feat: indexing bayesian average ranking for courses in algolia

enterprise_catalog/apps/api/v1/views/enterprise_catalog_get_content_metadata.py

@@ -38,7 +38,7 @@ def enterprise_catalog(self):
 
     def get_permission_object(self):
         """
-        Retrieves the apporpriate object to use during edx-rbac's permission checks.
+        Retrieves the appropriate object to use during edx-rbac's permission checks.
 
         This object is passed to the rule predicate(s).
         """

enterprise_catalog/apps/api_client/constants.py

@@ -13,6 +13,7 @@
 DISCOVERY_COURSE_REVIEWS_ENDPOINT = urljoin(settings.DISCOVERY_SERVICE_API_URL, 'course_review/')
 DISCOVERY_OFFSET_SIZE = 200
 DISCOVERY_CATALOG_QUERY_CACHE_KEY_TPL = 'catalog_query:{id}'
+DISCOVERY_AVERAGE_COURSE_REVIEW_CACHE_KEY = 'average_course_review'
 
 # Enterprise API Client Constants
 ENTERPRISE_API_URL = urljoin(settings.LMS_BASE_URL, '/enterprise/api/v1/')

enterprise_catalog/apps/api_client/discovery.py

@@ -7,9 +7,11 @@
 import requests
 from celery.exceptions import SoftTimeLimitExceeded
 from django.conf import settings
+from django.core.cache import cache
 
 from .base_oauth import BaseOAuthClient
 from .constants import (
+    DISCOVERY_AVERAGE_COURSE_REVIEW_CACHE_KEY,
     DISCOVERY_COURSE_REVIEWS_ENDPOINT,
     DISCOVERY_COURSES_ENDPOINT,
     DISCOVERY_OFFSET_SIZE,
@@ -122,7 +124,7 @@ def _retrieve_course_reviews(self, request_params):
                 break
         return response.json()
 
-    def get_course_reviews(self, course_keys=None):
+    def get_course_reviews(self, course_keys=None, set_average_course_rating=False):
         """
         Return results from the discovery service's /course_review endpoint as an object of key = course key, value =
         course review. If course_keys is specified, only return results for those course keys.
@@ -149,6 +151,15 @@ def get_course_reviews(self, course_keys=None):
             )
         }
 
+        if set_average_course_rating:
+            rolling_rating_sum = 0.0
+            for value in results.values():
+                rolling_rating_sum += float(value.get('avg_course_rating'))
+
+            total_average_course_rating = rolling_rating_sum / len(results)
+            cache_key = DISCOVERY_AVERAGE_COURSE_REVIEW_CACHE_KEY
+            cache.set(cache_key, total_average_course_rating, settings.DISCOVERY_AVERAGE_COURSE_REVIEW_CACHE_TIMEOUT)
+
         return results
 
     def get_metadata_by_query(self, catalog_query):

enterprise_catalog/apps/catalog/algolia_utils.py

@@ -4,10 +4,14 @@
 import time
 
 from dateutil import parser
+from django.core.cache import cache
 from django.utils.translation import gettext as _
 
 from enterprise_catalog.apps.api.v1.utils import is_course_run_active
 from enterprise_catalog.apps.api_client.algolia import AlgoliaSearchClient
+from enterprise_catalog.apps.api_client.constants import (
+    DISCOVERY_AVERAGE_COURSE_REVIEW_CACHE_KEY,
+)
 from enterprise_catalog.apps.catalog.constants import (
     COURSE,
     EXEC_ED_2U_COURSE_TYPE,
@@ -85,6 +89,7 @@
     'normalized_metadata',
     'reviews_count',
     'avg_course_rating',
+    'course_bayesian_average',
 ]
 
 # default configuration for the index
@@ -137,6 +142,7 @@
     'customRanking': [
         'asc(visible_via_association)',
         'asc(created)',
+        'desc(course_bayesian_average)',
         'desc(recent_enrollment_count)',
     ],
 }
@@ -331,6 +337,21 @@ def get_algolia_object_id(content_type, uuid):
     return None
 
 
+def get_course_bayesian_average(course):
+    """
+    https://www.algolia.com/doc/guides/managing-results/must-do/custom-ranking/how-to/bayesian-average/
+    """
+    cache_key = DISCOVERY_AVERAGE_COURSE_REVIEW_CACHE_KEY
+    total_average_rating = cache.get(cache_key)
+    avg_review = course.get('avg_course_rating')
+    ratings_count = course.get('reviews_count')
+    if avg_review is not None and ratings_count is not None:
+        bayes_avg = ((avg_review * ratings_count) + (total_average_rating * 15)) / (ratings_count + 15)
+    else:
+        bayes_avg = 0
+    return bayes_avg
+
+
 def get_course_language(course):
     """
     Gets the human-readable language name associated with the advertised course run. Used for
@@ -1204,6 +1225,7 @@ def _algolia_object_from_product(product, algolia_fields):
             'learning_type_v2': get_learning_type_v2(searchable_product),
             'reviews_count': get_reviews_count(searchable_product),
             'avg_course_rating': get_avg_course_rating(searchable_product),
+            'course_bayesian_average': get_course_bayesian_average(searchable_product),
         })
     elif searchable_product.get('content_type') == PROGRAM:
         searchable_product.update({

enterprise_catalog/settings/base.py

@@ -380,9 +380,11 @@
 
 # How long we keep API Client data in cache. (seconds)
 ONE_HOUR = 60 * 60
+ONE_DAY = ONE_HOUR * 24
 ENTERPRISE_CUSTOMER_CACHE_TIMEOUT = ONE_HOUR
 DISCOVERY_CATALOG_QUERY_CACHE_TIMEOUT = ONE_HOUR
 DISCOVERY_COURSE_DATA_CACHE_TIMEOUT = ONE_HOUR
+DISCOVERY_AVERAGE_COURSE_REVIEW_CACHE_TIMEOUT = ONE_DAY
 
 # URLs
 LMS_BASE_URL = os.environ.get('LMS_BASE_URL', '')