Film Script Analysis -- Case Study 2 -- Antz¶

-----20 Movies Script Visualizations will be done------This could be done for the 1000+ movie script I segmented--- I just randomly chose 20 movie scripts from the segmented movies....

The aim of this project is to provide detailed insights into different movies analyzed focusing on the characters, their dialogues, scene locations, emotional and sentiment analysis of the whole movie and the individual characters, character's interaction with one another and finally gender distribution in the each movie analyzed.¶

To execute this project, the following objectives were done:¶

Web scraping of the movie scripts (Over 1000+ movies were scraped from imsdb website)
Movies segmentation into Scenes --> Scene Location, Scene Action/Description, Scene Dialogues, Scene Characters (All the movies scraped were segmented except those that do not follow the "Screenplay format i.e. INT / EXT)"
Character extraction and appearances plot ---> Here, characters were plotted based on how many times they appeared and spoke in each scene and across the movie.
Character Interaction Mapping --> We mapped out the connection between all the characters in the movie and also the interaction between the Top 10 characters in the movie.
Here, we looked at the Most mentioned character based on the Scene dialogues and also the characters each character mention the most in their conversation.
Similar to Number 5., Here looked at who a specific character talks with the most in the Movie.
Emotional and Sentiment Analysis across the whole movie and for each individual character, However for this project we limited it to only the Top 10 characters. ---> This gives us the character's emotion when he/she appears in the movie.
Additional Scene Informations --> Exact Scene Locations, Scenes with dialogs and no dialogs, Scenes that occurred during the Day or in the Night, Scenes location based on Outdoor or Indoor appearances.
Gender Distribution in the movie

*(python Code) Modules for this project: imsbd_moviescript_scraper_AND_Scene_Segmentation.py, dialogue_appearance.py, characters_extract.py, xter_interaction.py, characters_mt.py, emotions.py, movie_info.py, gend_distribution_plot.py*

Tools: Python libraries

In [ ]:

In [1]:

#Import all the necessary python modules needed for this analysis

from characters_extract import extract_characters
from dialogue_appearance import scene_dialogues
from xter_interaction import interaction
from emotions import emotions_sentiments
from characters_mt import character_mentions
from gend_distribution_plot import gender
from movie_info import scene_info_plots

In [3]:

import glob
import random
import secrets
import re

import cufflinks as cf
import networkx as net
import itertools
# plotly
from plotly.subplots import make_subplots
import plotly.graph_objects as go
import pandas as pd
import chart_studio.plotly as py
import plotly.graph_objs as go
from plotly import tools
from plotly.offline import init_notebook_mode, iplot
init_notebook_mode(connected=True)
import plotly.express as px

1. Load in the Segmented Movie scripts stored in .pkl format¶

In [4]:

films = []
for f in glob.glob('Films/*'):
    film_name = re.sub(r'.pkl|Films\\', '', f)
    films.append(film_name)

In [5]:

#Number of Films we segmented into scenes, scene_actions, characters, and characters dialogue
print('Number of films available for Analysis:  ', len(films), ' Movies')

Number of films available for Analysis:   1037  Movies

In [23]:

#Random 10 films from the 1000 films scraped from the internet
films_list = random.sample(films, 10)

In [25]:

#Randomly select film to analyze
film = secrets.choice(films_list)
print(film)

Antz

In [26]:

##load the scenes, dialogues, characters into dataframe

df_film = pd.read_pickle('Films/' + film + '.pkl')
df_film_dialogue = pd.read_pickle('Dialogues/' + film + '.pkl')
df_film_characters = pd.read_pickle('Characters/' + film + '.pkl')

In [27]:

#Randomly generate 10 scenes from the movie script
df_film.sample(10)

Out[27]:

	Scene_Names	Scene_action	Scene_Characters	Scene_Dialogue	Contents
33	EXT. INSECTOPIA, TOP OF CAN NIGHT	As Ant Team Six lands and takes up position ab...	[TRACKER ANT, MANDIBLE, COMMANDO, MANDIBLE]	[Yesyesyesyesyes One of the commandoes puts h...	As Ant Team Six lands and takes up position a...
9	INT. TOWN CENTER NIGHT	The ant army has gathered in a huge HALL in fr...	[FORMICA, FORMICA, FORMICA, FORMICA, COLONEL, ...	[First of all, let me make one thing clear. N...	The ant army has gathered in a huge HALL in f...
5	INT. BALA'S QUARTERS DAY	Bala enters, followed by her handmaidens, who ...	[HANDMAIDEN, HANDMAIDEN, HANDMAIDEN, BALA, HAN...	[swooning over General Formica The General's b...	Bala enters, followed by her handmaidens, who...
28	EXT. GRASS JUNGLE DAY	Z and Bala are lost, wandering through the grass	[BALA, BALA, BALA]	[I swear, we have passed this blade of grass t...	Z and Bala are lost, wandering through the gr...
37	INT. FORMICA'S CHAMBERS DAY	Formica sits at his desk, going over some plan...	[FORMICA, BALA, FORMICA, BALA, FORMICA, BALA, ...	[Princess Bala. Good. Where's Z? Bala replies...	Formica sits at his desk, going over some pla...
11	EXT. TERRAIN NEAR TERMITE STUMP NIGHT	Z looks up to see...looming high above them......	[COLONEL, BARBATUS, BARBATUS, BARBATUS, COLONE...	[ATTAAAAAAAAACK The front line of ants starts ...	Z looks up to see...looming high above them.....
20	INT. KITCHEN DAY	Z backs up into the kitchen, still pulling Bal...	None	None	Z backs up into the kitchen, still pulling Ba...
3	INT. TOWN CENTER DAY	The huge, spacious main chamber of the colony....	[GENERAL FORMICA, CARPENTER, GENERAL FORMICA, ...	[Cut the chitchat down there We have spoiled t...	The huge, spacious main chamber of the colony...
40	INT. MID SKYLIGHT TUNNEL	Bala, tied and gagged. She's at the point in ...	None	None	Bala, tied and gagged. She's at the point in...
16	EXT. ANT MOUND DAY	Some guard ants are looking out across the san...	[GUARD ANT]	[Look They are back The army's back Alert t...	Some guard ants are looking out across the sa...

In [28]:

#check how many scenes the movie script has
df_film.shape

Out[28]:

(57, 5)

In [29]:

#Randomly select characters and their corresponding dialogues
df_film_dialogue.sample(10)

Out[29]:

	characters	Character_dialogue
115	FORMICA	Dammit, I am proud to be an ant. he looks out ...
184	FORMICA	May I cut in? Z Oh, of course
404	QUEEN	We are Z. Now, the workers are all shouting th...
71	WEAVER	Poor guy's had one too many scouting missions....
97	SOLDIER	An individual? Never heard of it.
281	BALA	You were right...you were right Z, it's beauti...
125	ANT SOLDIERS	to the tune of When Johnny Comes Marching Home...
75	BALA	And do Princesses do improper things?
288	MALE WASP	They are Eurotrash, dear. We hear a loud RUMBL...
220	WORKER	I heard about this guy. turning to the other w...

2. Character appearance plots¶

In [30]:

ext = extract_characters(df_film, df_film_dialogue, df_film_characters, film)

In [37]:

antz_characters = ext.extract_character_plot()

In [32]:

dia = scene_dialogues(df_film, film)

In [33]:

df_xter_app = dia.character_appearances(antz_characters)

Lets check the appearance of the top 3 characters across the Movie¶

In [34]:

print('Movie Characters:  \n', antz_characters)

Movie Characters:  
 ['BALA', 'FORMICA', 'WEAVER', 'AZTECA', 'QUEEN', 'BARBATUS', 'WORKER', 'HANDMAIDEN', 'MANDIBLE', 'CARPENTER', 'MALE WASP', 'MOTIVATIONAL COUNSELLOR', 'FLY', 'TRACKER ANT', 'COLONEL', 'ANT OFFICER', 'FOREMAN', 'GENERAL FORMICA', 'SOLDIER', 'CRICKET', 'DRUNK SCOUT', 'BARKER', 'ANT SOLDIERS', 'FEMALE WASP', 'BEETLE', 'LADYBUG', 'SOLDIER ANT', 'WORKERS', 'WASP']

In [35]:

df_bala_count, df_bala_dialogue = dia.xter_count_perscene(antz_characters[0])

In [36]:

dia.scene_dialogue_plot(df_bala_count)

In [38]:

df_fm_count, df_fm_dialogue = dia.xter_count_perscene(antz_characters[1])
dia.scene_dialogue_plot(df_fm_count)

In [39]:

df_wv_count, df_wv_dialogue = dia.xter_count_perscene(antz_characters[2])
dia.scene_dialogue_plot(df_wv_count)

In [40]:

df_fm_bl_count, df_fm_bl_dialogue = dia.xter_count_perscene(antz_characters[:2])
dia.scene_dialogue_plot(df_fm_bl_count)

In [41]:

df_fm_wv_count, df_fm_wv_dialogue = dia.xter_count_perscene(gotd_characters[1:3])
dia.scene_dialogue_plot(df_fm_wv_count)

3. Character Interaction Mapping¶

In [42]:

interact = interaction(df_film, film)

In [43]:

graph_list = interact.character_interaction()

Note the Net.MultiGraph code below will throw an error, BUT this does not stop it from displaying the Character Network Multigraph figure,¶

To run it, remove the hashtags i.e. comment it out¶

In [49]:

# G = net.MultiGraph()
# for scene in graph_list:
#     nodes = list(itertools.combinations(scene,2))
#     for pair in nodes:
#         G.add_edges_from([pair])
# page_ranked_nodes = net.pagerank_numpy(G,0.95)
# net.enumerate_all_cliques(G)
# between_nodes = net.betweenness_centrality(G, normalized=True, endpoints=True)

In [46]:

interact.character_interaction_plot(G, page_ranked_nodes)

Let break it down to the Top 10 Characters¶

In [47]:

#Remember to Re-run the above multigraph code aafter running this code line
graph_list = interact.top10_character_interaction(gotd_characters[:10])

In [50]:

interact.character_interaction_plot(G, page_ranked_nodes)

4. Characters mentioned the most in Scene Dialogues (Most-talked about characters in the movie.......) and who each character mention the most in their specific dialogue¶

In [51]:

xtr = character_mentions(df_film, antz_characters, film)

In [52]:

xter_mentions = xtr.most_mentioned()

Lets the check the appearance of the 5 most mentioned characters in the Movie¶

In [73]:

xtr.top_xters_mentions(xter_mentions, 5)

Lets check who specific characters mention the most in their dialogues in the movie¶

In [54]:

print(antz_characters)

['BALA', 'FORMICA', 'WEAVER', 'AZTECA', 'QUEEN', 'BARBATUS', 'WORKER', 'HANDMAIDEN', 'MANDIBLE', 'CARPENTER', 'MALE WASP', 'MOTIVATIONAL COUNSELLOR', 'FLY', 'TRACKER ANT', 'COLONEL', 'ANT OFFICER', 'FOREMAN', 'GENERAL FORMICA', 'SOLDIER', 'CRICKET', 'DRUNK SCOUT', 'BARKER', 'ANT SOLDIERS', 'FEMALE WASP', 'BEETLE', 'LADYBUG', 'SOLDIER ANT', 'WORKERS', 'WASP']

In [55]:

df_bala = xtr.talk_about_xters(df_film_dialogue, 'BALA')

In [56]:

df_fm = xtr.talk_about_xters(df_film_dialogue, 'FORMICA')

In [72]:

df_queen = xtr.talk_about_xters(df_film_dialogue, 'QUEEN')

Lets check who specific characters talk with the most in the movie¶

In [57]:

df_wv = xtr.most_talked_with('WEAVER')

In [59]:

df_az = xtr.most_talked_with(antz_characters[3])

We could continue to check this for every character but I will pause here....¶

5. Emotional Arcs and Sentiment Analysis across the Movie and for each characters¶

In [60]:

etn = emotions_sentiments(df_film, film)

In [61]:

df_film_sentiment = etn.film_sentiment('darkorchid')

In [62]:

df_film_emotion = etn.film_emotional_arc()

Lets check the emotional content for the Top 10 characters in the Movie¶

In [63]:

df_top10_emotions =  etn.emotional_content_plot(df_film_dialogue, antz_characters, 11)

Lets check emotional arcs for specific characters¶

Emotions at scenes where the character is not present is set to ZERO¶

In [64]:

df_ba_emotions = etn.emotional_arc_xter_plot(df_film_emotion, 'BALA')

In [65]:

df_WV_emotions = etn.emotional_arc_xter_plot(df_film_emotion, 'WEAVER')

In [66]:

df_fm_emotions = etn.emotional_arc_xter_plot(df_film_emotion, 'FORMICA')

As usual, this can be done for all characters in the movie, but we will pause here....¶

4. Movie additional information: Exact Scene location, Indoor/Outdoor Scenes percent, Time of Day occurences in the movie....¶

In [67]:

info = scene_info_plots(df_film, film)

In [68]:

info.extract_scene_locations()

In [69]:

info.pie_plots()

5. Gender Distribution in the Movie¶

In [70]:

gd = gender(antz_characters, film)

In [71]:

df_gender = gd.gender_types(px.colors.sequential.Inferno)

[nltk_data] Downloading package names to C:\Users\Adeboye
[nltk_data]     Adeniyi\AppData\Roaming\nltk_data...
[nltk_data]   Package names is already up-to-date!

This should give a screen writer a broad overview of the underlying insights that could be discovered from an existing screen play or movie script.¶

Thanks for visualizing this extensive film script analysis...¶

Remember all codes for this analysis are already in python modules....¶

In [ ]: