
Improved Accuracy of Sentiment Analysis Movie Review Using Support Vector Machine Based Information Gain
Author(s) -
Reza Maulana,
Panny Agustia Rahayuningsih,
Windi Irmayani,
Dedi Saputra,
Wanty Eka Jayanti
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1641/1/012060
Subject(s) - support vector machine , computer science , information gain , sentiment analysis , information gain ratio , artificial intelligence , value (mathematics) , relevance vector machine , feature selection , selection (genetic algorithm) , machine learning , convergence (economics) , data mining , vector space model , algorithm , pattern recognition (psychology) , economics , economic growth
The quality of a movie can be known from the opinions or reviews of previous audiences. This classification of reviews is grouped into positive opinions and negative opinions. One of the data mining algorithms that are most frequently used in research is the Support Vector Machine because it works well as a method of classifying text but has a very sensitive deficiency in the selection of features. The Information Gain method as feature selection can solve problems faster and more stable convergence levels. After testing on two movie review datasets are Cornell and Stanford datasets. The results obtained on the Cornell dataset is the Support Vector Machine algorithm to produce an accuracy of 83.05%, while for the Support Vector Machine based on Information Gain, the accuracy value is 85.65%. Increased accuracy reached 2.6%. Then, the results obtained on the Stanford dataset is the Support Vector Machine algorithm yields a value of 86.46%, while for the Support Vector Machine based on Information Gain, the accuracy value is 86.62%. Increased accuracy reached 0.166%. Support Vector Machine based Information Gain on the problem of movie review sentiment analysis proved to provide more accurate value.