[open-linguistics] Linguistic Linked Open Data cloud, sketch

Christian Chiarcos christian.chiarcos at web.de
Tue Jan 17 12:01:26 UTC 2012


Dear all,

a key result of the last telco were commitments by several participants to  
provide different types of linguistic data, including various  
lexical-semantic resources (marked green), corpora (marked orange), and  
meta data repositories (marked blue).

I have summarized possible links between resources we discussed at this  
occasion in the diagram under https://cacoo.com/diagrams/jmtxae5nl0vuYblW.  
For the abbreviations used, please see the minutes of the last telco under  
http://okfnpad.org/OWLG. (Unfortunately, cacoo does not support arrows,  
so, all [more or less] horizontal lines are to be read as directed edges  
pointing from left to right, whereas vertical lines indicate bidirectional  
linking -- suggestions for alternative software for collaborative graph  
drawing are highly welcome.)

The diagram can be edited online. So far, only resources were considered  
that the participants of the telco were directly involved with. Please  
feel free to add additional resources already available from the LOD cloud  
or that can be provided online.

The linking between the resources as shown there is partly hypothetical  
and not yet implemented, and should be discussed more thoroughly in the  
next telco. Additionally, one should take linking granularity into  
account. As Judith pointed out before, linking Wiktionary dumps at the  
lexeme level would be easy, but Wiktionary *sense* level alignment could  
be tricky (no stability for sense IDs).

In my opinion, would be sufficient have a coarse-grained (and partially  
trivial) linking between resources at the moment, to release the data in  
this form openly and to invite others to develop their own and possibly  
less trivial linkings. For example, I would expect that an algorithm that  
automatically creates a sense-level linking out of an lexeme-level linking  
(e.g., on the basis of the comments associated with it) for, say, WordNet  
and Wiktionary, should also be applicable to different Wiktionary dumps.

Best,
Christian




More information about the open-linguistics mailing list