Hi Dan, Thank you so much for your input and sorry for late reply, some how I missed your message and just checked the web interface!
My set of abbreviations are limited and all I need is expanding them in one-to-one relationship, so I think I think a simple dictionary should suffice. The closed I could find is an acronym with their full form from v 1.4 that could be helpful to someone else, here is the link: http://opennlp.sourceforge.net/models-1.4/english/coref/acronyms In my specific use case, I am going to normalize the data before using the parser, so it would be easier to do this outside of opennlp, was hopping to find something built-in. Thanks again and hope this file could help someone else. Best regards, Ahmed On Fri, Apr 24, 2015 at 7:15 PM, Daniel Russ <[email protected]> wrote: > Hi Ahmed, > > Of course there is no easy solution or you would not have needed > help. I am not familiar with any available models for abbreviations > finding. Are you concerned with disambiguating St. (Street) from St. > (Saint), or do you have a list of abbreviations you just want to replace? > If you are just doing string replacement, a simple hashmap should suffice. > Otherwise, first you need to identify that a word is an abbreviation, and > then what to replace it with. You might want to have a map from > abbreviation to a list of potential meaning, and train a model to > disambiguate between potential meanings. For example St is in the map > pointing to {Street, Saint}, then depending on the words before and after > (and whether they are capitalized) decide between the possible outcomes. > > Someone with more experience than me can probably give you better > advice. I hope that I can get you started. Also let us know if your get a > working model. It may help others later. > > Dan > > > > On Apr 24, 2015, at 12:54 PM, Ahmed Aboulenein wrote: > > > Hi ALL, > > > > Is there an easy way to expand the abbreviation in a String? > > > > For example: "I live close to main st." becomes "I live close to main > > street." > > > > Many thanks, > > Ahmed > >
