reuters_ner.py

import spacy	
from matplotlib import cm
from matplotlib.pyplot import plt

nlp = spacy.load('en_core_web_sm')
ner_collection = {"Location":[],"Person":[],"Date":[],"Quantity":[],"Organisation":[]}
location = []
person = []
date = []
quantity = []
organisation = []
def ner_text(text):
    doc = nlp(text)
    ner_collection = {"Location":[],"Person":[],"Date":[],"Quantity":[],"Organisation":[]}
    for ent in doc.ents:
        if str(ent.label_) == "GPE":
            ner_collection['Location'].append(ent.text)
            location.append(ent.text)
        elif str(ent.label_) == "DATE":
            ner_collection['Date'].append(ent.text)
            person.append(ent.text)
        elif str(ent.label_) == "PERSON":
            ner_collection['Person'].append(ent.text)
            date.append(ent.text)
        elif str(ent.label_) == "ORG":
            ner_collection['Organisation'].append(ent.text)
            quantity.append(ent.text)
        elif str(ent.label_) == "QUANTITY":
            ner_collection['Quantity'].append(ent.text)
            organisation.append(ent.text)
        else:
            continue
    return ner_collection
   articles_word_limit['ner_data'] = articles_word_limit['text'].map(lambda x: ner_text(x))
    
location_name = []
location_count = []
for i in location_counts.most_common()[:10]:
    location_name.append(i[0].upper())
    location_count.append(i[1])


fig, ax = plt.subplots(figsize=(15, 8), dpi=100)
ax.barh(location_name, location_count, alpha=0.7,
         # width = 0.5,
        color=cm.Blues([i / 0.00525 for i in [ 0.00208, 0.00235, 0.00281, 0.00317, 0.00362,
                                              0.00371, 0.00525, 0.00679, 0.00761, 0.00833]])
        )
plt.rcParams.update({'font.size': 10})
rects = ax.patches
for i, label in enumerate(location_count):
    ax.text(label+100 , i, str(label), size=10, ha='center', va='center')
ax.text(0, 1.02, 'Count of Location name Extracted from Reuters Articles', 
        transform=ax.transAxes, size=12, weight=600, color='#777777')
ax.xaxis.set_ticks_position('bottom')
ax.tick_params(axis='y', colors='black', labelsize=12)
ax.set_axisbelow(True)
ax.text(0, 1.08, 'TOP 10 Location Mention in Reuters Articles',
        transform=ax.transAxes, size=22, weight=600, ha='left')
ax.text(0, -0.1, 'Source: http://kdd.ics.uci.edu/databases/reuters21578/reuters21578.html',
        transform=ax.transAxes, size=12, weight=600, color='#777777')
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)
ax.spines['left'].set_visible(False)
ax.spines['bottom'].set_visible(False)
plt.tick_params(axis='y',which='both', left=False, top=False, labelbottom=False)
ax.set_xticks([])
plt.show()