Restrict the use of banned languages on your discord server. Use the language detection model with hugging face to detect the language used by members of your discord channel.
If you want to prevent your server members from using languages other than the approved set, the bot will warn them. This code is just a starting point for more advanced responses, which can easily be built on top of existing codebase.
echo "<HERE-YOUR-TOKEN>" > discord_token.txt
docker-compose up
The following repo is a simple but useful way to restrict the use of certain languages on your discord server. It is based on the XLM-RoBERTa model trained on the language identification dataset.
The message is sent to the bot (API), and if it contains an unwanted language, the user receives a warning.
The code itself is super simple and should be treated more as a starting point for future modifications. 99% of the "business logic" can be found in on_message function.
To make the bot useful you will need a token. Here you can find a step by step tutorial.
Once you have a working token you need to save it to the discord_token.txt file. You can do this by running:
echo "<HERE-YOUR-TOKEN>" > discord_token.txt
where <HERE-YOUR-TOKEN> is something like: OTIIzaaTcjY2xOussUy.YbkUkg.Eg.zhsN2ytTud1ClfeHGaTS
This will be used to authorize the discord client.
docker-compose up
All request will be logged, so you can actively observe what the bot is doing.
Q: How to change the language or message?
A: In config.yaml you can change the allowed languages, the returned message or the length of sentences that are not checked by the bot. The last one if useful cause the model sometimes tend to have a problem with shorter messages like "xd" or ":(".
min_post_char_length: 15
accepted_languages: [en]
message: "Please use English on this server"Q: What languages are available in the model?
A: arabic (ar), bulgarian (bg), german (de), modern greek (el), english (en), spanish (es), french (fr), hindi (hi), italian (it), japanese (ja), dutch (nl), polish (pl), portuguese (pt), russian (ru), swahili (sw), thai (th), turkish (tr), urdu (ur), vietnamese (vi), and chinese (zh)
Q: What about misspellings of words. Will the model still work?
A: Yes, the model seems to do great even with misspelled sentences.
Q: Can I deploy that using AWS Lambda/Azure Functions
A: As far as I know, you can't. The delay in running the lambda will be too great to be useful, but let me know if you know how to make it work. I would be happy to run it this way. Though services like Azure Container Instance or a simple EC2 machine should work great.
Q: Can I use this repo to harass minorities on my server?
A: No! The license prohibits you from doing that.
Q: What are the limitations of the model?
A: It might have the problem with short messages, hence the min_post_char_length heuristics. It can also perform poorly when presented with languages it has never seen (ML models tend to perform poorly when presented with data outside the training set distribution).