#! /usr/bin/env python

# -*- coding: utf8 -*-

import timbl

import os

#We are building a very simple context-aware translator Word Sense Disambiguator for the word "bank", based on the occurrence of some keywords in the same sentence:

# The features are binary and represent presence or absence of certain keywords. We choose:

# - money

# - sit

# - river

#They have a value of 0 or 1 (but note that Timbl support string features just as well!)

#The classes we predict are:

# - financial

# - furniture

# - geographic

#Build the classifier training

classifier = timbl.TimblClassifier("wsd-bank", "-a 0 -k 1" ) #wsd-bank will be the prefix of any files written for timbl

classifier.append( (1,0,0), 'financial') #append is used to add training instances

classifier.append( (0,1,0), 'furniture')

classifier.append( (0,0,1), 'geographic')

#Train the classifier

classifier.train()

#Save

classifier.save()

#We start anew and load the classifier again (of course we could have just skipped this and the save step and continued immediately)

classifier = timbl.TimblClassifier("wsd-bank", "-a 0 -k 1" ) #wsd-bank will be the prefix of any files written for timbl

classifier.load() #even if this is omitted it will still work, the first classify() call will invoke load()

#Let's classify an instance:

classlabel, distribution, distance = classifier.classify( (1,0,0) )

if classlabel == "financial":

print("Classified correctly! Our accuracy is " + str(classifier.getAccuracy()))

#Let's classify an ambiguous one:

winningclasslabel, distribution, distance = classifier.classify( (1,1,1) )

for classlabel, score in distribution.items():

print(classlabel + ": " + str(score))

print("Distance: ", distance)

#We again start anew and build a test file

if os.path.exists("testfile"): #delete if it already exists

os.unlink("testfile")

classifier = timbl.TimblClassifier("wsd-bank", "-a 0 -k 1 +v n+di+k" ) #add some extra verbosity flags

classifier.load()

classifier.addinstance("testfile", (1,0,0),'financial' ) #addinstance can be used to add instances to external files (use append() for training)

classifier.addinstance("testfile", (0,1,0),'furniture' )

classifier.addinstance("testfile", (0,0,1),'geograpic' )

classifier.addinstance("testfile", (1,1,0),'geograpic' ) #this one will be wrongly classified as financial & furniture

classifier.test("testfile")

print("Accuracy: ", classifier.getAccuracy())

print("Best neighbours: ", classifier.bestNeighbours()) #this only works with the extra verbosity flags and only if python-timbl is compiled with gcc