Bangla E-Mail Body to Subject Generation Using Sequence to Sequence RNNs

Abstract

"In recent years, subject generation has emerged as

one of the major challenges for deep learning and natural

language processing (NLP). A brief comment on a lengthy email

body is condensed in the subject generation. Our goal is to

develop a Bengali subject generator that is effective and efficient

and can produce a clear and insightful subject from a given

Bengali email body. To do this, we have gathered a variety of

emails body, including educational, commercial, etc. And will

use our model to generate subjects from those texts. Our model

uses bi-directional RNNs in the encoding layer along with

LSTMs and the decoding layer we used an attention model. Our

model generates subjects using a sequence-to-sequence model.

While developing this model, we encountered difficulties with

text pre-processing, missing word counting, vocabulary

counting, identifying unfamiliar words, word embedding, and

other tasks. Our primary objectives in this model were to

generate a subject and lessen its training loss of it. By crafting a

succinct, fluent topic from an email body, we effectively

decreased the training loss in our study trial to 0.001"