tweet repetition testing w/ mismatched goldstandard-players

example.py

from bson.objectid import ObjectId
from analysis.tw import *
import pprint

client = pymongo.MongoClient('mongodb://algdev.sportsmanias.com')
db = client['sportmanias']['tweets']

parent = Tw(db.find_one({'id_str':'890067420207042561'}))

a = Tw(db.find_one({'id_str':'889706097598054400'}))

b = Tw(db.find_one({'id_str':'889694791969030144'}))


print('parent.text: {}'.format(parent.text))
print('a.text: {}'.format(a.text))
print('b.text: {}'.format(b.text))

print('\n\n')

print('parent.players: {}'.format(parent.players))
print('a.players: {}'.format(a.players))
print('b.players: {}'.format(b.players))

print('\n\n')
print("parent compared to a")
pprint.pprint(parent.compare(a))
print('\n\n')
print("parent compared to b")
pprint.pprint(parent.compare(b))

example2.py

from bson.objectid import ObjectId
from analysis.tw import *
import pprint

print('\n\n\n\n\n')
client = pymongo.MongoClient('mongodb://algdev.sportsmanias.com')
db = client['sportmanias']['tweets']


tweets = ['891850453935300608','891843946401325056']
#tweets = ['892498705911296005', '892174106652704768']
#tweets = ['892497331500838912', '892485281793921024', '892123146119204865']
#tweets = ['892496175063982080', '892489097637384192']
#tweets = ['892492385111900160', '892144636491423744']


a = Tw(db.find_one({'id_str':tweets[0]}))
b = Tw(db.find_one({'id_str':tweets[1]}))



print('a.text: {}'.format(a.text))
print('b.text: {}'.format(b.text))

print('\n\n')

print('a.players: {}'.format(a.players))
print('b.players: {}'.format(b.players))

print('\n\n')
print("a compared to b:\n")
pprint.pprint(a.compare(b))

example2_output.txt

a.text: Ravin pitching for Dodgers.
b.text: Paredes pitching for Dodgers.



a.players: ['Josh Ravin']
b.players: [None]



a compared to b:

{'common-links': [],
 'duplicate': False,
 'links': [],
 'links2': [],
 'numbers': [],
 'numbers2': [],
 'players': ['Josh Ravin'],
 'players2': [None],
 'reason': 'goldstandard players mismatch',
 'similarity': 0.99999999999999989,
 'similarity_threshold': 0.85,
 'text': 'Ravin pitching for Dodgers.',
 'text2': 'Paredes pitching for Dodgers.'}