def simpleSentiment(trainingset, devin, devout):
temp = parseFile(trainingset)
tokens = temp[0]
tags = temp[1]
print("Training started.")
emissionTable = calculateEmission(tokens, tags, 3)
print("Training Completed.\n")
inputTokens = parseFileInput(devin)
inputTags = []
pprint(inputTokens)
# max(stats, key=stats.get)
for i in inputTokens:
if i == None:
inputTags.append("#SPACE#")
else:
if i in emissionTable:
inputTags.append(
max(emissionTable[i], key=emissionTable[i].get))
else:
inputTags.append(
max(emissionTable['#UNK#'],
key=emissionTable['#UNK#'].get))
pprint(inputTags)
writeout(
inputTokens, inputTags,
'/Users/aditya/Desktop/Machine Learning/Project/MLProject/EN/EN/output'
)
def simpleSentimentAnalysis(train, devin, devout):
traindata = parseTrainFile(train)
tokens = traindata[0]
tags = traindata[1]
emissionParams = calculateEmission(tags, tokens, 3)
inputTokens = parseFileInput(devin)
predictedTags = simple(inputTokens, emissionParams)
writeout(inputTokens, predictedTags, devout)
def maxMarginalSentimentAnalysis(train, devin, devout):
traindata = parseTrainFile(train)
tokens = traindata[0]
tags = traindata[1]
emissionParams = calculateEmission(tags, tokens, 3)
transitionParams = calculateTransition(tags)
inputTweets = parseMMInput(devin)
predictedTags = []
for tweet in inputTweets:
predictedTag = maxMarginal(tweet, transitionParams, emissionParams)
for t in predictedTag:
predictedTags.append(t)
predictedTags.append(None)
writeout(parseFileInput(devin), predictedTags, devout)
def improvedMaxMarginalSentimentAnalysis(train, devin, devout):
"""
Improve the parsing of train data and input data and update model parameters (Refer to report)
Using max-marginal decoding algorithm to predict the most likely tag sequence for the given tokens
"""
traindata = parseTrainP5(train)
tokens = traindata[0]
tags = traindata[1]
emissionParams = calculateEmission(tags, tokens, 3)
transitionParams = calculateTransition(tags)
inputTweets = parseInputP5(devin)
for i in range(0, 1):
predictedTags = []
emissionParam = calculateEmission(tags, tokens, 3)
transitionParam = calculateTransition(tags)
for tweet in inputTweets:
predictedTag = maxMarginal(tweet, transitionParam, emissionParam)
for t in predictedTag:
predictedTags.append(t)
predictedTags.append(None)
for word in tweet:
tokens.append(word)
tokens.append(None)
for tag in predictedTags:
tags.append(tag)
predictedTags = []
for tweet in inputTweets:
predictedTag = maxMarginal(tweet, transitionParams, emissionParams)
for t in predictedTag:
predictedTags.append(t)
predictedTags.append(None)
writeout(parseFileInput(devin), predictedTags, devout)