Apr 14, 2019, 20:21 pm
Would TPB be interested in a filter for spam comments, the ones advertising streaming sites?
I am an advanced beginner with machine learning and could design and train a model that identifies such comments specifically for TPB. Then you could reject them automatically.
To train, I would need a corpus of spam comments and legitimate comments, especially ones that closely resemble spam but are not. The training would yield measures for classification errors - non-spam positives and spam negatives. Comment evaluation would yield a class and confidence level. Ambiguous comments could be sent to a moderator or CAPTCHA barrier.
The filter requires Python and PyTorch. Once trained, it could be run on a CPU or GPU (50x faster). I estimate about 10 ms per comment classification.
I would offer this software development for free in order to solve a practical problem and to develop my skills.
Please respond by PM if you are interested.
I am an advanced beginner with machine learning and could design and train a model that identifies such comments specifically for TPB. Then you could reject them automatically.
To train, I would need a corpus of spam comments and legitimate comments, especially ones that closely resemble spam but are not. The training would yield measures for classification errors - non-spam positives and spam negatives. Comment evaluation would yield a class and confidence level. Ambiguous comments could be sent to a moderator or CAPTCHA barrier.
The filter requires Python and PyTorch. Once trained, it could be run on a CPU or GPU (50x faster). I estimate about 10 ms per comment classification.
I would offer this software development for free in order to solve a practical problem and to develop my skills.
Please respond by PM if you are interested.
