class RelevantRespGenerator:
# a constant controls the number of relevant responses
RESPONSE_NUM = 3
def __init__(self, brown_ic, spotify_client):
self.brown_ic = brown_ic
self.client = spotify_client
def create_phrase_measurer(self, search_query):
if Config.s_measurer == 'levenshtein':
self.create_levenshtein_measurer(search_query)
elif Config.s_measurer == 'naive':
self.create_naive_measurer(search_query)
elif Config.s_measurer == 'semantic':
self.create_semantic_measurer(search_query)
else:
raise ValueError("Only levenshtein, naive and semantic are allowed for measurer")
# Create the Levenshtein Distance to measure the similarity between search and response phrases
def create_levenshtein_measurer(self, search_query):
self.measurer = LevenPhraseSimMeasurer(search_query)
# Create the Semantic Similarity measurer to measure the similarity between search and response phrases
def create_semantic_measurer(self, search_query):
self.measurer = NLPPhraseSimMeasurer(self.brown_ic, search_query)
# Create a naive identical words measurer to measure similarity between search and response phrases
def create_naive_measurer(self, search_query):
self.measurer = NaivePhraseSimMeasurer(search_query)
# Generate the most relevant responses from Spotify with given search_query
def generate_response(self, search_query):
search_query = search_query.lower()
# Get a list of Track objects which are the response tracks from Spotify
playlist = self.get_playlist(search_query)
# for each response track, compute the semantic similarity between search_query and response track name
self.create_phrase_measurer(search_query)
for track in playlist:
phrase_sim = self.measurer.measure_phrase_sim(track.name)
#print track.name.encode('utf-8') + ":" + str(phrase_sim)
track.set_similarity(phrase_sim)
# sort all the Track objects and get top Track objects
playlist = sorted(playlist)
# get top RESPONSE_NUM tracks from playlist
return playlist[0 : RelevantRespGenerator.RESPONSE_NUM]
def get_playlist(self, search_query):
return self.client.get_playlist(search_query)
def create_semantic_measurer(self, search_query):
self.measurer = NLPPhraseSimMeasurer(self.brown_ic, search_query)
def create_naive_measurer(self, search_query):
self.measurer = NaivePhraseSimMeasurer(search_query)
def create_levenshtein_measurer(self, search_query):
self.measurer = LevenPhraseSimMeasurer(search_query)
