Classification of Mauritania dialect using ML Algorithms
summary
Local Mauritania dialect data harvested from Facebook and labeled into positive, negative and neutral classes gets cleaned, tokenized (Arabic normalization, stopword removal) and vectorized with bag-of-words and TF-IDF features. Models tested are SVM, logistic regression, and random forest trained with an 80/20 train/test split. Logistic regression consistently outperforms the others, reaching roughly 73% accuracy and about 82% precision on the positive class while SVM sits near 58%; author recommends larger labeled data, BERT-style fine tuning, and real-time deployment.
AI-written summary · about
chapters (15)
- 0:00Mauritania dialect research intro
- 1:30Sentiment analysis goals
- 2:15Mauritanian dialect classification focus
- 2:36Data to evaluation methodology
- 4:01Hassaniya dialect in social media
- 4:45Text preprocessing & tokenization
- 5:52TF & TF-IDF feature extraction
- 6:54ML algorithms for classification
- 7:39Implementation tools & env
- 8:22SVM, logistic regression & RF results
- 10:09Arabic & Mauritanian text processing
- 11:39Class distribution & precision
- 12:01Future research recommendations
- 13:28Conclusion & findings summary
- 14:08Q&A
related talks
27mMachine Learning Methods in Detecting Alzheimer’s Disease from Speech and Language
22mFacebook data and machine learning predicting elections?
48mLearning Machine Learning
11mClassifying Spoken Location Names
45mHow multilingual is your NLP model?
35mAnalysing user comments with Doc2Vec and Machine Learning classification
46mPractical Machine Learning and Rails
36mMachine Learning Sucks