARQL access to relational data [2]. [2] describes a method to support SPARQL queries over relational databases using D2RQ mapping language and with rule inference support. The mapping based SPARQL to SQL query translation is closely combined with rule inference.
 
I have to say regret for this night's conference call since I will have a meeting early tomorrow morning (my time). Thanks for your understanding.
 
[1] Effective and efficient semantic web data management over DB2, http://portal.acm.org/citation.cfm?id=1376616.1376735

[2] Semantic Enhancement for Enterprise Data Management, http://data.semanticweb.org/papers/iswc/2009/in-use/paper157.pdf

[3] An Effective SPARQL Support over Relational Databases, http://www.springerlink.com/content/u8217106407787n8/

 
Best regards,
 
Li Ma
----- 原始邮件 -----
发件人：Juan Sequeda <juanfederico@gmail.com>
收件人：ashok.malhotra@oracle.com
主题：Re: Patent application on converting SPARQL to SQL
日期：2009-12-15 21:43:57

As you know, we (UT Austin) have been working with Microsoft, but internally, there is a lot of miscommunication (I guess this may be common in large corporartions).


We are working with the lead guy in the SQL Server optimization group. He was not aware of this and I asked him about this patent. He is looking into this.燑/div> 


And, if I am not wrong, Microsoft bought Intellidiension [1], so this patent may be coming from them.


I know Eric has already a lot of experience in SPARQL to SQL. We are also writing up a report of the SPARQL to SQL stuff thats in Oracle (our intent is to submit to ESWC2010).燑/div> 


At the moment, the most complete and formal work that I know of is from Chebotko et al. Here is a TR [2] and here is the journal version [3]. Another work is from Cleveland Clinic from their rdf store [4].


I've been going through Chebtoko's stuff. Very complete... too complete IMO. He defines a relational alegbra for sparql and then proves that the sparql alegbra can be translated to the new sparql relational alegbra. In our work, we haven't "complicated" our selves that much. I don't think Eric has either. But this is for another discussion.


Cheers and talk to you all later.




[1]燑a href="http://www.intellidimension.com/">http://www.intellidimension.com/

[2]燑a href="http://www.cs.wayne.edu/~artem/main/research/TR-DB-052006-CLJF.pdf">http://www.cs.wayne.edu/~artem/main/research/TR-DB-052006-CLJF.pdf

[3]燑a href="http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6TYX-4W32KH3-1&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&_docanchor=&view=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=08d07b63a9c6a905139e0f6634cd76f9">http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6TYX-4W32KH3-1&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&_docanchor=&view=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=08d07b63a9c6a905139e0f6634cd76f9

[4]燑a href="http://portal.acm.org/citation.cfm?id=1620437">http://portal.acm.org/citation.cfm?id=1620437



Juan Sequeda, Ph.D Student
Dept. of Computer Sciences
The University of Texas at Austin
www.juansequeda.com
www.semanticwebaustin.org



On Tue, Dec 15, 2009 at 8:29 AM, ashok malhotra <ashok.malhotra@oracle.com> wrote:

This is a patent application. 燭he patent has not been granted yet. 燫egardless of what we feel are the chances
for its success we should argue against it because, if it is granted, all vendors may end up paying Microsoft to
license the technology. 
What we need are publicly available papers that describe SPARQL to SQL mapping. 營n the reference from Ivan Herman
Jim Hendler says they did some work in this area. 燩erhaps Virtuoso has something in its product literature that we can cite. 燭here may be other papers that RDB2RDF WG members have published.

The technique seems to depend on a particular method for storing the triples. 營 could not figure this out as I did not
get the figures with the application. 營f, of course, the technique works only with a special storage strategy then it does
not apply if you store the triples in a different way.
All the best, Ashok





Michael Hausenblas wrote:

All,

Just stumbled upon [1] that there is a pending SPARQL to SQL patent
application [2] by Microsoft ... anyone knows more about this?

Cheers,
燤ichael

[1] http://twitter.com/ivan_herman/status/6686176991
[2] http://www.faqs.org/patents/app/20090138437

燑br>