On Wed, 2018-05-02 at 17:54 +0000, Eren Erdemli wrote:
> Hi Robert,
> 
> :) I suppose this is where it gets confusing
> 
> I am thinking of structuring it with BT 2x2 max depth and length,
> 
> i.e
> 
> "a"     -> /dict/a (virtual=false)
> "axe"  -> /dict/ax/axe
> "apache" -> /dict/ap/ac/apache

Where would the language fit it. Is it a property of the word entry
node? If it is, how would you handle multiple languages?

> 
> 
> Another question arises with this is how do we back reference in to
> foreign
> representation.
> i.e where lets say axe = hache,
> 
> maybe use of  mix:referencable and reference to it
> "apache" -> /dict/ap/ac/apache/@references['xxxxxxx-xxxxxx']
> 
> or when you have a word that has same spelling in multiple languages

I would use simple path properties rather than references. References
tend to bloat the UUID index.

One option is to structure by language, e.g.

/dict/en_EN/ax/axe

You could add mix:title for title and description properties [1] to
your node. Also you could have a multi-valued 'translations' properties
which is of type path and points to the translations. Not sure if that
is enough and maybe you want to have a node per translation to add more
metadata.

Robert


[1]: https://docs.adobe.com/docs/en/spec/jsr170/javadocs/jcr-2.0/javax/
jcr/nodetype/package-summary.html#mix:title
> 
> 
> Eren
> 
> 
> On 2 May 2018 at 14:28, Robert Munteanu <[email protected]> wrote:
> 
> > Hi Eren,
> > 
> > On Mon, 2018-04-30 at 20:36 +0300, Eren Erdemli wrote:
> > > Greetings,
> > > 
> > > Might be a naive question,
> > > 
> > > I have a multi dictionary app built on mysql which is around 15
> > > years
> > > old.
> > > and have around 500GB of data
> > > 
> > > I wonder if using oak mongomk is suitable option for porting this
> > > app,
> > 
> > Maybe :-)
> > 
> > How do you see the data structured in Oak? I guess the first idea
> > would
> > be to somehow avoid having a large number (>10k) of orderable child
> > nodes directly below a single node.
> > 
> > Robert
> > 

Reply via email to