From cecd4f770f93687b47bc08436d3c139390cfed95 Mon Sep 17 00:00:00 2001
From: rlx <rlx@rolux.org>
Date: Mon, 22 Jan 2018 12:55:02 +0100
Subject: [PATCH] update recommendation engine

---
 recommendation_engine.py | 36 +++++++++++++++++++++++++-----------
 1 file changed, 25 insertions(+), 11 deletions(-)

diff --git a/recommendation_engine.py b/recommendation_engine.py
index 7b68dcf..7482277 100644
--- a/recommendation_engine.py
+++ b/recommendation_engine.py
@@ -32,6 +32,21 @@ class Engine:
         return clips[index:] + clips[:index - 1]
 
     def get_videos(self, user):
+        channels = {'keywords': 7, 'screenings': 7, 'random': 2}
+        sliders = {'dau': -1, 'physics': 0, 'sex': 1}
+        # For each playlist, compute keyword score
+        score = {}
+        for playlist in self.playlists:
+            score[playlist['name']] = random.random()
+            for tag in [tag for tag in playlist['tags'] if tag in sliders]:
+                score[playlist['name']] += sliders[tag]
+        # Select highest scoring playlists
+        playlists = sorted(
+            self.playlists,
+            key=lambda playlist: -score[playlist['name']]
+        )
+        videos = playlists[:channels['keywords']]
+        playlists = playlists[channels['keywords']:]
         # Count tags for the user
         count = {}
         for event in user['events']:
@@ -39,23 +54,22 @@ class Engine:
                 count[event['data']['product']] = count.get(
                     event['data']['product'], 0
                 ) + 1
-        score = {}
         # For each tag in playlist, increment score by count
-        for playlist in self.playlists:
+        for playlist in playlists:
             score[playlist['name']] = random.random()
-            for tag in playlist['tags']:
+            for tag in [tag for tag in playlist['tags'] if tag not in sliders]:
                 score[playlist['name']] += count.get(tag, 0)
-        # Select 16 highest scoring playlists
-        playlists = sorted(
-            self.playlists,
+        # Select highest scoring playlists
+        videos += sorted(
+            playlists,
             key=lambda playlist: -score[playlist['name']]
-        )[:16]
+        )[:16 - channels['keywords']]
         # Shuffle playlists (randomize layout) and shift clips (randomize start)
-        random.shuffle(playlists)
+        random.shuffle(videos)
         return [{
-            'clips': self._shift_clips(playlist['clips']),
-            'name': playlist['name']
-        } for playlist in playlists]
+            'clips': self._shift_clips(video['clips']),
+            'name': video['name']
+        } for video in videos]
 
     def update(self):
         # Get all storylines with tags