Igor Brigadir igorbrigadir

Key-Value Memory Networks for Directly Reading Documents

Knowledge Bases (KBs) are effective tools for Question Answering (QA) but are often too restrictive (due to fixed schema) and too sparse (due to limitations of Information Extraction (IE) systems).
The paper proposes Key-Value Memory Networks, a neural network architecture based on Memory Networks that can leverage both KBs and raw data for QA.
The paper also introduces MOVIEQA, a new QA dataset that can be answered by a perfect KB, by Wikipedia pages and by an imperfect KB obtained using IE techniques thereby allowing a comparison between systems using any of the three sources.
Link to the paper.

$ uname -r

curl https://raw.githubusercontent.com/jsonlines/guide/master/datagov100.json > data.json

$ < data.json jq '.name' | head -n 6

	#!/usr/bin/env python

	"""

	Twitter's API doesn't allow you to get replies to a particular tweet. Strange
	but true. But you can use Twitter's Search API to search for tweets that are
	directed at a particular user, and then search through the results to see if
	any are replies to a given tweet. You probably are also interested in the
	replies to any replies as well, so the process is recursive. The big caveat
	here is that the search API only returns results for the last 7 days. So

	# Find your tweets from this date in earlier years.

	# OS X's date function doesn't have the required date math capabilities.
	# This requires GNU date. You can get it on OS X with `brew install coreutils`

	GDATE=/usr/local/bin/gdate
	USERNAME=realDonaldTrump

	export URL="https://twitter.com/search?q=from%3A$USERNAME%20("
	for yearsago in {1..12}; do

	import datetime
	from dateutil.relativedelta import relativedelta

	def today(date=None,iso=False):
	if date is None: date=datetime.date.today()
	if iso: return date.isoformat()
	else: return date

	def yesterday(date=None,iso=False):
	if date is None: date=today()

	"""
	how to count as fast as possible
	(numbers from Python 3.5.2 on a Macbook Pro)
	YMMV, but these results are pretty stable for me, say +/- 0.1s on repeated runs
	"""

	from collections import Counter, defaultdict
	import random

	random_numbers = [random.randrange(10000) for _ in range(10000000)]

	#!/usr/bin/env python
	# -- coding: utf-8 --

	import six
	import os

	if six.PY3:
	from urllib.request import urlopen
	else:
	from urllib2 import urlopen

	# -- coding: utf-8 --

	import requests
	from io import BytesIO, SEEK_SET, SEEK_END

	class ResponseStream(object):
	def __init__(self, request_iterator):
	self._bytes = BytesIO()
	self._iterator = request_iterator