2 Replies Latest reply: Mar 20, 2006 5:52 PM by 478870 RSS

    query  regarding oracle10g RDF store

    478870
      hi

      I have few questions regarding the oracle10g rdf functions,

      1) Right now it supports just RDF-schema and as it does not support many other constructs, are you guys planning to include Owl-schema to RDF store

      2) Is there anyway that I can add myself new schema like OWL-schema to the existing RDF schema

      3) This question is regarding the loading process. Its took me around 12 hours to load 150mb file( around 1.4 million triples). I am totally scared as I need to upload the total Uniprot RDF which contains around 80 million triples, it might take weeks or may be months if there are any network problems.Are there any suggestions to speed up the loading process.Right now I am using the "NTripleConverter.java" loader.

      4)Shall we expect any new java api supporting the RDF store, as right now there are only two associated functions(something with features of sparql)

      5)This is a problem which i discussed even previously in the board regarding "re-loading" same data after deleting the models and tables associated with the previous data. I feel that this problem is due to constant checking of duplicates when re-loading the same data. Any suggestions to over come this problem?

      6)When can we expect the new version as there were few bugs in the present version?

      Sorry for the lengthy questionaire as we really need some help in this new technology.

      thanks chandra
        • 1. Re: query  regarding oracle10g RDF store
          276592
          The answer to these questions was originally posted at:
          query  regarding oracle10g RDF store

          ---
          hi

          I have few questions regarding the oracle10g rdf functions,

          1) Right now it supports just RDF-schema and as it does not support many other constructs, are you guys planning to include Owl-schema to RDF store
          We are considering adding support for a subset of OWL and adding an interface to integrate external OWL inferencing engines.

          >
          >
          2) Is there anyway that I can add myself new schema like OWL-schema to the existing RDF schema
          If they can be expressed as user-defined rules, then yes (note that it will not be possible to express all of OWL as user-defined rules). See the documentation for defining and adding user-defined rules to the database.

          >
          >
          3) This question is regarding the loading process. Its took me around 12 hours to load 150mb file( around 1.4 million triples). I am totally scared as I need to upload the total Uniprot RDF which contains around 80 million triples, it might take weeks or may be months if there are any network problems.Are there any suggestions to speed up the loading process. Right now I am using the "NTripleConverter.java" loader.
          With 10.2.0.2 (the patch is now available on metalink) we support a fast batch loader. Please upgrade your database to 10.2.0.2. The java files for the fast batch loader will be available shortly on OTN at http://www.oracle.com/technology/tech/semantic_technologies/index.html under 'Sample Code'. Your loading speed should improve significantly. Some loading figures are below:

          The test machine is a Hyperthreaded Intel P4 3.0 GHz processor with 512 KB cache and 2 GB physical memory. The operating system is Red Hat Enterprise Linux AS release 3 (Taroon Update 4) Kernel Rev 2.4.21-27, with Oracle Database 10g Enterprise Edition 10.2.0.2 Release Candidate 4 plus RDF patch (SGA size 600MB). The times reported are the mean results of 3 trials with warm caches (except for the 100-million-triple dataset, which was only loaded once). The UniProt dataset was used for the tests.

          Triples Loading Time

          10 K 20 sec
          100 K 3 min
          1 M 27 min
          5 M 2 hr 35 min
          10 M 6 hr
          100 M 77 hr

          >
          >
          4)Shall we expect any new java api supporting the RDF store, as right now there are only two associated functions(something with features of sparql)
          Do you mean a Java API for a function like SDO_RDF_MATCH?

          >
          >
          5)This is a problem which i discussed even previously in the board regarding "re-loading" same data after deleting the models and tables associated with the previous data. I feel that this problem is due to constant checking of duplicates when re-loading the same data. Any suggestions to over come this problem?
          This problem has been addressed (and the load consequently made faster when there are duplicates) and a patch will be available on metalink shortly. I will publish the Note number on this forum when it is available.

          >
          >
          6)When can we expect the new version as there were few bugs in the present version?
          Additional patches for bugs will be available on metalink shortly. I will publish the Note numbers on this forum when they are available.

          >
          >
          Sorry for the lengthy questionaire as we really need some help in this new technology.

          thanks chandra
          Melli
          • 2. Re: query  regarding oracle10g RDF store
            478870
            Thanks Stephen for the suggestions.We will upgrading to the new version by next week and will let you know if the problem still persists.

            chandra