Tired of labelling your dataset the boring way? Introducing Tinder for Reviews. Now label your dataset as if you are scrolling on tinder!
A demo of how the works.
2024-06-07_21-31-08.mov
pip install -r requirements.txt
- Save your SQLite database in
database
folder and go throughuser_config.py
and setup your SQLite database accordingly.
An example user_config.py with headers such as ["id", "review_content", "positive", "negative", "neutral", "polarity"]
has been added to examples/user_config.py along with a demo database at demoDB/feedback.db
table_name = "feedback"
primary_key = "id" # primary key of your database. If None, it will be auto generated.
DATABASE_PATH = './demoDB/feedback.db'
# what all columns/headers does your sql data have?
headers = ["id", "review_content", "positive",
"negative", "neutral", "polarity"]
# which column/header do you want to be shown on the review card?
column_for_review = "review_content"
review_type = "text" # options: "text" (support for image, videos soon)
polarity_column = "polarity" # if set to None, a new column will be created with the name "polarity"
# what does each swipe add value to? please make sure these columns are int
# polarity wise, right is positive (+1), up is neutral (+0), and left is negative (-1)
left_swipe = ["negative", "negative"] # [column name, display name (label on the website)]
up_swipe = ["neutral", "neu"] # [column name, display name (label on the website)]
right_swipe = ["positive", "positive"] # [column name, display name (label on the website)]
# A list of tuple which has username and then password
users = [("admin", "admin")]
# setup host here for the server.
# future tests might require the testcase to run the server
# so to make sure it runs on the same host, you can set it up here.
host = "localhost"
port = 5556
python main.py
and open http://localhost:5556
(or your custom host) in your browser.
- right swipes or green button = positive*.
- left swipe or red button = negative*.
- white button "neu" = neutral*.
* or whatever you have configured in user_config.py
Run tests using pytest
on your terminal.
pytest --verbose -s
You can use this to label your dataset for sentiment analysis, feedback analysis, etc in a fun way. Well, that's what I am using this for :P