Method nearest_neighbors

pattern/vector/__init__.py:1047–1056

Returns a list of (similarity, document)-tuples in the model, sorted by cosine similarity to the given document.

(self, document, top=10)

    similarity = cosine_similarity
    
    def nearest_neighbors(self, document, top=10):
        """ Returns a list of (similarity, document)-tuples in the model, 
            sorted by cosine similarity to the given document.
        """
        v = ((self.cosine_similarity(document, d), d) for d in self.documents)
        # Filter the input document from the matches.
        # Filter documents that score zero, and return the top.
        v = [(w, d) for w, d in v if w > 0 and d.id != document.id]
        v = heapq.nsmallest(top, v, key=lambda v: (-v[0],v[1]))
        return v
        
    similar = related = neighbors = nn = nearest_neighbors