Symbolics实验 - 微软概念图

概念图是公开关系图，访问方法：

import urllib

import json

import ssl



def http(x):

ssl._create_default_https_context = ssl._create_unverified_context

response = urllib.request.urlopen(x)

data = response.read()

return data.decode('utf-8')



def query(x):

return json.loads(http("https://concept.research.microsoft.com/api/Concept/ScoreByProb?instance={}&topK=10".format(urllib.parse.quote(x))))



query('microsoft')

如果要更新title的话，用NewsApi.org的数据，调api：

newsapi_key = '<your API key here>'def get_news(country='us'):

res = json.loads(http("https://newsapi.org/v2/top-headlines?country={0}&apiKey={1}".format(country,newsapi_key)))

return res['articles']



all_titles = [x['title'] for x in get_news('us')+get_news('gb')]

用TextBlob库萃取title的名词：

import sys!{sys.executable} -m pip install textblob

!{sys.executable} -m textblob.download_corpora

from textblob import TextBlob



w = {}for x in all_titles:

for n in TextBlob(x).noun_phrases:

if n in w:

w[n].append(x)

else:

w[n]=[x]{ x:len(w[x]) for x in w.keys()}

如果是要将title替换成一般词句（避免深奥专业词），要对所有词句用一次REST：

w = {}

for x in all_titles:

 for noun in TextBlob(x).noun_phrases:

   terms = query(noun.replace(' ','%20'))

 for term in [u for u in terms.keys() if terms[u]>0.1]:

  if term in w:

   w[term].append(x)

  else:

   w[term]=[x]

调用：

x:len(w[x])

for x in w.keys()

  if len(w[x])>3

输出：

{'city': 9,

'brand': 4,

'place': 9,

'town': 4,

'factor': 4,

'film': 4,

'nation': 11,

'state': 5,

'person': 4,

'organization': 5,

'publication': 10,

'market': 5,

'economy': 4,

'company': 6,

'newspaper': 6,

'relationship': 6}

打印title的具体描述：

print('\nECONOMY:\n'+'\n'.join(w['economy']))

print('\nNATION:\n'+'\n'.join(w['nation']))print('\nPERSON:\n'+'\n'.join(w['person']))