news_titles = [
"Half of all older adults are worried about dementia, survey says",
"Telus invests $500,000 to bring wireless service to First Nation’s community in B.C.",
"Asteroid Bigger Than the Eiffel Tower Approaching Earth at 20 Times the Speed of Sound",
"Night shift: Bottas beats Vettel by 0.012s for pole as Hamilton starts in P5",
"Exclusive: John Barnes discusses Liverpool FC’s January transfer plans",
"Maharashtra: Balasaheb, Pawar relation beyond politics",
"Which words has Collins Dictionary included in its Brexicon?",
"Paralympics: Australian cycling champion Modra killed in road collision",
"Multistate salmonella outbreak causes one death: CDC",
]
LABEL_STUDIO_CONFIG ='''<View>
<Text name="post" value="$text"></Text>
<Choices name="label" toName="title" choice="multiple">
<Choice value="Economy"></Choice>
<Choice value="Technology"></Choice>
<Choice value="Sports"></Choice>
<Choice value="Science"></Choice>
<Choice value="Entertainment"></Choice>
<Choice value="Other"></Choice>
</Choices>
</View>'''
with WebUI(frontend_dir='label_studio', context={"label_studio":{"config":LABEL_STUDIO_CONFIG}, 'title':title}) as tru: # start http server in background
for item in news_titles:
label = tru.ask({"data": {"text": item}})
print("data: {}, label: {}".format(item, label)) #output result
'value': 'economy',
'shortcut': "['s']"
},
{
'label': 'technology (d)',
'value': 'technology',
'shortcut': "['d']"
},
{
'label': 'sports (f)',
'value': 'sports',
'shortcut': "['f']"
},
{
'label': 'other (k)',
'value': 'other',
'shortcut': "['k']"
},
{
'label': 'skip (enter)',
'value': 'skip',
'shortcut': "['enter']"
},
]
with WebUI(context={'title': title}) as tru: # start http server in background
for item in news_titles:
label = tru.ask({"html": item},
meta={"buttons": random.sample(controls, k=3)})
print("data: {}, label: {}".format(item, label.y)) #output result
'shortcut': "['s']"
},
{
'label': 'technology (d)',
'value': 'technology',
'shortcut': "['d']"
},
{
'label': 'sports (f)',
'value': 'sports',
'shortcut': "['f']"
},
{
'label': 'other (k)',
'value': 'other',
'shortcut': "['k']"
},
{
'label': 'skip (enter)',
'value': 'skip',
'shortcut': "['enter']"
},
]
with WebUI(context={
'buttons': controls,
'title': title
}) as tru: # start http server in background
for item in news_titles:
label = tru.ask({"html": item})
print("data: {}, label: {}".format(item, label.y)) #output result
""",
"""On August 10, 2015, Google Inc. announced plans to create a new public holding company, Alphabet Inc. Google CEO Larry Page made this announcement in a blog post on Google's official blog.[11] Alphabet would be created to restructure Google by moving subsidiaries from Google to Alphabet, narrowing Google's scope. The company would consist of Google as well as other businesses including X Development, Calico, Nest, Verily, Malta, Fiber, Makani, CapitalG, and GV.[6][12][13] Sundar Pichai, Product Chief, became the new CEO of Google, replacing Larry Page, who transitioned to the role of running Alphabet, along with Google co-founder Sergey Brin.[14][15]
"""
]
CONFIG = '''
<View>
<Labels name="ner" toName="text">
<Label value="Person"></Label>
<Label value="Organization"></Label>
<Label value="Date"></Label>
<Label value="Location"></Label>
</Labels>
<Text name="text" value="$text"></Text>
</View>
'''
INTERFACES = [
"basic", "load", "controls", "submit", "completions", "side-column"
]
with WebUI(
frontend_dir='label_studio',
context={"label_studio": {
"config": CONFIG,
"interfaces": INTERFACES
}}) as tru: # start http server in background
for data in texts:
completions = tru.ask({"data": {"text": data}})
print("{}".format(json.dumps(completions, ensure_ascii=False)),
flush=True) #output result
def get_annotations(data, default_sampling_strategy="random"):
"""Prompts annotator for label from command line and adds annotations to data
Keyword arguments:
data -- an list of unlabeled items where each item is
[ID, TEXT, LABEL, SAMPLING_STRATEGY, CONFIDENCE]
default_sampling_strategy -- strategy to use for each item if not already specified
"""
controls = [
{'label':'disaster-related (a)', 'value':1, 'shortcut':"['a']"},
{'label':'not related (x)', 'value':0, 'shortcut':"['x']"},
{'label':'back (arrow left)', 'value':2, 'shortcut':"['arrowleft']"},
{'label':'save and exit (e)', 'value':"s", 'shortcut':"['e']"},
]
with WebUI(context={'buttons':controls, 'title':'News annotation'}) as tru: # start http server in background
ind = 0
tru.update({"html":"<br><h3>Instruction</h3><br><pre>"+detailed_instructions+"</pre>"})
while ind <= len(data):
if ind < 0:
ind = 0 # in case you've gone back before the first
if ind < len(data):
textid = data[ind][0]
text = '<h5>'+data[ind][1]+'</h5>'
label = data[ind][2]
strategy = data[ind][3]
if textid in already_labeled:
print("Skipping seen "+label)
ind+=1
else:
label = tru.ask({"html": text})
label = str(label.y)
tru.update({"html":""})
if label == "2":
ind-=1 # go back
elif label == "s":
break # save and exit
else:
if not label == "1":
label = "0" # treat everything other than 1 as 0
data[ind][2] = label # add label to our data
if data[ind][3] is None or data[ind][3] == "":
data[ind][3] = default_sampling_strategy # add default if none given
ind+=1
else:
#last one - give annotator a chance to go back
tru.update({"html":"<br><pre>"+last_instruction+"</pre>"})
label = tru.ask({"html": text})
label = str(label.y)
if label == "2":
ind-=1
else:
ind+=1
return data
controls = [
{
'label': 'Cat (a)',
'value': 'cat',
'shortcut': "['a']"
},
{
'label': 'Dog (x)',
'value': 'dog',
'shortcut': "['x']"
},
{
'label': 'Skip (enter)',
'value': '',
'shortcut': "['enter']"
},
]
def img_tag(filename):
return "<img src=\"/data/{}\">".format(filename)
with WebUI(data_dir='./images', context={
'buttons': controls,
'title': title
}) as tru: # start http server in background
for item in images:
label = tru.ask({"html": img_tag(item)})
print("image: {}, label: {}".format(item, label.y)) #output result