[jira] Issue Comment Edited: (CLEREZZA-20) Stable serialization

["Daniel Spicar (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CLEREZZA-20?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12798651#action_12798651 ] 

Daniel Spicar edited comment on CLEREZZA-20 at 1/11/10 12:37 PM:
-----------------------------------------------------------------

I attached a solution for stable serialization of graphs in Clerezza based on blank node labeling algorithm by Jeremy Carroll and Minimum Self-contained Graph algorithm by Giovanni Tummarello, Christian Morbidoni, Paolo Puliti, Francesco Piazza, Università Politecnica delle Marche, Italy

It doesn't guarantee a deterministic result. 

At the Moment it makes use of the Jena serializer for the initial step. This may be changed easily. Using a custom N-Triples serializer could potentially also increase performance as some later steps could be merged into the initial serializing process

      was (Author: daniel.spicar):
    A solution for stable serialization of graphs in Clerezza based on blank node labeling algorithm by Jeremy Carroll and Minimum Self-contained Graph algorithm by Giovanni Tummarello, Christian Morbidoni, Paolo Puliti, Francesco Piazza, Università Politecnica delle Marche, Italy

It doesn't guarantee a deterministic result. 

At the Moment it makes use of the Jena serializer for the initial step. This may be changed easily. Using a custom N-Triples serializer could potentially also increase performance as some later steps could be merged into the initial serializing process
  
> Stable serialization
> --------------------
>
>                 Key: CLEREZZA-20
>                 URL: https://issues.apache.org/jira/browse/CLEREZZA-20
>             Project: Clerezza
>          Issue Type: New Feature
>            Reporter: Reto Bachmann-Gmür
>         Attachments: StableSerializerProvider.java
>
>
> We shall provide an N-TRIPLES serializer that provides some stability in the serializations, i.e. that small changes in the graph cause a small change in the serialized form (and that the same graph often results in the same serialization). While full deterministic serialization is too computation expensive in the general case, some relatively simple adpatations in the b-node labeling algorithm and the order of the triples could give usable stability, thus making the nt-documents (such as the documentation.nt files) much more suitable for storage within a a VCS.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.