Algorithm to get the data related to words and their relationship with other words

Algorithm_Get_Words_Metadata.MD

Calculate Negativity and positivity through word relationships

significative relationships = word weight > 500

if word "que" has "significative relationships" with word la" then is negative

Example of Negative Texts

• NO QUIERO QUE SEA LA ULTIMA PELICULA DE JOHNNY DEEP EN PIRATAS DEL CARIBE

• yo creo y lo digo mui en lo personal que la 4 fue la mas aburrida pues faltaron personajes que hacen de estas peliculas mas interesantes para mi estubo mejor la 3 y eso que soy bien fan de piratas del caribe

Test Model Negative Positive:

{
	word: "que"
	relationships: [
		{word: "la", weight: 1},
		{word: "hacen", weight: 1},
		{word: "soy", weight: 1},
	]
}

Test Model Positive:

{
	word: "que"
	relationships: [
		{word: "la", weight: 3221},
		{word: "hacen", weight: 23},
		{word: "soy", weight: 2655},
	]
}

Algorithm to get the data related to words and their relationship with other words

def add_word(big_data_relation, word, previous_word, next_word):
	is_new_word = True

	relationship_words = []
	if previous_word:
		relationship_words.append(previous_word)

	if next_word:
		relationship_words.append(next_word)

	# Loop all the list of words and if it find the word in the list then add its relationships to it
	for i in range(0, len(big_data_relation)):
		if big_data_relation[i]['parent_word'] == word:
			is_new_word = False
			is_new_relationship = True
			for relationship_word in relationship_words:

				for j in range(0, len(big_data_relation[i]['relationships'])):
					if big_data_relation[i]['relationships'][j]['word'] == relationship_word:
						is_new_relationship = False
						big_data_relation[i]['relationships'][j]['weight'] += 1

				if is_new_relationship:
					relationship_model = {'word': "", 'weight': 0}
					relationship_model['word'] = relationship_word
					relationship_model['weight'] = 1
					big_data_relation[i]['relationships'].append(relationship_model)

	if is_new_word:
		word_model = {'parent_word': "", 'relationships': []}
		word_model['parent_word'] = word
		if previous_word:
			relationship_model = {'word': "", 'weight': 0}
			relationship_model['word'] = previous_word
			relationship_model['weight'] = 1
			word_model['relationships'].append(relationship_model)
		if next_word:
			relationship_model = {'word': "", 'weight': 0}
			relationship_model['word'] = next_word
			relationship_model['weight'] = 1
			word_model['relationships'].append(relationship_model)

		big_data_relation.append(word_model)

def format_text(text):
	return text.lower().replace('.', '').replace(',', '').replace('?', '').replace('!', '')

# This should be trained with a positive or negative list of text when the big_data_relations is created then
# you should have the relationships and the words for that polarity, for example, if you have input a lot of negative
# texts your big_data_relation would have a list of the words and relationships for negative words.

big_data_relation = []

while True:
	text = raw_input('Write a text: ')
	list_of_words = format_text(text).split(' ')
	n = len(list_of_words)
	for i in range(0, n):
		current_word = list_of_words[i]
		previous_word = list_of_words[i-1] if i > 0 else None
		next_word = list_of_words[i+1] if i < n - 1 else None
		add_word(big_data_relation, list_of_words[i], previous_word, next_word)

	print "----------\n%s\n---------" % big_data_relation