class WebPageSimilarity:
def __init__(self):
self.structuralSimilarity = StructuralSimilarity()
self.contentSimilarity = ContentSimilarity()
#self.visualSimilarity = VisualSimilarity()
self.linkSimilarity = LinkSimilarity()
def calculate_similarity(self, url1, url2):
similarity_dict = {}
similarity_dict['content_similarity'] = self.contentSimilarity.main(url1, url2)
similarity_dict['link_similarity'] = self.linkSimilarity.main(url1, url2)
#similarity_dict['visual_similarity'] = self.visualSimilarity(url1, url2)
similarity_dict['structural_similarity'] = self.structuralSimilarity.main(url1, url2)
return similarity_dict
def calculate_weights(self, listOfCalculatedSimilarities, listOfActualSimilarities):
a = np.array(listOfCalculatedSimilarities)
b = np.array(listOfActualSimilarities)
result = np.linalg.lstsq(a, b)
return result
def __init__(self):
self.structuralSimilarity = StructuralSimilarity()
self.contentSimilarity = ContentSimilarity()
#self.visualSimilarity = VisualSimilarity()
self.linkSimilarity = LinkSimilarity()