Lipstick Expert

Create environment

pip install -r requirements.txt

Create Database

Get all video_id

python get_video_id.py

You can change query_list in the main of get_video_id.py to get different video_id. And we get video_id.txt.

Get video infomation

please download the dataset used for sentiment analysis training at first.

tar -xvzf posdata.tar.gz
rm -rf posdata.tar.gz
tar -xvzf negdata.tar.gz
rm -rf negdata.tar.gz

How to get video infomation

Please name all your video_id data with the format like video_{num}.txt. One video_id dataset should contain no more than 100 video_id to avoid data loss. Result data for each dataset would be restored in data_{num}.json. You can run get_video_info.py like this:

python get_video_info.py num

After you get all .json, you can merge all these json files by comment the parts in "main" that are not commented, and uncomment the parts that are commented, and then run:

python get_video_info.py

Database

After the first phrase of data collection, you could get database named video_info_database.json.

Get product information

First of all, we store the all color information we collected before in color.json, you should guarantee that color.json in the same folder of get_db.py. Besides, you need the database you get from the last step, if you database named video_info_database.json, you can just run the code like.

python get_db.py video_info_database.json

Database

After running this python file, you could get database with both video and product information named database_with_sephora.json

Train and calculate score

After we get the final json file, we rename it as final.json with sephora data inside, we run:

python scoring.py

The function will read in the json file and otuput a processed json file "score_list.json" with scoring infomation inside which is needed by UI part.