Aliaksandr Autayeu

Thanks Aliaksandr for detailed explanation. This was very helpful. Sensekey will work for my use case. I think editing a wordnet in a database might be easier to handle. Is there any specific reason for not implementing editing in DatabaseBackedInstance? Regards, Vinod

Thanks Aliaksandr, this was helpful. Below are the steps that one should follow to update the gloss of a sense. // make editable dict.edit(); // get synset and update the gloss Synset synset = dict.getSynsetAt(pos, offsetId); synset.setGloss("new gloss"); // save, close and open dictionary dict.save(); dict.close(); dict = Dictionary.getFileBackedInstance(wordnet.dir); While editing senses (gloss only), I ran into an issue where offset id of a sense was changed after some other sense was edited....

Thanks Aliaksandr, this was helpful. Below are the steps that one should follow to update the gloss of a sense. START - - - // make editable dict.edit(); // get synset and update the gloss Synset synset = dict.getSynsetAt(pos, offsetId); synset.setGloss("new gloss"); // save, close and open dictionary dict.save(); dict.close(); dict = Dictionary.getFileBackedInstance(wordnet.dir); - - - END - - - While editing senses (gloss only), I ran into an issue where offset id of a sense was changed after some...

There was no need for editing WordNet in a database in my use cases. And anyway database is secondary to files due to offset semantics. In "editable database" you still have to implement offset calculation logic, or violate their semantics. Or abandon offsets, which is undesirable in most "WordNet" cases. And when it's not critical (that is, when you don't need offsets), you often have a lot more freedom in choosing format, so when moving to databases it is often better\easier to depart from WordNet...

First, about offsets as identifiers. This gives some background to your question. Most people use offsets as "sense identifiers". But they are bad sense identifiers. The reason they are used because they are convenient for fast access and fixed for a particular WordNet version. However, sense key is a much better sense identifier - it does not change across WordNet versions and it does not change when dictionary is edited. But in some access patterns, in some circumstances, using it requires one...

Hi Vinod, I'm sorry, I don't have much time to look into this, but perhaps the following unit test can help you? It tests gloss editing. Please, see testCreateSynsetRecreate in DictionaryEditTester.java:384. Best regards, Aliaksandr

dict2db.bat is generated during build and is present in the binary release: https://sourceforge.net/projects/extjwnl/files/extjwnl-2.0.3.zip/download in the bin folder. But indeed, it is a convenience wrapper around net.sf.extjwnl.utilities.DictionaryToDatabase.

Perhaps DictionaryToDatabase.txt is relevant for your question. It explains how to populate MySQL database with WordNet data.