Popis: |
Labeling dataset remains one of the big challenges for machine learning practitioners in developing countries especially those in Africa. Indeed, the effectiveness of machine learning models depends on the volume of training data available. The more the training dataset is huge, the more the models are effective. In this paper, we present a NLP system to annotate tweet related to meningitis based on linguistic features. We defined different assertion type of tweet mentioning the keyword meningitis. To automate the features extraction, we implemented a NER model called MeNER for meningitis entity recognition. We defined different rules for each assertion for labeling purpose, and we performed a classification task using ANN. Our ANN model achieved obtained a best accuracy of 0.93. |