0x2620/dd-re
2
0
Fork 0
Code Issues 4 Pull requests Releases Wiki Activity

Merge branch 'iss1'

Browse source
This commit is contained in:
pythagoraswitch 2018-08-03 00:13:19 +02:00
commit c8b6699915
2 changed files with 115 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
.gitignore vendored
View file

@ -1,3 +1,7 @@
json/*
venv/*
__pycache__/
*.swp
*.ipynb
*.pyc
test.py

114
recommendation_engine.py
View file

@ -15,6 +15,7 @@ import ox
from utils import run_async
logger = logging.getLogger(__name__)
verbose = True
class Engine:
@ -46,6 +47,12 @@ class Engine:
'screenings': {'locked': True, 'value': 2}
},
'globalKeywords': {},
'gridChange': {
'nextClip': {'locked': False, 'value': 4},
'nextPlaylist': {'locked': False, 'value': 4},
'staySame': {'locked': False, 'value': 8}
}
}
self.update_keywords()
@ -80,16 +87,106 @@ class Engine:
def get_videos(self, user):
channels = {k: v.get('value', 0) for k, v in self.state['channels'].items()}
sliders = {k: v.get('value', 0) for k, v in self.state['globalKeywords'].items()}
grid_change = {k: v.get('value', 0) for k, v in self.state['gridChange'].items()}
# check if there were grid events for all indexes.
grid_events = {}
(nc, np, ns) = (grid_change.get("nextClip"), grid_change.get("nextPlaylist"), grid_change.get("staySame"))
# this assumes np + nc + ns = total number of videos in the grid view (16).
# Make sure sanity check exists in front-end (error if it does not add up to 16).
video_num = nc + np + ns
# for event in user.get('events', []):
# if event.get('event') == "grid" and event.get('data').get('index') not in grid_events:
# grid_events[event.get('data').get('index')] = event.get('data')
# if len(grid_events) == video_num:
# break
# # The version where the loop also extract play_index (requires "index" in play event data):
play_index = None
for event in user.get('events', []):
if event.get('event') == "grid" and event.get('data').get('index') not in grid_events:
grid_events[event.get('data').get('index')] = event.get('data')
if event.get('event') == "play" and not play_index:
play_index = event.get('data').get('index')
if len(grid_events) == video_num and play_index:
break
prev_grid_list = sorted([v for v in grid_events.values()], key=lambda k:k['index'])
# if there were no grid events for all, initialize all grids.
if len(prev_grid_list) < video_num:
return self.get_recommendations(user)
else:
# played index is excluded from the random shuffle and deterministically added to staySame pool.
video_indx = [*range(play_index)]+[*range(play_index+1,video_num)]
# video_indx = list(range(video_num))
random.shuffle(video_indx)
video_indx.append(play_index)
next_clip_index = video_indx[:nc]
next_playlist_index = video_indx[nc:nc+np]
stay_same_index = video_indx[nc+np:]
rec_list = []
# select next clip for nextClip pool except when the playlist has only one clip.
for i in next_clip_index:
for playlist in self.playlists:
if playlist.get('name')== prev_grid_list[i].get('playlist'):
if len(playlist["clips"]) == 1:
next_playlist_index.append(i)
break
# Discuss how this behavour should be: should it switch to a new playlist if it is the end of the playlist clip sequence already?
elif prev_grid_list[i].get('playlistPosition') + 1 == len(playlist['clips']):
playlist_pos = 0
else:
playlist_pos = prev_grid_list[i].get('playlistPosition') + 1
rec_list.append((i, {
'clips': playlist['clips'],
# 'position': random.randrange(len(playlist['clips'])),
'position': playlist_pos,
'name': playlist['name'],
'tags': playlist['tags'],
}))
# randomly select playlists (excluding the playlists from the current grid once "playlist" is recorded for grid events)
# for nextPlaylist pool.
vids_exclude = [e.get("playlist") for e in prev_grid_list]
video = self.get_recommendations(user, vids_exclude)
rec_list += [(i, video[i]) for i in next_playlist_index]
#staySame pool
rec_list += [(i,{}) for i in stay_same_index]
rec_list = sorted(rec_list, key=lambda k:k[0])
return [e[1] for e in rec_list]
# NOTE for future improvement: vids_exclude element unit could be clip or in/out time pairs, rather than playlist.
# The same playlist could be played in the grid view as long as these are differenct clips or separate times.
def get_recommendations(self, user, vids_exclude = []):
channels = {k: v.get('value', 0) for k, v in self.state['channels'].items()}
sliders = {k: v.get('value', 0) for k, v in self.state['globalKeywords'].items()}
gridChange = {k: v.get('value', 0) for k, v in self.state['gridChange'].items()}
# Exclude playlists from the most recent grid
playlists = self.playlists
if len(vids_exclude) > 0:
for playlist in playlists:
if playlist["name"] in vids_exclude:
playlists.remove(playlist)
# For each playlist, compute user keyword score
user_keywords = user.get('keywords', {})
score = {}
for playlist in self.playlists:
for playlist in playlists:
score[playlist['name']] = random.random()
for tag in [tag for tag in playlist['tags'] if tag in user_keywords]:
score[playlist['name']] += user_keywords[tag]
# Select highest scoring playlists
playlists = sorted(
self.playlists,
playlists,
key=lambda playlist: -score[playlist['name']]
)
videos = playlists[:channels['userKeywords']]
@ -131,8 +228,19 @@ class Engine:
'tags': video['tags'],
} for video in videos]
def get_next(self, user, position):
video = self.get_videos(user)[position]
grid_events = {}
video_num = 16
for event in user.get('events', []):
if event.get('event') == "grid" and event.get('data').get('index') not in grid_events:
grid_events[event.get('data').get('index')] = event.get('data')
if len(grid_events) == video_num:
break
prev_grid_list = sorted([v for v in grid_events.values()], key=lambda k:k['index'])
vids_exclude = [e.get("playlist") for e in prev_grid_list]
video = self.get_recommendations(user, vids_exclude)[position]
return video
def update_state(self, data):