diff --git a/ai/recommender/base.py b/ai/recommender/base.py
index 4ecb0e900..9e71a0111 100644
--- a/ai/recommender/base.py
+++ b/ai/recommender/base.py
@@ -179,22 +179,23 @@ def _train_empty_rec(self,
         # load serialized rec, or initialize from the given
         # knowledgebase
         logger.info(f"load_serialized_rec='{load_serialized_rec}'")
+        logger.info(f"self.serialized_rec_path='{self.serialized_rec_path}'")
 
         if load_serialized_rec == "always":
             if not os.path.exists(self.serialized_rec_path):
                 raise ValueError(f"load_serialized_rec='{load_serialized_rec}'"
                         " but cannot load serialized recommender:"
-                        " '{self.serialized_rec_path}'")
+                        f" '{self.serialized_rec_path}'")
             self.load(self.serialized_rec_path, knowledgebase_results)
 
         elif load_serialized_rec == "if_exists":
             if os.path.exists(self.serialized_rec_path):
                 logger.info(f"Loading serialized recommender:"
-                        " {self.serialized_rec_path}")
+                        f" '{self.serialized_rec_path}'")
                 self.load(self.serialized_rec_path, knowledgebase_results)
             else:
                 logger.warn(f"Not loading serialized recommender, file does "
-                        "not exist: {self.serialized_rec_path}")
+                        f"not exist: '{self.serialized_rec_path}'")
                 if knowledgebase_results is not None:
                     logger.info(f"Initializing new recommender from provided "
                             "knowledgebase")
diff --git a/ai/recommender/surprise_recommenders.py b/ai/recommender/surprise_recommenders.py
index f20edc030..ff8ce5768 100644
--- a/ai/recommender/surprise_recommenders.py
+++ b/ai/recommender/surprise_recommenders.py
@@ -52,7 +52,7 @@
 import logging
 
 logger = logging.getLogger(__name__)
-logger.setLevel(logging.DEBUG)
+logger.setLevel(logging.INFO)
 ch = logging.StreamHandler()
 formatter = logging.Formatter('%(module)s: %(levelname)s: %(message)s')
 ch.setFormatter(formatter)
@@ -100,15 +100,16 @@ def __init__(self,
         self.ml_type = ml_type
 
         if metric is None:
-            logger.debug('metric is None, setting...')
+            logger.warning('metric is None, setting...')
             self.metric='bal_accuracy' if self.ml_type=='classifier' else 'mse'
         else:
             self.metric = metric
         assert(self.metric is not None)
 
-        logger.debug('self.algo_name: '+self.algo_name)
-        logger.debug('ml_type: '+self.ml_type)
-        logger.debug('metric: '+self.metric)
+        logger.info('initlizing SurpriseRecommender')
+        logger.info('self.algo_name: '+self.algo_name)
+        logger.info('ml_type: '+self.ml_type)
+        logger.info('metric: '+self.metric)
 
         self.min_epochs = 10
         self.max_epochs = 100
@@ -439,7 +440,7 @@ def _update_model(self,results_data):
         self.algo.n_epochs = min(len(results_data),self.max_epochs)
         self.algo.n_epochs = max(self.algo.n_epochs,self.min_epochs)
 
-        logger.debug('fitting self.algo...')
+        logger.info('fitting self.algo...')
         self.algo.partial_fit(self.trainset)
-        logger.debug('done.')
-        logger.debug('model SVD updated')
+        logger.info('done.')
+        logger.info('model SVD updated')
diff --git a/docker-compose-production.yml b/docker-compose-production.yml
index c131773ba..fd0a94429 100644
--- a/docker-compose-production.yml
+++ b/docker-compose-production.yml
@@ -12,6 +12,7 @@ services:
     volumes:
       - "/appsrc"
       - "./data/datasets/user:/appsrc/data/datasets/user"
+      - "./data/recommenders/pennaiweb:/appsrc/data/recommenders/pennaiweb"
       - "./target:/appsrc/target"
     ports:
       - "5080:5080"