IMPORTANT: To use this notebook, you'll need to
- Install IPython Notebook (easiest way: use Anaconda)
- Download this notebook and all other Python scripts used here from https://github.com/mboudour/WordNets/blob/master/EUSN2016_LiteraryTextNetworksWorkshop
- Run
ipython notebookin the same directory where notebook and scripts were put
This work is licensed under a Creative Commons Attribution 4.0 International License.
Importing Python modules¶
In [1]:
import random
import nltk
import codecs
from textblob import TextBlob
import networkx as nx
import matplotlib.pyplot as plt
import pandas as pd
import os
import imp
import seaborn as sns # pip install seaborn
sns.set_style("white") # For seaborn to show axes in iPython Notebook
from lightning import Lightning
from numpy import random, asarray, sqrt, arctan2, pi, clip
from seaborn import color_palette
from sklearn import datasets
from colorsys import hsv_to_rgb
# utilsdir='/Users/mosesboudourides/Dropbox/Python Projects/EUSN2016_LiteraryTextNetworksWorkshop/utils/'
# utilsdir='/home/mab/Dropbox/Python Projects/utils/'#tools.py'
utilsdir='/home/mab/Dropbox/Python Projects/EUSN2016_LiteraryTextNetworksWorkshop/utils/'
%matplotlib inline
%load_ext autoreload
I. Importing the Text & Names of Characters¶
In [2]:
filename = 'texts/ABronte_AgnesGrey.txt'
titlename = "Anne Bronte's Agnes Grey"
central_hero = 'Nancy Brown'
online_list_of_characters = 'https://en.wikipedia.org/wiki/Agnes_Grey'
nn1 = "Anne Bronte's Agnes Grey selected terms"
nn2 = "Anne Bronte's Agnes Grey Characters"
vname1 = 'vids/AgnesGrey.gif'
vname2 = 'vids/AgnesGrey.mp4'
nn3 = "['Nancy Brown']"
filename1 = 'S_out_graphs/Nancy Brown_graph.graphml'
# https://en.wikipedia.org/wiki/Agnes_Grey
# Agnes Grey
# Edward Weston
# Richard Grey
# Alice Grey
# Mary Grey
# Mrs. Bloomfield
# Mr. Bloomfield
# Matilda Murray
# Rosalie Ashby (formerly Rosalie Murray)
# John Murray
# Charles Murray
# Mr. Murray
# Mrs. Murray
# Lord Thomas Ashby
# Mr. Hatfield
# Nancy Brown
# Tom Bloomfield
# Mr. Richardson
dici={'Agnes Grey':'Agnes Grey', 'Agnes':'Agnes Grey',
'Edward Weston':'Edward Weston','Edward':'Edward Weston','Weston':'Edward Weston',
'Richard Grey':'Richard Grey','Richard':'Richard Grey',
'Alice Grey':'Alice Grey','Alice':'Alice Grey',
'Mary Grey':'Mary Grey','Mary':'Mary Grey',
'Mrs. Bloomfield':'Mrs. Bloomfield',
'Mr. Bloomfield':'Mr. Bloomfield',
'Matilda Murray':'Matilda Murray','Matilda':'Matilda Murray',
'Rosalie Ashby':'Rosalie Ashby','Rosalie':'Rosalie Ashby',
'John Murray':'John Murray', 'John':'John Murray',
'Charles Murray':'Charles Murray','Charles':'Charles Murray',
'Mr. Murray':'Mr. Murray',
'Mrs. Murray':'Mrs. Murray',
'Thomas Ashby':'Lord Thomas Ashby','Thomas':'Lord Thomas Ashby',
'Mr. Hatfield':'Mr. Hatfield','Hatfield':'Mr. Hatfield',
'Nancy Brown':'Nancy Brown','Nancy':'Nancy Brown','Brown':'Nancy Brown',
'Tom Bloomfield':'Tom Bloomfield','Tom':'Tom Bloomfield','Bloomfield':'Tom Bloomfield',
'Mr. Richardson':'Mr. Richardson','Richardson':'Mr. Richardson'}
In [3]:
f = codecs.open(filename, "r", encoding="utf-8").read()
num_lines = 0
num_words = 0
num_chars = 0
for line in f:
words = line.split()
num_lines += 1
num_words += len(words)
num_chars += len(line)
print "%s has number of words = %i and number of characters = %i" %(titlename,num_words,num_chars)
blob = TextBlob(f)
ndici={i.lower():k for i,k in dici.items()}
dnici=[(i.split()[0],i.split()[1]) for i in ndici.keys() if len(i.split())>1]
selectedTerms=ndici.keys()
In [4]:
%autoreload 2
tool= imp.load_source('tools', utilsdir+'tools.py')
# tool= imp.load_source('tools', '/Users/mosesboudourides/Dropbox/Python Projects/utils/tools.py')
# import tools as tool
create_pandas_dataframe_from_text=tool.create_pandas_dataframe_from_text
create_coo_graph=tool.create_coo_graph
dfst,sec_prot,coccurlist,occurlist,dflines=create_pandas_dataframe_from_text(blob,selectedTerms,ndici,titlename)
co_graph=create_coo_graph(coccurlist)
dfst.rename(columns={nn1:nn2},inplace=True)
# dfst.rename(columns={"Anne Bronte's Agnes Grey selected terms":"Anne Bronte's Agnes Grey Characters"},inplace=True)
dfst.sort_values(by='Frequencies').sort(["Frequencies"], ascending=[0])
Out[4]:
In [5]:
prot_pol_sub=dflines[['protagonists','#_of_protagonists','polarity','subjectivity']].reset_index()
prot_pol_sub['sentence_id']=prot_pol_sub.index
prot_pol_sub=prot_pol_sub[['sentence_id','protagonists','#_of_protagonists','polarity','subjectivity']]
cuts = 0
# prot_pol_sub = prot_pol_sub[prot_pol_sub['#_of_protagonists']>=cuts]
lp = prot_pol_sub['protagonists'].tolist()
lpn = []
for i in lp:
for j in i:
lpn.append(j)
# len(set(lpn))
print "The total number of sentences in %s is %i." %(titlename,len(prot_pol_sub))
# print "The total number of sentences in %s with at least %i characters in each one of them is %i." %(titlename,cuts+1,len(prot_pol_sub))
prot_pol_sub.rename(columns={'protagonists':'Lists_of_Characters','#_of_protagonists':'#_of_Characters','polarity':'Polarity','subjectivity':'Subjectivity'},inplace=True)
prot_pol_sub.sort(["#_of_Characters"], ascending=[0])
ddff = prot_pol_sub.drop('sentence_id', 1)
ddff.index.name = 'Sentence_ID'
ddff.head(10)
Out[5]:
II. Descriptive Statistics of Characters¶
In [6]:
nddd=ddff[ddff['Polarity'] !=0 ]#& ddff['Subjectivity'] !=0]
nddd=nddd[nddd['Subjectivity'] !=0]
# ddff=nddd
ddff[['#_of_Characters','Polarity','Subjectivity']].describe()
Out[6]:
In [7]:
corrmat = ddff.corr()
f, ax = plt.subplots(figsize=(12, 9))
sns.heatmap(corrmat, vmax=.8, square=True, annot=True)
Out[7]:
In [8]:
from pandas.tools.plotting import scatter_matrix
ntei='Scatter Matrix Plot of ' + titlename
f, ax = plt.subplots(figsize=(12,12))
# nddd
sss=scatter_matrix(ddff[['#_of_Characters','Polarity','Subjectivity']], alpha=0.9, color='black', diagonal='hist',ax=ax)
plt.suptitle(ntei,fontsize=18,fontweight='bold')
corr = ddff.corr().as_matrix() #nddd.corr().as_matrix()
for i, j in zip(*plt.np.triu_indices_from(sss, k=1)):
sss[i, j].annotate("pearson = %.3f" %corr[i,j], (0.8, 0.93), xycoords='axes fraction', ha='center', va='center')
The Histogram of the Number of Characters¶
In [9]:
from mpl_toolkits.axes_grid1.inset_locator import zoomed_inset_axes
from mpl_toolkits.axes_grid1.inset_locator import mark_inset
ndfl=dflines[dflines['#_of_protagonists']>0 ]
fig, ax = plt.subplots(figsize=[12, 10])
axes2 = zoomed_inset_axes(ax, 16, loc=7) # zoom = 6
dflines['#_of_protagonists'].plot.hist(ax=ax)
ax.set_xlabel('#_of_Characters')
ax.set_ylabel('Frequency')
ax.set_title('Histogram of # of characters')
x1, x2, y1, y2 = 2.95, 3., 0, 30
axes2.set_xlim(x1, x2)
axes2.set_ylim(y1, y2)
ndfl['#_of_protagonists'].plot.hist(ax=axes2)
axes2.set_ylabel('Frequency')
mark_inset(ax, axes2, loc1=2, loc2=4, fc="none", ec="0.5")
axes3 = zoomed_inset_axes(ax, 10, loc=10)
x1, x2, y1, y2 = 2, 2.1, 0, 60
axes3.set_xlim(x1, x2)
axes3.set_ylim(y1, y2)
ndfl['#_of_protagonists'].plot.hist(ax=axes3)
axes3.set_ylabel('Frequency')
mark_inset(ax, axes3, loc1=2, loc2=4, fc="none", ec="0.5")
plt.show()
Various Scatter Plots¶
In [10]:
x = nddd['Polarity']
y = nddd['Subjectivity']
z = nddd['#_of_Characters']
In [11]:
lgn = Lightning(ipython=True, host='http://public.lightning-viz.org')
Lightning initializedIn [12]:
series = [x,y]
lgn.line(series)
Out[12]:
In [13]:
viz = lgn.scatter(x, y, values = z, alpha=0.6, colormap='YlOrRd')
viz
Out[13]:
In [14]:
from ggplot import *
import matplotlib as mpl
ntei='Scatter Plot of ' + titlename
p = ggplot(aes(x='Polarity', y='Subjectivity',color='#_of_Characters'), data=ddff) #nddd)
p + geom_point() + ggtitle(ntei) + theme_matplotlib(rc={"figure.figsize": "12, 9"}, matplotlib_defaults=False)
Out[14]:
In [15]:
ntei=' The Sentiment Space of ' + titlename
f, ax = plt.subplots(figsize=(12,9))
ddff.plot.hexbin(x='Polarity',y='Subjectivity',gridsize=20,C='#_of_Characters',ax=ax,reduce_C_function=max,cmap='jet')#,title=ntei)
plt.xlim(-1.1, 1.1)
plt.ylim(-.1, 1.1)
plt.suptitle(ntei,fontsize=15,fontweight='bold')
Out[15]:
In [16]:
ntei='KDE Joint Plot of the Sentiment Space of ' + titlename
# f, ax = plt.subplots(figsize=(10,10))
# cmap = sns.cubehelix_palette(light=1, as_cmap=True)
# cmap = sns.cubehelix_palette(rot=-.4, as_cmap=True)
cmap = sns.cubehelix_palette(8, start=.5, rot=-.75, as_cmap=True)
ggn=sns.jointplot(x='Polarity',y='Subjectivity', data=nddd, kind ="kde",cmap=cmap,space=0, size=10) #, ax=ax) #kind="kde",
# sns.kdeplot(nddd['Polarity'],nddd['Subjectivity'], cmap=cmap, shade=True)
ggn.plot_joint(plt.scatter, c="r", s=30, linewidth=1, marker="+")
ggn.ax_joint.collections[0].set_alpha(0)
ggn.set_axis_labels("Polarity", "Subjectivity")
plt.suptitle(ntei,fontsize=15,fontweight='bold')
Out[16]:
In [17]:
# pols=nddd.Polarity.tolist()
# subj=nddd.Subjectivity.tolist()
# ntei=titlename+' in Sentiment Space'
# import numpy as np
# import matplotlib.pyplot as plt
# import matplotlib.animation as animation
# fig, ax = plt.subplots()
# plt.xlim(-1.1, 1.1)
# plt.ylim(-.1, 1.1)
# ax.set_xlabel('Polarity')
# ax.set_ylabel('Subjectivity')
# # arro=ax.arrow(pols[0], subj[0], pols[1]-pols[0], subj[1]-subj[0], head_width=0.03, head_length=0, fc='b', ec='b',
# # length_includes_head=False,
# # # # head_starts_at_zero=True
# # # # overhang=-.51
# # fill=False)
# def animate(i):
# col=(1.*i/(1.*len(pols)),.5,.5)
# plt.plot(pols[i],subj[i],'o',color=col, markersize=5)
# # return arro,
# ani = animation.FuncAnimation(fig, animate, np.arange(0, len(pols)-1),
# interval=25, blit=False)
# # ani.save(vname2)#, metadata={'artist':'Guido'})
# ani.save(vname1, writer='imagemagick')
# plt.show()
# print len(pols)
In [18]:
# %%bash
# ffmpeg -f gif -i vids/AgnesGrey.gif vids/AgnesGrey.mp4
In [19]:
import io
import base64
from IPython.display import HTML
video = io.open(vname2, 'r+b').read()
encoded = base64.b64encode(video)
HTML(data='''<video alt="test" controls>
<source src="data:video/mp4;base64,{0}" type="video/mp4" />
</video>'''.format(encoded.decode('ascii')))
# /Users/mosesboudourides/Dropbox/Python Projects/LiteratureNetworks/ArthurConanDoyle/SherlockHolmesStoriesNetwork/
Out[19]:
III. The Two-Mode Network of Characters and Sentences¶
In [20]:
%autoreload 2
# from tools import draw_network_node_color
sstt="%s Two-Mode Network of Sentences and Characters" %titlename
pos=nx.spring_layout(sec_prot)
nds=[nd for nd in sec_prot.nodes() if isinstance(nd,int)]
prot=[nd for nd in sec_prot.nodes() if nd not in nds]
for en,nd in enumerate(nds):
if en<len(nds)/2.:
pos[nd][0]=-1
pos[nd][1]=en*2./len(nds)
else:
pos[nd][0]=1
pos[nd][1]=(en-len(nds)/2.)*2./len(nds)
for en ,nd in enumerate(prot):
pos[nd][0]=0
pos[nd][1]=en*1./len(prot)
possit=tool.draw_network_node_color(sec_prot,sstt,pos=pos,with_edgewidth=False,withLabels=True,labfs=12,valpha=0.2,
ealpha=0.4,labelfont=15,with_node_weight=False,node_size_fixer=300.,node_col='polarity')
# possit=draw_network_node_color(sec_prot,sstt,pos=pos,with_edgewidth=False,withLabels=True,labfs=12,valpha=0.2,
# ealpha=0.4,labelfont=15,with_node_weight=False,node_size_fixer=300.,node_col='polarity')
In [21]:
possit=tool.draw_network_node_color(sec_prot,sstt,pos=pos,with_edgewidth=False,withLabels=True,labfs=12,valpha=0.2,
ealpha=0.4,labelfont=15,with_node_weight=False,node_size_fixer=300.,
node_col='subjectivity',colormat='Greens')
# possit=draw_network_node_color(sec_prot,sstt,pos=pos,with_edgewidth=False,withLabels=True,labfs=12,valpha=0.2,
# ealpha=0.4,labelfont=15,with_node_weight=False,node_size_fixer=300.,
# node_col='subjectivity',colormat='Greens')
IV. The Network of Sententially Co-Occurring Characters¶
In [22]:
%autoreload 2
# from tools import draw_network, make_graph_from_lists
tool= imp.load_source('tools', utilsdir+'tools.py')
plist = prot_pol_sub['Lists_of_Characters'].tolist()
pplist=prot_pol_sub['Polarity'].tolist()
nplist=prot_pol_sub['#_of_Characters'].tolist()
splist=prot_pol_sub['Subjectivity'].tolist()
G = tool.make_graph_from_lists(plist,pplist,nplist,splist)
posg=nx.spring_layout(G,scale=50,k=0.55,iterations=20)
# posg=nx.spring_layout(G,scale=50)#,k=0.55)#,iterations=20)
sstt="%s Network of Selected Characters \n(Sentences colored in polarity)" %titlename
possit=tool.draw_network(G,sstt,pos=posg,with_edgewidth=True,withLabels=True,labfs=15,valpha=0.2,ealpha=0.7,labelfont=15,
with_edgecolor=True,edgecolor='polarity',colormat='Blues')
In [23]:
sstt="%s Network of Selected Characters \n(Sentences Colored in Subjectivity)" %titlename
possit=tool.draw_network(G,sstt,pos=posg,with_edgewidth=True,withLabels=True,labfs=15,valpha=0.2,ealpha=0.7,labelfont=15,
with_edgecolor=True,edgecolor='subjectivity',colormat='Blues')
Centralities of Nodes in the Network of Sententially Co-Occurring Characters¶
In [24]:
from tools import draw_centralities_subplots
centrali=draw_centralities_subplots(G,pos=posg,withLabels=False,labfs=5,figsi=(15,22),ealpha=1,vals=True)
The table of Centralities of Nodes in the Network of Sententially Co-Occurring Characters¶
In [25]:
dfc=pd.DataFrame()
u=0
for i,k in centrali.items():
dfc.insert(u,i,k.values())
u+=1
dfc.insert(0,'Nodes',centrali[centrali.keys()[0]].keys())
dfc
Out[25]:
Communities of Nodes in the Network of Sententially Co-Occurring Characters¶
In [26]:
%autoreload 2
from tools import draw_comms, modul_arity, print_communities
part,nodper=print_communities(G,sstt)
ndfl=dflines[dflines['#_of_protagonists']>0 ]
# ndfl['#_of_protagonists'].plot.hist()
d=0.8
dd=0.8
c=1.2
cc=1.4
alpha=0.2
ealpha=0.2
vcc={}
sstta="The %s Communities of %s Network of Characters" %(max(part.values())+1,titlename)#sstt)
draw_comms(G,G.nodes(),[],[],[] ,part,part,d,dd,c,cc,alpha,ealpha,nodper,sstta,titlefont=20,labelfont=17,valpha=0.5)
V. Trajectories of Transitions among Sentential Co-Occurrences of Characters¶
More about Transitions and Trajectories in Temporal Networks:¶
http://www.slideshare.net/MosesBoudourides/trajectories1¶
In [27]:
# %autoreload 2
# !pip install --user pygexf
# !pip install --user pyinterval
# !pip install --user pyinter
# !pip install --user python-igraph
# import trajectories as trj
trj= imp.load_source('trajectories', utilsdir+'trajectories_t.py')
# import create_gexf_year as cgy
Transitions among Sentential Co-Occurrences of the Central Hero and Other Characters¶
In [28]:
def search_in_list(x):
l=x['protagonists']
return central_hero in l
ndfl=dflines[dflines['#_of_protagonists']>1 ]
dialogs=ndfl[ndfl.apply(search_in_list,axis=1) ==True]
protagonists=dialogs.protagonists.tolist()
start=range(1,len(protagonists)+1)
end=range(2,len(protagonists)+2)
polarities=dialogs.polarity.tolist()
subj=dialogs.subjectivity.tolist()
qq=0
figi=None
search_name=central_hero
G,ndls,pold,subjd=trj.creatTestGraph_pandas_bip(start,end,protagonists,search_name,polarities,subj)
trajpdfs=trj.main_work_search_name(G,ndls,qq,figi,search_name,verb=False,plot_first_mode=False)
Statistics of the Trajectory of Transitions among Sentential Co-Occurrences of the Central Hero and Other Characters¶
In [29]:
trajpdfs[nn3] #["['Agnes Grey']"]
# trajpdfs["['Sherlock Holmes']"]
Out[29]:
The Color of Nodes Corresponds to Sentential Sentiment Polarities Varying from Color Red (-1) to Color Green (1)¶
In [30]:
%autoreload 2
import igraph as ig
from tools import igraph_draw_traj
filname=filename1 #'S_out_graphs/Agnes Grey_graph.graphml'
g,visual_style,layout=igraph_draw_traj(filname,pold)
ig.plot(g, **visual_style)
Out[30]:
The Color of Nodes Corresponds to Sentential Sentiment Subjectivities Varying from Color Red (0) to Color Green (1)¶
In [31]:
g,visual_style,layout=igraph_draw_traj(filname,subjd,polar=False,layout=layout)
ig.plot(g, **visual_style)
Out[31]:
In [ ]: