Academia.eduAcademia.edu

Figure 2 – uploaded by Rhea Bharal

See full PDFdownloadDownload figure
Word2Vector is based on the fact that some words have similarities to one another which helps us to perform natural language processing.

Figure 2 Word2Vector is based on the fact that some words have similarities to one another which helps us to perform natural language processing.

Related Figures (6)

Figure 1: Negative vs Positive Tweets
Figure 1: Negative vs Positive Tweets
using the max operation. important information is lost about where exactly it appeared Unlike a CNN, RNN is suitable for temporal information, also called sequential data. However, it is difficult to train RNNs with learning long-term temporal dependencies as it causes time lag problems during training and disappearing gradient problems hence, Bi-LSTM, Using bidirectional network, the inputs will run in two directions, one from future to past and one from past to future and this is bette1 than LSTM as it preserves information in memory from the future and uses the two hidden states together for retaining information from both past and future at any given time. Thus it has proven to give good results in natural language processing
using the max operation. important information is lost about where exactly it appeared Unlike a CNN, RNN is suitable for temporal information, also called sequential data. However, it is difficult to train RNNs with learning long-term temporal dependencies as it causes time lag problems during training and disappearing gradient problems hence, Bi-LSTM, Using bidirectional network, the inputs will run in two directions, one from future to past and one from past to future and this is bette1 than LSTM as it preserves information in memory from the future and uses the two hidden states together for retaining information from both past and future at any given time. Thus it has proven to give good results in natural language processing
Figure 4: Bi-LSTM Cell Volume 10 Issue 9, September 2021
Figure 4: Bi-LSTM Cell Volume 10 Issue 9, September 2021
Figure 5: CNN-BiLSTM Layer
Figure 5: CNN-BiLSTM Layer
In figure 8, We can see the incorrectly and correctly classified tweets. The proportion in the principle diagonal are the correctly classified tweets. The accuracy of the model is 78.86% by using CNN-BiLSTM model. The performance of model on this dataset is better than other machine learning techniques. Therefore, CNN-BiLSTM is very good for classifying the tweets and that too in shorter period of training time.
In figure 8, We can see the incorrectly and correctly classified tweets. The proportion in the principle diagonal are the correctly classified tweets. The accuracy of the model is 78.86% by using CNN-BiLSTM model. The performance of model on this dataset is better than other machine learning techniques. Therefore, CNN-BiLSTM is very good for classifying the tweets and that too in shorter period of training time.
Figure 10: Accuracy comparison with different models Deep learning methods such as CNN-BiLSTM show better performance of sentiment classification with more amounts of training data) As shown in figure 10, Our model’s accuracy is better as compared to other models that have been trained on the same dataset Sentiment 140.
Figure 10: Accuracy comparison with different models Deep learning methods such as CNN-BiLSTM show better performance of sentiment classification with more amounts of training data) As shown in figure 10, Our model’s accuracy is better as compared to other models that have been trained on the same dataset Sentiment 140.
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved