How textual quality of online reviews affect classification performance: a case of deep learning sentiment analysis

Abstract Objectives The aim of the present study was to create and test an automatic system for assessing the technical quality of positioning in periapical radiography of the maxillary canines using deep learning classification and segmentation techniques. Methods We created and tested two deep learning systems using 500 periapical radiographs (250 each of good- and bad-quality images). We assigned 350, 70, and 80 images as the training, validation, and test datasets, respectively. The learning model of system 1 was created with only the classification process, whereas system 2 consisted of both the segmentation and classification models. In each model, 500 epochs of training were performed using AlexNet and U-net for classification and segmentation, respectively. The segmentation results were evaluated by the intersection over union method, with values of 0.6 or more considered as success. The classification results were compared between the two systems. Results The segmentation performance of system 2 was recall, precision, and F measure of 0.937, 0.961, and 0.949, respectively. System 2 showed better classification performance values than those obtained by system 1. The area under the receiver operating characteristic curve values differed significantly between system 1 (0.649) and system 2 (0.927). Conclusions The deep learning systems we created appeared to have potential benefits in evaluation of the technical positioning quality of periapical radiographs through the use of segmentation and classification functions.

Download Full-text

eWOW of Guests Regarding Their Hotel Experience

Handbook of Research on Social Media Applications for the Tourism and Hospitality Sector - Advances in Hospitality, Tourism, and the Services Industry ◽

10.4018/978-1-7998-1947-9.ch017 ◽

2020 ◽

pp. 295-308

Author(s):

Zelia Breda ◽

Rui Costa ◽

Gorete Dinis ◽

Amandine Angie Martins

Keyword(s):

Sentiment Analysis ◽

Online Reviews ◽

User Generated Content ◽

Sources Of Information ◽

Value For Money ◽

Quality Of Sleep ◽

Online Comments ◽

Analytical Response ◽

Source Of Information

Online comments are increasingly mentioned as an important source of information, simplifying consumers' buying decisions. Online user-generated content has become one of the main sources of information for tourists, who themselves become creators of their own online content. This chapter focuses on sentiment analysis of comments made on TripAdvisor regarding one resort located in the Algarve region, in Portugal. The resort has good reviews, which means that the eWOM is positive. The highest scores relate to the resort's cleanliness, location and quality of sleep, and those that were less relevant were the value for money, the rooms and the service. The most dominant emotion is joy, followed by an analytical response. Negative emotions, such as sadness and anger, were not found very often in the online reviews. These results could be explained by the quality of the service, the kindness of the staff, the facilities for children, the entertainment, and the location, attributes that were often highlighted in the comments.

Download Full-text

Comparison of Attention-based Deep LearningModels for EEG Classification

10.21203/rs.3.rs-279263/v1 ◽

2021 ◽

Author(s):

Giulia Cisotto ◽

Alessio Zanga ◽

Joanna Chlebus ◽

Italo Zoppis ◽

Sara Manzoni ◽

...

Keyword(s):

Deep Learning ◽

State Of The Art ◽

Classification Performance ◽

Classification Problems ◽

Large Variability ◽

Eeg Classification ◽

Time Frequency ◽

Promising Strategy ◽

Complex Reasoning

Abstract Deep Learning (DL) has recently shown promising classification performance in Electroencephalography (EEG) in many different scenarios. However, the complex reasoning of such models often prevent the user to explain their classification abilities. Attention, one of the most recent and influential ideas in DL, allows the models to learn which portions of the data are relevant to the final classification output. In this work, we compared three attention-enhanced DL models, the brand-new InstaGATs , an LSTM with attention and a CNN with attention. We used these models to classify normal and abnormal, including artifactual and pathological, EEG patterns in three different datasets. We achieved the state of the art in all classification problems, regardless the large variability of the datasets and the simple architecture of the attention-enhanced models. Additionally, we proved that, depending on how the attention mechanism is applied and where the attention layer is located in the model, we can alternatively leverage the information contained in the time, frequency or space domain of the EEG dataset. Therefore, attention represents a promising strategy to evaluate the quality of the EEG information, and its relevance for classification, in different real-world scenarios.

Download Full-text

Sentiment Analysis and Deep Learning Based Cyber Bullying Detection in Twitter Dataset

International Journal of Recent Technology and Engineering - 2 ◽

10.35940/ijrte.d6511.1110421 ◽

2021 ◽

Vol 10 (4) ◽

pp. 15-25

Author(s):

Sherly T.T ◽

◽

B. Rosiline Jeetha ◽

Keyword(s):

Deep Learning ◽

Sentiment Analysis ◽

Fruit Fly ◽

Classification Performance ◽

Cyber Bullying ◽

Research Approach ◽

Suggested Approach ◽

Being Bullied ◽

Physical Distress ◽

Reduced Time

When somebody, usually a teenager, abuses or harasses individual on the internet and other digital places, mainly on social networking platforms, this is termed as cyberbullying. Cyberbullying, like all types of bullying, produces psychological, emotional, and physical distress. Every individual's reaction to being bullied is diverse, but research has discovered certain common patterns. In a recent study, we introduced a technique called Hybrid Firefly Artificial Neural Networks (HFANN) to combat cyberbullying. Nevertheless, without considering the sentiment analysis features, accuracy of cyber bullying identification is lowered in this study. The Sentiment Analysis and Deep Learning based Cyber Bullying Detection (SADL-CDD) approach is used in the suggested research approach to address this issue. The punctuations, urls, html tags, and emoticons from the input tweet comments are removed first in this study project. Sentiment feature extraction is performed after pre-processing to improve classification accuracy. The Modified Fruit Fly Algorithm (MFFA) is used to choose the best features from the extracted features. Following feature selection, cyber bullying detection is carried out using a Hybrid Recurrent Residual Convolutional Neural Network (HRecRCNN). The experimental outcome of this study indicates the efficiency of the suggested approach. In comparison to current algorithms, the SADL-CDD method delivers improved classification performance with respect to reduced time complexity, greater precision, recall, f-measure, and accuracy.

Download Full-text

Imbalanced sentiment classification based on sequence generative adversarial nets

Journal of Intelligent & Fuzzy Systems ◽

10.3233/jifs-201370 ◽

2020 ◽

Vol 39 (5) ◽

pp. 7909-7919

Author(s):

Chuantao Wang ◽

Xuexin Yang ◽

Linkai Ding

Keyword(s):

Deep Learning ◽

Online Reviews ◽

Classification Performance ◽

Sentiment Classification ◽

Classification Task ◽

Algorithm Optimization ◽

Minority Class ◽

Sample Distribution ◽

Practical Applications ◽

Deep Model

The purpose of sentiment classification is to solve the problem of automatic judgment of sentiment tendency. In the sentiment classification task of text data (such as online reviews), the traditional deep learning model focuses on algorithm optimization, but ignores the characteristics of the imbalanced distribution of the number of samples in each classification, which will cause the classification performance of the model to decrease in practical applications. In this paper, the experiment is divided into two stages. In the first stage, samples of minority class in the sample distribution are used to train a sequence generative adversarial nets, so that the sequence generative adversarial nets can learn the features of the samples of minority class in depth. In the second stage, the trained generator of sequence generative adversarial nets is used to generate false samples of minority class and mix them with the original samples to balance the sample distribution. After that, the mixed samples are input into the sentiment classification deep model to complete the model training. Experimental results show that the model has excellent classification performance in comparing a variety of deep learning models based on classic imbalanced learning methods in the sentiment classification task of hotel reviews.

Download Full-text

Evaluation of Sentiment Analysis via Word Embedding and RNN Variants for Amazon Online Reviews

Mathematical Problems in Engineering ◽

10.1155/2021/5536560 ◽

2021 ◽

Vol 2021 ◽

pp. 1-10

Author(s):

Najla M. Alharbi ◽

Norah S. Alghamdi ◽

Eman H. Alkhammash ◽

Jehad F. Al Amri

Keyword(s):

Feature Extraction ◽

Deep Learning ◽

Sentiment Analysis ◽

Short Term Memory ◽

Online Reviews ◽

Word Embedding ◽

Learning Approaches ◽

Short Term ◽

Term Memory ◽

Long Short Term Memory

Consumer feedback is highly valuable in business to assess their performance and is also beneficial to customers as it gives them an idea of what to expect from new products. In this research, the aim is to evaluate different deep learning approaches to accurately predict the opinion of customers based on mobile phone reviews obtained from Amazon.com. The prediction is based on analysing these reviews and categorizing them as positive, negative, or neutral. Different deep learning algorithms have been implemented and evaluated such as simple RNN with its four variants, namely, Long Short-Term Memory Networks (LRNN), Group Long Short-Term Memory Networks (GLRNN), gated recurrent unit (GRNN), and update recurrent unit (UGRNN). All evaluated algorithms are combined with word embedding as feature extraction approach for sentiment analysis including Glove, word2vec, and FastText by Skip-grams. The five different algorithms with the three feature extraction methods are evaluated based on accuracy, recall, precision, and F1-score for both balanced and unbalanced datasets. For the unbalanced dataset, it was found that the GLRNN algorithms with FastText feature extraction scored the highest accuracy of 93.75%. This result achieved the highest accuracy on this dataset when compared with other methods mentioned in the literature. For the balanced dataset, the highest achieved accuracy was 88.39% by the LRNN algorithm.

Download Full-text

Deep Learning for Aspect-Based Sentiment Analysis on Indonesian Hotels Reviews

Kinetik Game Technology Information System Computer Network Computing Electronics and Control ◽

10.22219/kinetik.v6i3.1300 ◽

2021 ◽

Author(s):

Siwi Cahyaningtyas ◽

Dhomas Hatta Fudholi ◽

Ahmad Fathan Hidayatullah

Keyword(s):

Deep Learning ◽

Sentiment Analysis ◽

Data Preprocessing ◽

Sentiment Classification ◽

Accuracy Score ◽

Learning Methods ◽

Average Accuracy ◽

Quality Of Products

Tourism is one of the fastest-growing industries. Many travelers book hotels and share their experiences using travel e-commerce sites. To improve the quality of products and services, we can take advantage by analyzing their reviews. We can see the good and the bad thing reviews in every aspect of the hotel. However, research to analyze sentiment in every aspect using Indonesian hotel reviews is still relatively new. In this work, we propose to create an Aspect-based Sentiment Analysis (ABSA) using Indonesian hotel reviews to solve the problem. This research consists of four steps: collecting data, preprocessing, aspect classification, and sentiment classification. Our classification process compares with eight deep learning methods (RNN, LSTM, GRU, BiLSTM, Attention BiLSTM, CNN, CNN-LSTM, and CNN-BiLSTM). In aspect classification, we have six classes of aspects which are harga (price), hotel, kamar (room), lokasi (location), pelayanan (service), and restoran (restaurant). In sentiment analysis, we compared two scenarios to classify sentiments as positive or negative. The first one is to classify sentiment in all aspects, and the second one is to classify sentiment in every aspect. The results showed that LSTM achieved the best model for aspect classification with an accuracy value of 0.926. For sentiment classification, our experiments showed that classify sentiment in every aspect achieved a better result than classify sentiment in all aspects. The result showed that the CNN model gets an average accuracy score of 0.904.

Download Full-text

Exploring Consumer Emotions in Pre-Pandemic and Pandemic Times. A Sentiment Analysis of Perceptions in the Fine-Dining Restaurant Industry in Bucharest, Romania

International Journal of Environmental Research and Public Health ◽

10.3390/ijerph182413300 ◽

2021 ◽

Vol 18 (24) ◽

pp. 13300

Author(s):

Jacqueline-Nathalie Harba ◽

Gabriela Tigu ◽

Adriana AnaMaria Davidescu

Keyword(s):

Content Analysis ◽

Sentiment Analysis ◽

Online Reviews ◽

Restaurant Industry ◽

Capital City ◽

Choice Making ◽

Consumer Emotions ◽

Cognitive Choice ◽

Media Sentiment

This research paper aims to analyse how consumer emotions have evolved during the pandemic period in comparison with the pre-pandemic period in relation to restaurant demand in the Romanian fine-dining industry and uses valuable information based on social-media sentiment analysis and content analysis. Focusing on theories of consumer behaviour, the study aims to emphasize how, under the influence of an epidemic crisis caused by an infectious disease, individual behaviour adapts to the “new normal”, embracing a series of changes in the preferences, attitudes, and cognitive choice-making processes. The article takes into account a comparative analysis of the consumer emotions between the pre-COVID-19 pandemic period (2010–2019) and the pandemic period (2020–present), based on the online reviews provided by customers for five fine-dining restaurants from Bucharest, the capital city of Romania: The Artist, Relais & Chateaux Le Bistrot Francais, Casa di David, Kaiamo, and L’Atelier. The research was based on two mining analyses—content analysis and sentiment analysis—and explored the emotional intent of words, with the data being collected from TripAdvisor through web-scrapping. The empirical results defined the fine-dining experience during the pandemic as being associated with the quality of the dishes and also with the quality of the service. The overall consumer sentiment in the direction of the restaurants analyzed is positive. The sentiment research found that throughout the epidemic, the consumers’ attitudes about restaurants deteriorated. In this sense, consumers seem to be less satisfied with the restaurants’ services than before the pandemic. This is another thing that the restaurants had difficulties in when adapting their operations for the pandemic.

Download Full-text

Analytical review of approaches for tonality recognition of Russian text data

Information and Control Systems ◽

10.31799/1684-8853-2020-4-20-30 ◽

2020 ◽

pp. 20-30

Author(s):

Anastasiya Dvoynikova ◽

Alexey Karpov

Keyword(s):

Sentiment Analysis ◽

Online Reviews ◽

Russian Text ◽

Software Systems ◽

Text Data ◽

World Languages ◽

Analytical Review ◽

Audio Data ◽

Quality Of Products

Introduction: In recent years, sentiment analysis has found practical application in many areas, such as evaluating the quality of products and services based on customers’ online reviews, analyzing negative emotions in messages, forecasting stock markets or political situations based on news data. In this regard, a large number of systems and methods for Russian text sentiment analysis are being developed. Purpose: A detailed review of approaches, and comparative analysis of available databases in the field of Russian text sentimental analysis. Results: Our analytical review of the approaches to Russian text data sentiment analysis has shown that there are a large number of ways for preprocessing, vectorization and machine classification of the text data. Studying the available databases shows that the Russian text sentimental analysis is less developed than that for other major world languages. Studying the existing software systems for Russian text analysis reveals their low accuracy compared to English, which can be caused by the sophisticated structure of Russian. Discussion: In our further research, we plan to implement sentiment analysis of spoken speech using audio data. To do this, we will need to obtain a spelling transcription of speech for each speaker.

Download Full-text