def guess_keywords(obj, eng):
"""Workflow task to ask Magpie API for a keywords assessment."""
magpie_url = get_magpie_url()
if not magpie_url:
# Skip task if no API URL set
return
payload = prepare_magpie_payload(obj.data, corpus="keywords")
try:
results = json_api_request(magpie_url, payload)
except requests.exceptions.RequestException:
results = {}
if results:
labels = results.get('labels', [])
keywords = labels[:10]
keywords = [{'label': k[0], 'score': k[1], 'accept': k[1] >= 0.09} for
k in
keywords]
obj.extra_data["keywords_prediction"] = dict(
keywords=keywords
)
current_app.logger.info("Keyword prediction (top 10): {0}".format(
obj.extra_data["keywords_prediction"]["keywords"]
))
def guess_experiments(obj, eng):
"""Workflow task to ask Magpie API for a subject area assessment."""
magpie_url = get_magpie_url()
if not magpie_url:
# Skip task if no API URL set
return
payload = prepare_magpie_payload(obj.data, corpus="experiments")
results = json_api_request(magpie_url, payload)
if results:
all_predictions = results.get('labels', [])
selected_experiments = filter_magpie_response(
all_predictions,
limit=0.5,
)
selected_experiments = [
{'label': e[0], 'score': e[1]}
for e in selected_experiments
]
obj.extra_data["experiments_prediction"] = dict(
experiments=selected_experiments,
)
current_app.logger.info("Experiment prediction: {0}".format(
obj.extra_data["experiments_prediction"]["experiments"]
))
def test_json_api_request_retries_on_connection_error():
with requests_mock.Mocker() as requests_mocker:
body = {'foo': 'bar'}
requests_mocker.register_uri(
'POST', 'http://example.org/api', [
{'exc': requests.packages.urllib3.exceptions.ConnectionError},
{'json': body},
])
expected = {'foo': 'bar'}
result = json_api_request('http://example.org/api', {})
assert expected == result
def guess_experiments(obj, eng):
"""Workflow task to ask Magpie API for a subject area assessment."""
magpie_url = get_magpie_url()
if not magpie_url:
# Skip task if no API URL set
return
payload = prepare_magpie_payload(obj.data, corpus="experiments")
results = json_api_request(magpie_url, payload)
if results:
labels = results.get('labels', [])
experiments = filter_magpie_response(labels, limit=0.5)
obj.extra_data["experiments_prediction"] = dict(
labels=labels, experiments=experiments)
current_app.logger.info("Experiment prediction: {0}".format(
obj.extra_data["experiments_prediction"]["experiments"]))
def guess_coreness(obj, eng):
"""Workflow task to ask Beard API for a coreness assessment."""
predictor_url = get_beard_url()
if not predictor_url:
return
# FIXME: Have option to select different prediction models when
# available in the API
payload = prepare_payload(obj.data)
try:
results = json_api_request(predictor_url, payload)
except requests.exceptions.RequestException:
results = {}
if results:
scores = results.get('scores') or []
max_score = max(scores)
decision = results.get('decision')
scores = {
"CORE": scores[0],
"Non-CORE": scores[1],
"Rejected": scores[2],
}
# Generate a normalized relevance_score useful for sorting
# We assume a CORE paper to have the highest relevance so we add a
# significant value to seperate it from Non-Core and Rejected.
# Normally scores range from -2 / +2 so 10 is significant.
# Non-CORE scores are untouched, while Rejected is substracted -10.
# Finally this provides one normalized score of relevance across
# all categories of papers.
relevance_score = max_score
if decision == "CORE":
relevance_score += 10
elif decision == "Rejected":
relevance_score = (max_score * -1) - 10
# FIXME: Add top_words info when available from the API
obj.extra_data["relevance_prediction"] = dict(
max_score=max_score,
decision=decision,
scores=scores,
relevance_score=relevance_score
)
current_app.logger.info("Prediction results: {0}".format(
obj.extra_data["relevance_prediction"])
)
def guess_coreness(obj, eng):
"""Workflow task to ask Beard API for a coreness assessment."""
predictor_url = get_beard_url()
if not predictor_url:
return
# FIXME: Have option to select different prediction models when
# available in the API
payload = prepare_payload(obj.data)
try:
results = json_api_request(predictor_url, payload)
except requests.exceptions.RequestException:
results = {}
if results:
scores = results.get('scores') or []
max_score = max(scores)
decision = results.get('decision')
scores = {
"CORE": scores[0],
"Non-CORE": scores[1],
"Rejected": scores[2],
}
# Generate a normalized relevance_score useful for sorting
# We assume a CORE paper to have the highest relevance so we add a
# significant value to seperate it from Non-Core and Rejected.
# Normally scores range from -2 / +2 so 10 is significant.
# Non-CORE scores are untouched, while Rejected is substracted -10.
# Finally this provides one normalized score of relevance across
# all categories of papers.
relevance_score = max_score
if decision == "CORE":
relevance_score += 10
elif decision == "Rejected":
relevance_score = (max_score * -1) - 10
# FIXME: Add top_words info when available from the API
obj.extra_data["relevance_prediction"] = dict(
max_score=max_score,
decision=decision,
scores=scores,
relevance_score=relevance_score
)
current_app.logger.info("Prediction results: {0}".format(
obj.extra_data["relevance_prediction"])
)
def test_json_api_request_retries_on_connection_error():
with requests_mock.Mocker() as requests_mocker:
body = {'foo': 'bar'}
requests_mocker.register_uri('POST', 'http://example.org/api', [
{
'exc': requests.packages.urllib3.exceptions.ConnectionError
},
{
'json': body
},
])
expected = {'foo': 'bar'}
result = json_api_request('http://example.org/api', {})
assert expected == result
def guess_categories(obj, eng):
"""Workflow task to ask Magpie API for a subject area assessment."""
magpie_url = get_magpie_url()
if not magpie_url:
# Skip task if no API URL set
return
payload = prepare_magpie_payload(obj.data, corpus="categories")
results = json_api_request(magpie_url, payload)
if results:
labels = results.get('labels', [])
categories = filter_magpie_response(labels, limit=0.22)
categories = [{
'label': c[0],
'score': c[1],
'accept': c[1] >= 0.25
} for c in categories]
obj.extra_data["categories_prediction"] = dict(categories=categories)
current_app.logger.info("Category prediction: {0}".format(
obj.extra_data["categories_prediction"]["categories"]))
def guess_categories(obj, eng):
"""Workflow task to ask Magpie API for a subject area assessment."""
magpie_url = get_magpie_url()
if not magpie_url:
# Skip task if no API URL set
return
payload = prepare_magpie_payload(obj.data, corpus="categories")
results = json_api_request(magpie_url, payload)
if results:
labels = results.get('labels', [])
categories = filter_magpie_response(labels, limit=0.22)
categories = [{'label': c[0], 'score': c[1],
'accept': c[1] >= 0.25} for c in categories]
obj.extra_data["categories_prediction"] = dict(
categories=categories
)
current_app.logger.info("Category prediction: {0}".format(
obj.extra_data["categories_prediction"]["categories"]
))
def guess_keywords(obj, eng):
"""Workflow task to ask Magpie API for a keywords assessment."""
magpie_url = get_magpie_url()
if not magpie_url:
# Skip task if no API URL set
return
payload = prepare_magpie_payload(obj.data, corpus="keywords")
try:
results = json_api_request(magpie_url, payload)
except requests.exceptions.RequestException:
results = {}
if results:
labels = results.get('labels', [])
keywords = labels[:10]
keywords = [{
'label': k[0],
'score': k[1],
'accept': k[1] >= 0.09
} for k in keywords]
obj.extra_data["keywords_prediction"] = dict(keywords=keywords)
current_app.logger.info("Keyword prediction (top 10): {0}".format(
obj.extra_data["keywords_prediction"]["keywords"]))
