Injecting background knowledge into the word vectors
- Place your texts under the
corporafolder. Script assumes one text per line, splits sentences on dot. - Launch TextReader.py. It will train word vectors and save them into
model.w2vfile - Fill free to implement a proper script that will actually take parameters and won't use the hardcoded ones
java -cp thewikimachine.jar org.fbk.cit.hlt.thewikimachine.xmldump.WikipediaTextExtractor
-d <path-to-dump.xml>
-o <path-to-output-directory>
-t <amount of threads>