Beispiel #1
0
def find_match_pair(L1, L2, L1_translation, L2_translation, Greek_search_dict, Latin_text, Latin_word_num, Greek_text, Greek_word_num, max_distance_greek = None):
	maxScore = None
	currScore = 0
	bestMatch = None

	# G1 and G2 each represent an array of all the locations where a matched word appears in the corresponding document
	G1_translation_array, G2_translation_array = get_GreekPair(L1_translation, L2_translation, Greek_search_dict)

	# If either of the translation arrays are empty, return a None value
	if G1_translation_array == None or G2_translation_array == None:
		return None

	# Iterates through all possible matches for the translations of L1 and L2
	for G1_pos in G1_translation_array:
		for G2_pos in G2_translation_array:

			# Makes sure that distance between two greek words is constrained
			if max_distance_greek == None or (abs(G2_pos - G1_pos) <= max_distance_greek):
				currScore = calc_match_pair_score(L1.pos, L1.occurences, L2.pos, L2.occurences, Latin_word_num, G1_pos, len(G1_translation_array), G2_pos, len(G2_translation_array), Greek_word_num)

				if (maxScore == None) or (currScore > maxScore):
					maxScore = currScore
					bestMatch = xling_match(L1.pos, L2.pos, Latin_text, G1_pos, G2_pos, Greek_text, maxScore)
			else:
				continue

	if (bestMatch == None):
		return None
	else:
		return bestMatch
Beispiel #2
0
def calc_match_phrase_score(search, translated_indices, scoreKeeper, Greek_word_num, Greek_search_dict, Greek_text):
	sum_of_word_occurences = 0

	f_g = 0
	for word_index in translated_indices:
		num_word_occurences = float(len(Greek_search_dict[Greek_text[word_index]]))
		f_g = + float(Greek_word_num / num_word_occurences)

	if (f_g <= 0):
		return

	max_d_g = 0

	# Length is calculated by subtracting pos of first word from pos of last word
	for i in range(len(translated_indices)):
		for j in range(len(translated_indices)):

			d_g = abs(translated_indices[len(translated_indices) - 1] - translated_indices[0])

			if (d_g > max_d_g):
				max_d_g = d_g

	# Case where multiple words map to same word
	if (max_d_g == 0):
		max_d_g = 1

	# f_l is 1 since all the latin words we have are contained in the search phrase
	f_l = 1

	# d_l is search_phrase.len
	d_l = search.search_len

	score = (f_l + f_g) / (d_l + max_d_g)

	curr_match = xling_match(0, search.search_len - 1, search.text, translated_indices[0], translated_indices[len(translated_indices) - 1], Greek_text, score)

	scoreKeeper.add_newMatch(curr_match)

	return