Yes, basically, you have too large of a branching factor and that causes the beam search to fail. You do need to determinize and minimize.
I've written Python bindings for OpenFST that make this fairly easy; I am going to try to get all of this up on a server soon (we're moving right now). I'll announce to the group when that's done. Tom On Wed, Jul 15, 2009 at 11:59, Pierpaolo Monaco<[email protected]> wrote: > I'm trying to build my own language models. > I have a list of all the Italian town, and I'm trying to recognize them, in > handwritten uppercase character, for proof. > I am building the language model using openFST. > For now i'm not optimizing the model, so every word in the graph have a > different path, starting from a common start state and all ending in a own > state, that is the last letter of the word. > The model works until the words that i insert in my "dictionary" are less > then 100. > > overco...@overcomer-laptop:/media/disk/temp/prova/elab$ ocropus page Doc\ > 1RF400.tif > TORINO > > > When I insert 100 or more words i get the error. > > overco...@overcomer-laptop:/media/disk/temp/prova/elab$ ocropus page Doc\ > 1RF400.tif > [beam search failed] > > What's going on? There is a limit of a number of arcs going out from one > state? > > Regards, > > -- > ----------------------------- > Pierpaolo Monaco > ---------------------------- > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "ocropus" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/ocropus?hl=en -~----------~----~----~----~------~----~------~--~---
