Fuzzy Twin SVM based-Profile Categorization approach

Mabrouk, Olfa; Hlaoua, Lobna; Omri, Mohamed Nazih

doi:10.1109/fskd.2018.8687216

Cited by 10 publications

(2 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As can see multinomial the best model where win 11 times than other models. While Figures 3,4,5,6 show the highest accuracy among four models of NB and compares it with word count. The accuracy of TF Multinomial was 85.65 at 17 000 words, TF Bernoulli 85.34 at 11 000 words, TF Complement 85.65 at 17 000 words, and finally got a TF Gaussian accuracy of 81.05 at 1000 words.…”

Section: Dataset Descriptionmentioning

confidence: 93%

“…When reading a restaurant review, a well-trained SA suite would deduce that the individual critic had a positive opinion about the cuisine but had a service poor opinion. The amount of information that are subjective are available on the Internet has increased significantly during the previous two decades [4] [5]. The numbers of websites have sprung up because the users can create and collaborate whatever they wish [6].…”

Section: Context and Issuesmentioning

confidence: 99%

See 1 more Smart Citation

Machine Learning-based Opinion Extraction Approach from Movie Reviews for Sentiment Analysis

Jassim

Hamed

Omri

2022

Preprint

Self Cite

View full text Add to dashboard Cite

The field of sentiment mining, also known as sentiment mining,sentiment analysis, sentiment mining, sentiment extraction, etc., has seen asurge in the university environment. Various ways to automate the processof sentiment analysis have been tested by researchers in machine learning,data mining, natural language processing, and other fields. They examine thefeelings embedded in people’s opinions and beliefs that affect multiple areas,including companies’ services and products. A movie review has to go throughmany processes to be able to detect and name feelings and achieve greateraccuracy. Due to the structure of the language, the difficulties have been increased, its grammar and dictionary management. As part of this work, wepropose a new approach for extracting words from a specific text and thenclassifying them. Thanks to a phase of pre-processing and extraction of wordsdepending on the frequency, our approach makes it possible to select between500 and 20,000 words with a vectorial construction by applying the techniquesof ”Term Frequency” (T F) and that of the ”Term frequency-Inverse DocumentFrequency” (T F − IDF). Four different Naive Bayes models were thus considered and used (Complement, Multinomial, Bernoulli and Gaussian). We evaluated our proposed approach against different standard measures namelyprecision, accuracy, recall, F1 − score and kappa. The results revealed thatthe Na¨ıve Bayes multinomial system obtained the best results with an accuracyof 86.46%.

show abstract

Section: Dataset Descriptionmentioning

confidence: 93%