Does DBPedia vectors have vectors for the predicates?

idio / wiki2vec

Generating Vectors for DBpedia Entities via Word2Vec and Wikipedia Dumps. Questions? https://gitter.im/idio-opensource/Lobby

601 stars 137 forks source link

Does DBPedia vectors have vectors for the predicates? #18

Open zhq2009 opened 8 years ago

zhq2009 commented 8 years ago

Hello,

We are currently using the DBPedia dataset, "English Wikipedia (Feb 2015) 1000 dimension - No stemming - 10skipgram". Does DBPedia vectors have vectors for the predicates?

Thank you

dav009 commented 8 years ago

No, it only contains vectors for the entities.

I'm not sure what you exactly mean "vectors for the predicates". can you give me some examples of the predicates you want vectors for?

zhq2009 commented 8 years ago

Thank you for your help.

As on the DBPedia website http://dbpedia.org/page/Barack_Obama, the predicates are on the left part of the web page, such as "birthDate", "office", "predecessor", "residence" and "successor" etc.

Thank you very much

dav009 commented 8 years ago

Those predicates come from infoboxes fieldnames and passing them through a dbpedia mapping, right?. If so then :

you could potentially generate them as the wikipedia text is being processed and add them to the corpus.
add them to the corpus after the wiki2vec steps have completed. i.e If you have pairs like : successor(A, B) you can add the text:
- DBPEDIA_ID/A DBPEDIA_REL/successor DBPEDIA_ID/B for example to both the text of entity A and entity B