I trained a neural network to learn Arabic morphology资料
本文为去找网小编(www.7zhao.net)为您推荐的I trained a neural network to learn Arabic morphology，希望对您有所帮助，谢谢！
A fun little project to play with Jupyter Notebooks, Scikit-learn, and neural nets with Keras.
To train a neural network to learn Arabic morphology.
Scripts for data mining
2 iterations of the model: roots.ipynb and roots-with-noroots.ipynb.
roots-with-noroots.ipynbis named weird, but it just means we are more intelligent about tracking words that are "mabniyy", or undeclined.
It's not very accurate (about 50%) so it's pretty addictive to work on. Surely, someone, somewhere, has done this better, but we aren't solving world hunger here, just having some nerdy fun.
Pull requests welcome :)
The output is formatted as "accuracy: [correctAnswer, input]" with output on the end if incorrect.
Correct: ['تبديلي', 'بدل'] Missed: ['الألياف', 'ليف'] Predicted: للف Correct: ['تنْويت', 'نوت'] ... Correct: ['متوسّط', 'وسط'] Correct: ['تنفيذي', 'نفذ'] Missed: ['المتبقية', 'بقا'] Predicted: ربب Missed: ['متقدم', 'قدم'] Predicted: ودم Missed: ['الأساسي', ''] Predicted: سسس Correct: ['تغيير', 'غير'] Correct: ['سطر', 'سطر'] Correct: ['سليم', 'سلم'] Correct: ['جملة', 'جمل'] Missed: ['الذهب', 'ذهب'] Predicted: هبب Correct: ['طرفية', 'طرف'] Correct: ['متوسطة', 'وسط'] Missed: ['سوق', 'سوق'] Predicted: وقق Score: 58.8%
How to use
Just install Jupyter Notebook and run
jupyter notebookin this folder, and select one of the
以上为I trained a neural network to learn Arabic morphology文章的全部内容，若您也有好的文章，欢迎与我们分享！