Uning tuzilishini aniq va quyi oqim algoritmlari tomonidan oson boshqarish mumkin

(1) Yo'naltirilmagan, tortilmagan grafik
Yuqorida ta'kidlab o'tilganidek, grafik tasviri qanday bo'lishidan qat'i nazar, yuqori darajadagi usul har doim bir xil bo'ladi, bu holda 4 bosqichdan iborat :

1. 
   Hujjatlarni oldindan qayta ishlash;
   
2. 
   Ob'ektlarni aniqlash (grafikdagi tugunlar);
   
3. 
   Aloqalarni aniqlang (grafikning qirralari);
   
4. 
   Haqiqiy grafikni yarating.
   

Men hozir ushbu bosqichlarning har birini (va tegishli ilovalarni) ko'rib chiqaman. Ushbu bosqichlarning har birini amalga oshirish, ayniqsa ob'ekt va aloqadorlar, biz boshqa grafik tasvirlarni yaratish uchun o'zgartirgan narsamizdir.
import networkx as nx
import matplotlib.pyplot as plt

def build_graph(doc):

# preprocess document for standardization
pdoc = preprocess_document(doc)
# get graph nodes
nodes = get_entities(pdoc)
# get graph edges
edges = get_relations(pdoc)
# create graph structure with NetworkX
G = nx.Graph()
G.add_nodes_from(nodes)
G.add_edges_from(edges)
return G

Asosiy funktsiyani Python amalga oshirish. Kirish = matnli hujjat (string). Chiqish = NetworkX grafik ob'ekti.

Hujjatni oldindan qayta ishlash
Yuqoridagi spetsifikatsiyalarga ko'ra, alohida so'zlar tugunlar, jumlalardagi bigrammalar esa qirralardir. Shunday qilib, dastlabki ishlov berish bosqichi hujjatni jumlalarga va har bir jumlani so'zlarga bo'lishi kerak . Bundan tashqari, biz o'zgaruvchanlikni minimallashtirishni xohlayotganimiz uchun u barcha matnni kichik harf bilan yozishi kerak (masalan, "Salom" va "salom" so'zlari bir xil atama hisoblanadi). Vazifa taʼrifida tinish belgilariga eʼtibor bermaganimiz uchun ular ham olib tashlanadi.
def preprocess_document(document, sentence_spliter='.', word_spliter=' ', punct_mark=','):
# lowercase all words and remove trailing whitespaces
document = document.lower().strip()
# remove unwanted punctuation marks
for pm in punct_mark:
document = document.replace(pm, '')
# get list of sentences which are non-empty
sentences = [sent for sent in document.split(sentence_spliter) if sent != '']
# get list of sentences which are lists of words
document = []
for sent in sentences:
words = sent.strip().split(word_spliter)
document.append(words)
return document