ORIGINAL_ARTICLE
NB vs. SVM: A Contrastive Study for Sentiment Classification on Two Text Domains
Thanks to the availability of websites like Twitter, user-generated content is being published on the Internet every second. Sentiment Classification is one of the most attractive fields in text mining, which classifies reviews into positive and negative classes. Pre-processing is an important goal when these textual contexts are employed through machine learning techniques. Without effective pre-processing methods, inaccurate results will be achieved. This article aims to investigate the role of pre-processing in the Sentiment Classification problem. The main idea in this paper comes from using sampling techniques. This paper suggests classifying the tweets and reviews using supervised classifiers. We applied a set of pre-processing stages consisting of n-grams and samplings on two well-known datasets. Our results are worthwhile for companies to monitor the people's sentiment about their brands and for many other applications. We have provided further evidence to confirm the superiority of our model. Experimental results reveal that the proposed model outperforms the existing methods and can improve the performance of Sentiment Classification in terms of accuracy, precision, recall, and F1 criteria.
https://journal.research.fanap.com/article_130674_f1eb81de64a5d717e38f9f58312752a4.pdf
2021-06-01
1
12
10.22034/jaisis.2021.279225.1025
Text Mining
Sentiment Classification
Supervised methods
Movie Reviews
Twitter
Razieh
Asgarnezhad
razyehan@gmail.com
1
Department of Computer Engineering, Isfahan (Khorasgan) Branch, Islamic Azad University, Isfahan, Iran
LEAD_AUTHOR
S. Amirhassan
Monadjemi
monadjemi@eng.ui.ac.ir
2
Senior Lecturer, School of continuing and lifelong education, National University of Singapore, 119077, Singapore.
AUTHOR
ORIGINAL_ARTICLE
The Market Structure and Pricing Models of Cloud Services
The macroeconomic policies of developing counteries have repeatedly emphasized the creation of a suitable platform for the growth and emergence of knowledge-based businesses and domestic production. Using cloud services, among others, could flourish ecommerce industry for the related countries and skyrocket the revenue for information and communication technology companies. In this regard, governments will consider supportive agenda for private copmponies to enter the lucrative market of cloud services. However, this is not enough and such componies require to have profound understanding over the market structure and therby the performance of the coresponding main players. Specifically, financial issues and the way of absorbing the potential consumers through adopting appropriate business models are of paramount importance. This study works as a guide for the new followers of the market by identifing the Oligopoly structure of the cloud services market and introducing the pricing models of the cloud services.
https://journal.research.fanap.com/article_132361_f1f60a8f9145602fa29f86cb75045f02.pdf
2021-06-01
13
21
10.22034/jaisis.2021.285664.1027
Cloud services
Oligopoly
Business analysis
Pricing models
Edris
Salehi
e.salehi@um.ac.ir
1
Department of Economics, Yazd University, Yazd, Iran
LEAD_AUTHOR
Seyed Nezamudin
Makiyan
nmakiyan@yazd.ac.ir
2
Associate Professor, Economics Department, Yazd University
AUTHOR
ORIGINAL_ARTICLE
S&P500 Index Direction Prediction Using Textual Tweets and Their Corresponding Sentiment
In this paper, a novel method is proposed to predict the direction of Standard & Poor 500 (S&P500) index using the tweets in this regard as well as the index amount from the day before. At the beginning, using a dataset of all tweets and their corresponding posting times about S&P500 index, companies and securities are considered as features of the study. Next, these feature vectors are assigned three different labels based on the direction of the index change from the day before and whether the change is significant enough, creating a classification problem. Building a sentiment analysis tool based on T5 transformer which attempts to combine all the downstream tasks into a text-to-text format, sentiment feature is added to each tweet in the dataset. Lastly, after balancing the data and preprocessing the textual information through an NLP pipeline, a deep neural network is proposed to classify the processed data. It is shown that using the tweets and their corresponding sentiments, the proposed method for movement direction prediction of the S&P500 index outperformed other existing models.
https://journal.research.fanap.com/article_135403_775c4f187c40181e2720e9fc5fc5eaec.pdf
2021-06-01
22
28
10.22034/jaisis.2021.290661.1029
Natural Language Processing
Predictive Modeling
Sentiment analysis
deep neural networks
Transformer Learning
Parman
Mohammadalizadeh
p.mohammadalizadeh@eng.basu.ac.ir
1
Artificial Intelligence and Robotics
AUTHOR
Mohammadjavad
Jafari
mohammad.jafari@srbiau.ac.ir
2
Science and Research Branch, Islamic Azad University, Tehran, Iran
LEAD_AUTHOR
ORIGINAL_ARTICLE
Attention-Based and Positional-Aware Neural Networks for Next-Item Recommendation
Next-item recommendation intends to predict user interest over sequential items given user historical behaviors. There has been a lot of past works for the next-item recommendation, according to Markov Chains(MCs) and Recurrent Neural Networks(RNNs). MCs perform best in sparse datasets and RNNs perform better in denser datasets. To improve these methods, several recommendation systems have been built on MC and RNN architectures. However, these methods struggle to consider long-range dependencies and uncover complex relationships in next-item recommendation. Due to the limitations of these methods, we apply self-attention using neural networks. The proposed method can consider long-range dependencies using self-attention. Also, the proposed method can uncover complex relationships to capture an efficient features representation using neural networks such as long short-term memory(LSTM), Bi-Directional LSTM and convolutional neural network(CNN) presenting for sequence modeling. In this paper, to choose the best model, several neural network models are evaluated and the best model is selected regarding the performance of self-attention-based neural networks. At each time step, this method tries to identify which items are ‘relevant’ from a user’s action history and apply them to predict the next item. We are able to achieve a high accuracy rate and significantly outperform state-of-the-art next-item methods on sequential datasets.
https://journal.research.fanap.com/article_139007_eabf3cbd060fb46d5bb732a39e45b640.pdf
2021-06-01
29
40
10.22034/jaisis.2021.304225.1033
recommendation system
Deep Learning
sequential data
user behavior
attention model
Nemat
Saeidi
saeidi.n@eng.ui.ac.ir
1
Department of Artificial Intelligence Engineering, University of Isfahan, Isfahan, Iran
LEAD_AUTHOR
Hossein
Shahamat
h.shahamat@modares.ac.ir
2
Department of Electrical and Computer Engineering, Tarbiat Moders University, Tehran, Iran
AUTHOR
ORIGINAL_ARTICLE
ML Revolution In NLP: A Review Of Machine Learning Techniques In Natural Language Processing
In the current era, usually, people communicate with each other through the internet. This platform opens this opportunity for computer scientists to access huge information about human languages. However, this big data is unstructured from the computational point of view. Thus, computer scientists developed Natural Language Processing (NLP) methods for analyzing human language data by computers. Indeed, NLP is a way to analyze human language messages by computerized methods. On the other hand, due to the high capabilities of Machine Learning (ML), many researchers incorporated this approach in language processing techniques to improve the performance of NLP systems. In this paper, we aim to present a summarized review of the NLP techniques. considering the importance of the ML approach. Firstly, this paper introduces basic terminology for NLP. Secondly, according to the importance of ML history, the studied techniques are categorized into three groups: old-fashioned, conventional, and modern methods. The presented review in this study could be beneficial for ML and NLP reseachers in order to develop new ML techniques for NLP tasks.
https://journal.research.fanap.com/article_139143_e73ed1b90626c4f5c2457a88dfc6fccd.pdf
2021-06-01
41
48
10.22034/jaisis.2021.306194.1034
Human language data
Natural Language Processing
Machine Learning
Majid
Abedinzadeh Shahri
m.abedinzadeh@ut.ac.ir
1
School of Electrical and Computer Engineering, University of Tehran, Tehran, Iran
LEAD_AUTHOR
Rasool
Taban
rasooltaban@gmail.com
2
Center for Computational and Stochastic Mathematics, University of Lisbon, Portugal
AUTHOR
ORIGINAL_ARTICLE
The Role of Data Governance in Improving Urban Management: Case Study of Tehran Municipality
Today's cities face various challenges, including economic growth, environmental sustainability, social resilience, air pollution, and traffic. Given these challenges, we need information for analysis and forecasting to manage crises in specific situations. Many cities are investing in information technology research and developing policies to improve citizens' quality of life. Given the ICT trend for sustainable smart cities, the future planning process of the city is critical. Therefore, timely and reliable primary input is critical for decision-making analysis in real-time and reliable statistics, which means increasing the quality and quantity of data in all aspects. In this paper, the experiences of Tehran urban statistics and observatory center in providing open data is presented with regard to the implementation of anonymization processes using data governance measures. We describe the anonymization of some real data from municipality, which has been done by PDPC and present a checklist for classifying all information. Finally, the design of metadata storage is clarified. Essentially, one of the successful experiences is the use of researchers' analysis by providing no sensitive data as open data. This experience is based on studying global examples with the framework of DMBOK and required privacy and confidentiality of information. We used data governance in the organization to emphasize on all aspects of data management, including data production and sharing, data quality assurance, data integration, and related topics.
https://journal.research.fanap.com/article_139812_f6442e4fa2b7bd0d5838c72efb4769e4.pdf
2021-06-01
49
59
10.22034/jaisis.2021.299011.1032
Data Governance
Open data
data anonymity
Transparency
Sara
Bourbour Hosseinbeigi
bourbour-s@tehran.ir
1
Tehran Urban Statistics and Observatory Center
AUTHOR
Shirin
Hamed Ahangari
s.hcom9@gmail.com
2
Tehran Municipality ICT Organization
LEAD_AUTHOR