Here's the recent articles submitted by geremew asrat
Articles By geremew asrat
Public Opinion Analysis Of Amharic News Using Deep Learning
By: geremew asrat
Sentiment analysis analyzes people's opinions, sentiments, and emotions from user generated texts. In this study, different approach comparison the training of two different modes of deep neural network architectures: long short term memory and gate recurrent units. The applied dataset has been extracted from AMC Amharic official Facebook page and used to test the effectiveness and assess the robustness of our planned approach, along with constructing our proper corpus, following the setting up of a special vector of words using word2vec. On test the attained dataset via the LSTM and the GRU models, classification accuracy rates of respectively 95.5% and 96% have been reached. The present work appears conducted to apply the LSTM architecture to the context of Amharic sentiment classification area as appearing in social media. As a potential work perspective, we consider experimenting with larger dataset range useful for training other more highly thorough and deep architectures, along with attempting to plan architectural combinations in a bid to achieve even more remarkably improved accurate results.(read
entire article)
View : 156 Times
Category : Computers
