Problem with the existing approach
In the baseline approach, we got great accuracy. However, we ignored the following points, which we can be implemented in our revised approach:
We did not focus on word embedding-based techniques
Deep learning (DL) algorithms such as CNN can be helpful for us
We need to focus on these two points because word embedding-based techniques really help retain the semantics of the text. So we should use these techniques as well as the DL-based-algorithm, which helps us provide more accuracy because DL algorithms perform well when a nested data structure is involved. What do I mean by a nested data structure? Well, that means any written sentence or spoken sentence made up of phrases, phrases made of words, and so on. So, natural language has a nested data structure. DL algorithms help us understand the nested structure of the sentences from our text dataset.
