def average_time_per_package():
annotation_id_blacklist = get_package_annotations()
results_start = labelled_collection.aggregate(pipeline=[
{
"$match": {
'_id': {'$nin': annotation_id_blacklist}
}
},
{
"$group": {
"_id": {"pid": "$package_id", "uid": "$user_name"},
"start_time": {"$min": "$start_time"}
}
}
])
results_end = labelled_collection.aggregate(pipeline=[
{
"$match": {
'_id': {'$nin': annotation_id_blacklist}
}
},
{
"$group": {
"_id": {"pid": "$package_id", "uid": "$user_name"},
"end_time": {"$max": "$end_time"}
}
}
])
min_time_for_id = {(res['_id']['pid'], res['_id']['uid']): res['start_time'] for res in results_start}
times_in_ms = []
for result in results_end:
pid = result['_id']['pid']
uid = result['_id']['uid']
end_time = result['end_time']
if min_time_for_id.get((pid, uid), None) is not None:
start_time = min_time_for_id[(pid, uid)]
times_in_ms.append(end_time-start_time)
sorted_times = sorted(times_in_ms)
if (len(sorted_times)%2) == 0:
idx = int(len(sorted_times)/2)
return (sorted_times[idx] + sorted_times[idx-1])/2
else:
idx = int((len(sorted_times) - 1)/2)
return sorted_times[idx]
def median_time_per_annotation():
annotation_id_blacklist = get_package_annotations()
results = labelled_collection.aggregate(pipeline=[
{
"$match": {
'_id': {'$nin': annotation_id_blacklist}
}
},
{
"$group": {
"_id": "dummy",
"elapsed_times": {"$addToSet": "$elapsed_time"}
}
}
])
elapsed_times = list(results)[0]['elapsed_times']
sorted_elapsed_times = sorted(elapsed_times)
if (len(sorted_elapsed_times) % 2) == 0:
idx = int(len(sorted_elapsed_times) / 2)
return (sorted_elapsed_times[idx] + sorted_elapsed_times[idx - 1]) / 2
else:
idx = int((len(sorted_elapsed_times) - 1) / 2)
return sorted_elapsed_times[idx]
def number_of_annotators():
results = labelled_collection.aggregate(pipeline=[
{
"$group": {
"_id": {"uid": "$user_name"},
"convos": {"$sum": 1}
}
}
])
return len(list(results))
def number_of_annotators_filtered():
annotation_id_blacklist = get_package_annotations()
results = labelled_collection.aggregate(pipeline=[
{
"$match": {
'_id': {'$nin': annotation_id_blacklist}
}
},
{
"$group": {
"_id": {"uid": "$user_name"},
"convos": {"$sum": 1}
}
}
])
return len(list(results))
def total_time():
annotation_id_blacklist = get_package_annotations()
results = labelled_collection.aggregate(pipeline=[
{
"$match": {
'_id': {'$nin': annotation_id_blacklist}
}
},
{
"$group": {
"_id": "dummy",
"total_time": {"$sum": "$elapsed_time"}
}
}
])
total_time = list(results)[0]['total_time']
#convert to hrs
return total_time/1000/60/60
def get_package_annotations():
results = labelled_collection.aggregate(pipeline=[{
"$group": {
"_id": {
"pid": "$package_id",
"uid": "$user_name"
},
"convos": {
"$addToSet": "$convo_id"
}
}
}])
pres = package_collection.find({})
pid_to_cids = {
str(p['_id']): set([c['convo_id'] for c in p['package']])
for p in pres
}
annotation_blacklist = []
for res in results:
pid = res['_id']['pid']
convo_ids = set(res['convos'])
if pid_to_cids.get(pid, None) is None:
continue
expected_ids = set(pid_to_cids[pid])
if not convo_ids == expected_ids:
annotation_blacklist.append(res['_id'])
annotation_id_blacklist = []
for entry in annotation_blacklist:
annotations = labelled_collection.find({
'package_id': entry['pid'],
'user_name': entry['uid']
})
aids = [annotation['_id'] for annotation in annotations]
annotation_id_blacklist.extend(aids)
return annotation_id_blacklist