def re_create_config(event, context=None):
"""
End-point: (Temporary) To re-create the config after changes are made programmatically
"""
# get config. configuration index has 1 document
result = []
es = getESConn()
utility.createOrUpdateGlobalConfig(es, config_db=config_db)
time.sleep(0.3) # 0.3 sec wait to allow time for created index to be ready
query = {"query": retrieve.MatchAll()}
res = es.search(index=config_db, body=query)
if (res['hits']['total']['value'] > 0):
result = res['hits']['hits'][0]['_source']
response = {
"statusCode": 200,
"headers": headers,
"body": json.dumps(result)
}
return response
def get_config(event, context=None):
"""
End-point: Retrieves configuration
"""
# get config. configuration index has 1 document
result = []
es = getESConn()
if not es.indices.exists(index=config_db): # create config db if it does not exist
utility.createOrUpdateGlobalConfig(es, config_db=config_db)
time.sleep(0.3) # 0.3 sec wait to allow time for created index to be ready
query = {"query": retrieve.MatchAll()}
res = es.search(index=config_db, body=query)
if (res['hits']['total']['value'] > 0):
result = res['hits']['hits'][0]['_source']
response = {
"statusCode": 200,
"headers": headers,
"body": json.dumps(result)
}
return response
def all_projects(event, context=None):
"""
End-point: Retrieves all projects. Each project is separate CBR application.
"""
result = []
# retrieve if ES index does exist
es = getESConn()
if es.indices.exists(index=projects_db):
query = {}
query['query'] = retrieve.MatchAll()
res = es.search(index=projects_db, body=query)
for hit in res['hits']['hits']:
entry = hit['_source']
entry['id__'] = hit['_id']
result.append(entry)
response = {
"statusCode": 200,
"headers": headers,
"body": json.dumps(result)
}
return response
def cbr_retrieve(event, context=None):
"""
End-point: Completes the Retrieve step of the CBR cycle.
"""
start = timer() # start timer
result = {'recommended': {}, 'bestK': []}
es = getESConn() # es connection
# query["query"]["bool"]["should"].append(queryFnc)
queryAdded = False
params = json.loads(event['body']) # parameters in request body
# print(params)
queryFeatures = params['data']
proj = params.get('project', None)
if proj is None:
projId = params.get('projectId', None) # name of casebase
proj = utility.getByUniqueField(es, projects_db, "_id", projId)
proj_attributes = proj['attributes']
globalSim = params['globalSim']
k = params['topk']
query = {"query": {"bool": {"should": []}}}
query["size"] = int(k) # top k results
for entry in queryFeatures:
if ('value' in entry) and entry['value'] is not None and "" != entry['value'] and int(
entry.get('weight', 0)) > 0 and entry['similarity'] != "None":
queryAdded = True
field = entry['name']
similarityType = entry['similarity']
options = retrieve.get_attribute_by_name(proj['attributes'], field).get('options', None)
# print(options)
# fieldType = entry['type']
# use lowercase when field is specified as case-insensitive
value = entry['value'].lower() if similarityType == 'EqualIgnoreCase' else entry['value']
weight = entry['weight']
# isProblem = entry['unknown']
# strategy = entry['strategy']
qfnc = retrieve.getQueryFunction(field, value, weight, similarityType, options)
query["query"]["bool"]["should"].append(qfnc)
if not queryAdded: # retrieval all (up to k) if not query was added
query["query"]["bool"]["should"].append(retrieve.MatchAll())
# perform retrieval
counter = 0
res = es.search(index=proj['casebase'], body=query)
for hit in res['hits']['hits']:
entry = hit['_source']
entry.pop('hash__', None) # remove hash field and value
entry = retrieve.remove_vector_fields(proj_attributes, entry) # remove vectors from Semantic USE fields
if counter == 0:
result['recommended'] = copy.deepcopy(entry)
# entry['id__'] = hit['_id'] # using 'id__' to track this case (this is removed during an update operation)
entry['score__'] = hit['_score'] # removed during an update operation
result['bestK'].append(entry)
counter += 1
# Recommend: Get the recommended result using chosen reuse strategies for unknown attribute values and keep known attribute values supplied
if counter > 0:
for entry in queryFeatures:
if not entry['unknown'] and ('value' in entry) and entry['value'] is not None and "" != entry[
'value']: # copy known values
result['recommended'][entry['name']] = entry['value']
if entry.get('similarity') is not None and entry['unknown'] and entry[
'strategy'] != "Best Match": # use reuse strategies for unknown fields
if entry['strategy'] == "Maximum":
result['recommended'][entry['name']] = max(d[entry['name']] for d in result['bestK'])
if entry['strategy'] == "Minimum":
result['recommended'][entry['name']] = min(d[entry['name']] for d in result['bestK'])
if entry['strategy'] == "Mean":
result['recommended'][entry['name']] = np.mean([x[entry['name']] for x in result['bestK']])
if entry['strategy'] == "Median":
result['recommended'][entry['name']] = np.median([x[entry['name']] for x in result['bestK']])
if entry['strategy'] == "Mode":
result['recommended'][entry['name']] = statistics.mode([x[entry['name']] for x in result['bestK']])
end = timer() # end timer
result['retrieveTime'] = end - start
result['esTime'] = res['took']
response = {
"statusCode": 200,
"headers": headers,
"body": json.dumps(result)
}
return response
def cbr_retrieve(event, context=None):
"""
End-point: Completes the Retrieve step of the CBR cycle.
"""
start = timer() # start timer
result = {'recommended': {}, 'bestK': []}
# query['query']['bool']['should'].append(queryFnc)
queryAdded = False
params = json.loads(event['body']) # parameters in request body
# print(params)
queryFeatures = params['data']
proj = params['project']
globalSim = params['globalSim']
k = params['topk']
query = {'query': {'bool': {'should': []}}}
query['size'] = int(k) # top k results
for entry in queryFeatures:
if ('value' in entry
) and entry['value'] is not None and "" != entry['value'] and int(
entry['weight']) > 0 and entry['similarityType'] != "None":
queryAdded = True
field = entry['field']
# fieldType = entry['type']
value = entry['value']
weight = entry['weight']
# isProblem = entry['unknown']
# strategy = entry['strategy']
similarityType = entry['similarityType']
qfnc = retrieve.getQueryFunction(field, value, weight,
similarityType)
query['query']['bool']['should'].append(qfnc)
if not queryAdded: # retrieval all (up to k) if not query was added
query['query']['bool']['should'].append(retrieve.MatchAll())
# print(query)
# perform retrieval
counter = 0
es = getESConn()
res = es.search(index=proj['casebase'], body=query)
for hit in res['hits']['hits']:
entry = hit['_source']
entry.pop('hash__', None) # remove hash field and value
if counter == 0:
result['recommended'] = copy.deepcopy(entry)
# entry['id__'] = hit['_id'] # using 'id__' to track this case (this is removed during an update operation)
entry['score__'] = hit['_score'] # removed during an update operation
result['bestK'].append(entry)
counter += 1
# Recommend: Get the recommended result using chosen reuse strategies for unknown attribute values and keep known attribute values supplied
if counter > 0:
for entry in queryFeatures:
if not entry['unknown'] and (
'value'
in entry) and entry['value'] is not None and "" != entry[
'value']: # copy known values
result['recommended'][entry['field']] = entry['value']
if entry['similarityType'] != "None" and entry['unknown'] and entry[
'strategy'] != "Best Match": # use reuse strategies for unknown fields
if entry['strategy'] == "Maximum":
result['recommended'][entry['field']] = max(
d[entry['field']] for d in result['bestK'])
if entry['strategy'] == "Minimum":
result['recommended'][entry['field']] = min(
d[entry['field']] for d in result['bestK'])
if entry['strategy'] == "Mean":
result['recommended'][entry['field']] = np.mean(
[x[entry['field']] for x in result['bestK']])
if entry['strategy'] == "Median":
result['recommended'][entry['field']] = np.median(
[x[entry['field']] for x in result['bestK']])
if entry['strategy'] == "Mode":
result['recommended'][entry['field']] = statistics.mode(
[x[entry['field']] for x in result['bestK']])
end = timer() # end timer
result['retrieveTime'] = end - start
result['esTime'] = res['took']
response = {
"statusCode": 200,
"headers": headers,
"body": json.dumps(result)
}
return response
