def narrowGraphTo10MainCharacters(G, speakers):
mainCharacters = []
speakersDictionary = {}
#Putting all the speakers in a dictionary
for speaker in speakers:
if speaker in speakersDictionary:
speakersDictionary[speaker] = speakersDictionary[speaker] + 1
else:
speakersDictionary[speaker] = 1
#sorting in order to get the main characters
sortedSpeakers = sorted(speakersDictionary.items(), key=lambda kv: kv[1],reverse=True)
mainCharactersNumber=10
for i in range(mainCharactersNumber):
mainCharacters.append(sortedSpeakers[i][0])
print()
G3 = nx.Graph()
for u, v, d in G.edges(data=True):
if (u in mainCharacters and v in mainCharacters):
G3.add_edge(u, v, weight=d['weight'])
return G3
def voronoi2(G,movie,isDirected):
global color_map
if isDirected==0:
G2 = nx.Graph()
else:
G2 = nx.DiGraph()
#Converting frequencies to distances
for u, v, d in G.edges(data=True):
G2.add_edge(u, v, weight=(1 / d['weight']))
# Choosing 2 Ankers and as a result -> Dividing the Graph to 2 partitions
if movie == 'Captain America - Civil War':
anker1 = 'STEVE ROGERS'
anker2 = ['TONY STARK', 'NATASHA ROMANOFF']
else:
anker1 = 'WAYNE'
anker2 = ['BANE', 'SELINA']
color_map = []
for v in G2:
# print('v=' +str(v)," to anker:",anker1,'|',spw(G2, v, anker1)," to anker:",anker2,'|',spw(G2, v, anker2))
if (v == anker1):
color_map.append('blue')
elif (v in anker2):
color_map.append('red')
elif (spw(G2, v, anker1) < spw(G2, v, anker2[0]) and spw(G2, v, anker1) < spw(G2, v, anker2[1])):
color_map.append('blue')
else:
color_map.append('red')
return G2
def __init__(self, dimension: int, graph: nx.Graph = None, id: str = ''):
"""
Initializes the NetworkX graph representation of
intersecting and overlapping Regions.
Args:
dimension:
The number of dimensions in all of the Regions
within the graph; the dimensionality of all Regions.
graph:
The internal NetworkX graph representation or
implementation for a graph of intersecting or
overlapping Regions.
id:
The unique identifier for this RIGraph.
Randonly generated with UUID v4, if not provided.
"""
assert isinstance(dimension, int) and dimension > 0
assert graph == None or isinstance(graph, nx.Graph)
self.dimension = dimension
self.id = id = id if len(id) > 0 else str(uuid4())
if graph == None:
self.G = nx.Graph(id=id, dimension=dimension)
else:
self.G = G = graph
self.dimension = G.graph.setdefault('dimension', dimension)
self.id = G.graph.setdefault('id', id)
def __init__(self, dimension: int, graph: nx.Graph = None, id: str = ''):
"""
Initializes the NetworkX graph representation of
intersecting and overlapping Regions.
Args:
dimension:
The number of dimensions in all of the Regions
within the graph; the dimensionality of all Regions.
graph:
The internal NetworkX graph representation or
implementation for a graph of intersecting or
overlapping Regions.
id:
The unique identifier for this RIGraph.
Randonly generated with UUID v4, if not provided.
"""
assert dimension > 0
# TODO maybe add 'assert all(isinstance( self.G.nodes(... , Region)))' ?
self.dimension = dimension
self.id = id if len(id) > 0 else str(uuid4())
self.G = graph if graph is not None else nx.Graph()
def buildGraphFromListUndirectedUnweighted(speakers):
edges = []
G = nx.Graph()
for i in range(0, len(speakers) - 2):
if(speakers[i]!=speakers[i+1]):
G.add_edge(speakers[i], speakers[i + 1], weight=1)
return G
def buildGraphFromListUndirectedWeighted(speakers):
edges = []
G = nx.Graph()
for i in range(0, len(speakers) - 2):
if(speakers[i]!=speakers[i+1]):
if (G.has_edge(speakers[i], speakers[i + 1])):
G[speakers[i]][speakers[i + 1]]['weight'] += 1
else:
G.add_edge(speakers[i], speakers[i + 1], weight=1)
return G
def addGraph(fname, toOutput = 2):
G = nx.Graph()
with open(fname, "r") as f1:
for line in f1:
tmp = line.strip().split(" ")
G.add_edge(tmp[0], tmp[1])
fout = open("file"+str(file_id)+"_out4", "w")
for g in nx.connected_components(G):
if toOutput!=2:
print g
else:
fout.write(g.__str__()+"\n")
fout.close()
return G
def data_clust(self, origin, nodes, links):
#print("links",links)
edge_list = [tuple(elem.values()) for elem in links]
g = nx.Graph()
g.add_edges_from(edge_list)
#clust_coefficients = nx.clustering(g)
partition = community.best_partition(g)
clustered_nodes = nodes
part_set = set()
number_of_clusters = 0
for i, node in enumerate(clustered_nodes):
node['cluster'] = partition[i]
if partition[i] not in part_set:
part_set.add(partition[i])
number_of_clusters += 1
return clustered_nodes, links, number_of_clusters
def grafica():
G = nx.Graph()
G.add_edges_from([(1, 2), (1, 3)])
def votingActualAlgo(G, firstAnker, secondAnker, movieName):
#(i,j) are vertixes we anchored
#drawGraph(G,0)
directedG = nx.Graph().to_directed()# from ex1 we already have directed graph
sum = 0
allEdgesWeight = {}
#For each v we sum all the edges wehight
for v in G:
vNeighbors = neighbors(G, v)
for u in vNeighbors:
sum += G[v][u]['weight']
allEdgesWeight[v] = sum
sum = 0
#print('list of main characters')
#print(allEdgesWeight.keys())
m = nx.convert_matrix.to_numpy_matrix(G)
#print('The graph converted to matrix')
#print(m)
#divide each directed egde in the sum of edges weight from vertex V (- normelaize the matrix)
for k in range(m.shape[0]):
#print(m[k].sum())
m[k]/=m[k].sum()
#print('Normalized M - directed(!) - weight of the edge is divided by the sum of the edge from this vertex')
#print(m)
names = list(allEdgesWeight.keys())
i= names.index(firstAnker)
j=names.index(secondAnker)
#print('\n\n\n')
#Setting Ankers - "Eater Vertex"
m[i]=0
m[i,i]=1
m[j] = 0
m[j,j] = 1
#print('VOTING - MATRIX WITH ANKORS!\n\n', m)
#print('Multipyng\n\n')
for k in range (0,4):
m=np.dot(m,m)
#print ('Mult -',k,'- :\n',m)
list_for_return = list()
ank1 = list()#anckor 1 full list with 0's
ank2 = list()#anckor 2 full list with 0's
ank1_list = list() # list without 0's
ank2_list = list() # list without 0's
for k in range (0,10):
if m[k, i] >= m[k, j]:
ank1.append(m[k, i])
ank1_list.append(m[k, i])
ank2.append(0)
else:
ank2.append(m[k, j])
ank2_list.append(m[k, j])
ank1.append(0)
print('list of groups for the movie',movieName)
names=list(allEdgesWeight.keys())
groupI1=[]
groupJ2=[]
for k in range(len(names)):
if ank1[k]>ank2[k]:
groupI1.append(names[k])
else:
groupJ2.append(names[k])
print(names[i],'=',groupI1,' | ',names[j],'=',groupJ2) #1~i 2~J
# print(ank1)
# print(ank2)
print('\n')
#
# for u, v, d in G.edges(data=True):
# directedG.add_edge(u, v, weight=d['weight'])
# directedG.add_edge(v, u, weight=d['weight'])
#
# for v in directedG:
# vSize = allEdgesWeight[v]
# for u in directedG:
# if(directedG.has_edge(v,u)):
# directedG[v][u]['weight'] = directedG[v][u]['weight']/vSize
#
list_for_return.append(directedG)
list_for_return.append(ank1_list)
list_for_return.append(ank2_list)
list_for_return.append(ank1)
list_for_return.append(ank2)
return list_for_return
#!/usr/bin/env python
# coding: utf-8
from networkx import networkx as nx
from itertools import combinations
from collections import defaultdict
import matplotlib.pyplot as plt
import scipy
import os
import json
import database_handler
#from data_reader import get_stackoverflow_data
BASE_PATH = 'data'
G = nx.Graph()
#xml_dict = get_stackoverflow_data(f"{BASE_PATH}/bitcoin")
# example on how to use the xml
#for row in xml_dict['Users'][:2]:
# # row objects have tag & attrib
# print(row.tag, row.attrib)
# Extraction of features into individual json files for each directory
database_handler.prepare_json_databases(BASE_PATH)
G.add_nodes_from(os.listdir(BASE_PATH))
edges = []
# Each directory represents a site
