class BuildGenreCountry(object):
def __init__(self, titles):
self.titles = titles
self.graph = Graph("Graph for relationship among countries and genres")
self.distinct_genre = {}
self.genre_count = 1
def build_graph(self):
for title in self.titles:
links = []
genres = title.listed_in.split(", ")
for genre in genres:
node_link = Node(genre, Constants.GENRE_PREFIX_LABEL)
node_link.id = self._get_genre_id(genre)
link = Link(node_link)
links.append(link)
#create nodes (countries)
title_countries = title.country.split(", ")
for country in title_countries:
if country:
if country.endswith(','):
country = country[:-1]
new_node = Node(country, Constants.COUNTRY_PREFIX_LABEL)
new_node.add_link(links)
self.graph.add_node_withid_and_merge_links(new_node)
def most_common_genre_by_country(self, country_id):
country_node = next(
(e for e in self.graph.nodes if e.id == country_id), None)
if country_node:
return max(country_node.links, key=attrgetter('weight'))
else:
return None
def _get_genre_id(self, genre):
genre_id = ''
for key in self.distinct_genre:
if self.distinct_genre[key] == genre:
genre_id = genre
if genre_id == '':
self.distinct_genre[self.genre_count] = genre
genre_id = genre
self.genre_count += 1
return '{0}{1}'.format(Constants.GENRE_PREFIX_LABEL, genre_id)
class BuildTitleSimilarityGraph(object):
def __init__(self, titles):
self.titles = titles
self.graph = Graph("Graph for relationship titles based on description text")
def build_graph(self):
for title in self.titles:
links = []
title_node = Node(title.show_id, Constants.TITLE_PREFIX_LABEL, '{0}{1}'.format(Constants.TITLE_PREFIX_LABEL, title.title))
for link_title in self.titles:
if link_title.show_id == title.show_id:
continue
similarity_coeff = BuildTitleSimilarityGraph._jaccard_similarity_text(title.description, link_title.description)
if Constants.MIN_SIMILARITY_COEFFICIENT <= similarity_coeff:
weight = int(similarity_coeff * 100)
link_node = Node(link_title.show_id, Constants.TITLE_PREFIX_LABEL, '{0}{1}'.format(Constants.TITLE_PREFIX_LABEL, link_title.title))
link = Link(link_node, weight)
links.append(link)
title_node.add_link(links)
self.graph.add_node(title_node)
def get_most_similar(self, title_id):
most_similar = next((e for e in self.graph.nodes if e.id == title_id), None)
if most_similar:
return max(most_similar.links, key = attrgetter('weight'))
else:
return None
@staticmethod
def _jaccard_similarity_text(str1, str2):
a = set(str1.lower().split())
b = set(str2.lower().split())
c = a.intersection(b)
return float(len(c)) / (len(a) + len(b) - len(c))
class BuildActorGenreGraph(object):
def __init__(self, titles):
self.titles = titles
self.graph = Graph("Graph for relationship among actors and genres")
self.distinct_genre = {}
self.genre_count = 1
def build_graph(self):
for title in self.titles:
links = []
genres = title.listed_in.split(", ")
for genre in genres:
node_link = Node(genre, Constants.GENRE_PREFIX_LABEL)
node_link.id = self._get_genre_id(genre)
link = Link(node_link)
links.append(link)
#create nodes (actors)
actors = title.cast.split(", ")
for actor in actors:
if actor:
if actor.endswith(','):
actor = actor[:-1]
new_node = Node(actor, Constants.ACTOR_PREFIX_LABEL)
new_node.add_link(links)
self.graph.add_node_withid_and_merge_links(new_node)
def _get_genre_id(self, genre):
genre_id = ''
for key in self.distinct_genre:
if self.distinct_genre[key] == genre:
genre_id = genre
if genre_id == '':
self.distinct_genre[self.genre_count] = genre
genre_id = genre
self.genre_count += 1
return '{0}{1}'.format(Constants.GENRE_PREFIX_LABEL, genre_id)
class BuildGenreYearGraph(object):
def __init__(self, titles):
self.titles = titles
self.graph = Graph("Graph for relationship among genres and years")
self.distinct_genre = {}
self.genre_count = 1
def build_graph(self):
for title in self.titles:
links = []
# Building the links
genres = title.listed_in.split(", ")
for genre in genres:
node_link = Node(genre, Constants.GENRE_PREFIX_LABEL)
node_link.id = self._get_genre_id(genre)
link = Link(node_link)
links.append(link)
year = title.release_year
year_node = Node(year, Constants.YEAR_PREFIX_LABEL)
year_node.add_link(links)
self.graph.add_node_withid_and_merge_links(year_node)
def _get_genre_id(self, genre):
genre_id = ''
for key in self.distinct_genre:
if self.distinct_genre[key] == genre:
genre_id = genre
if genre_id == '':
self.distinct_genre[self.genre_count] = genre
genre_id = genre
self.genre_count += 1
return '{0}{1}'.format(Constants.GENRE_PREFIX_LABEL, genre)
def __init__(self, titles):
self.titles = titles
self.graph = Graph("Graph for relationship among countries and genres")
self.distinct_genre = {}
self.genre_count = 1
def __init__(self, titles):
self.titles = titles
self.graph = Graph("Graph for relationship titles based on description text")