Review sentiment analysis based on deep neural network

With the rapid development of the Internet and related technologies, network data has shown a spurt of growth, mainly in the form of text. With this growth trend of data, text classification has become an increasingly important research topic. The use of deep learning technology to represent text ha...

Full description

Saved in:
Bibliographic Details
Main Author: Song, Wanzhen
Other Authors: Mao Kezhi
Format: Thesis-Master by Coursework
Language:English
Published: Nanyang Technological University 2020
Subjects:
Online Access:https://hdl.handle.net/10356/143417
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-143417
record_format dspace
spelling sg-ntu-dr.10356-1434172023-07-04T16:38:07Z Review sentiment analysis based on deep neural network Song, Wanzhen Mao Kezhi School of Electrical and Electronic Engineering EKZMao@ntu.edu.sg Engineering::Electrical and electronic engineering With the rapid development of the Internet and related technologies, network data has shown a spurt of growth, mainly in the form of text. With this growth trend of data, text classification has become an increasingly important research topic. The use of deep learning technology to represent text has received great attention from researchers. Based on this, this paper studies and implements the sentiment classification problem of comments. By investigating the research status of sentiment analysis, combined with the current cutting-edge technologies in the field of machine learning and deep learning, the five deep learning models are explored and compared. The goal is to understand the principles of the models and choose more suitable ones in different situations to improve the accuracy of sentiment classification. The work done in this dissertation mainly includes the following parts: 1) Build CNN, BiLSTM, BiLSTM-At, RCNN, Transformer with the concept of attention mechanism. Compare the effects of different models on eight different types of datasets. 2) To improve the effectiveness of the model, this dissertation sets comparative experiments on parameters selection, to determine parameters that affect the model performance, and explain how they affect the model performance. 3) An in-depth exploration of the principle of different models. Find out the weight of each word in determining the sentiment of the text. Visualize three classic models. Behind the model that determine the sentiment of the text. 4) According to 3) draw the advantages and disadvantages of different models and determine the suitable condition for different models. Master of Science (Electronics) 2020-08-31T06:13:13Z 2020-08-31T06:13:13Z 2020 Thesis-Master by Coursework https://hdl.handle.net/10356/143417 en application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering::Electrical and electronic engineering
spellingShingle Engineering::Electrical and electronic engineering
Song, Wanzhen
Review sentiment analysis based on deep neural network
description With the rapid development of the Internet and related technologies, network data has shown a spurt of growth, mainly in the form of text. With this growth trend of data, text classification has become an increasingly important research topic. The use of deep learning technology to represent text has received great attention from researchers. Based on this, this paper studies and implements the sentiment classification problem of comments. By investigating the research status of sentiment analysis, combined with the current cutting-edge technologies in the field of machine learning and deep learning, the five deep learning models are explored and compared. The goal is to understand the principles of the models and choose more suitable ones in different situations to improve the accuracy of sentiment classification. The work done in this dissertation mainly includes the following parts: 1) Build CNN, BiLSTM, BiLSTM-At, RCNN, Transformer with the concept of attention mechanism. Compare the effects of different models on eight different types of datasets. 2) To improve the effectiveness of the model, this dissertation sets comparative experiments on parameters selection, to determine parameters that affect the model performance, and explain how they affect the model performance. 3) An in-depth exploration of the principle of different models. Find out the weight of each word in determining the sentiment of the text. Visualize three classic models. Behind the model that determine the sentiment of the text. 4) According to 3) draw the advantages and disadvantages of different models and determine the suitable condition for different models.
author2 Mao Kezhi
author_facet Mao Kezhi
Song, Wanzhen
format Thesis-Master by Coursework
author Song, Wanzhen
author_sort Song, Wanzhen
title Review sentiment analysis based on deep neural network
title_short Review sentiment analysis based on deep neural network
title_full Review sentiment analysis based on deep neural network
title_fullStr Review sentiment analysis based on deep neural network
title_full_unstemmed Review sentiment analysis based on deep neural network
title_sort review sentiment analysis based on deep neural network
publisher Nanyang Technological University
publishDate 2020
url https://hdl.handle.net/10356/143417
_version_ 1772825527315333120