ANALISIS SENTIMEN OPINI PUBLIK BERITA KEBAKARAN HUTAN MELALUI KOMPARASI ALGORITMA SUPPORT VECTOR MACHINE DAN K-NEAREST NEIGHBOR BERBASIS PARTICLE SWARM OPTIMIZATION
ANALISIS SENTIMEN OPINI PUBLIK BERITA KEBAKARAN HUTAN MELALUI KOMPARASI ALGORITMA SUPPORT VECTOR MACHINE DAN K-NEAREST NEIGHBOR BERBASIS PARTICLE SWARM OPTIMIZATION
复制标题
分析情绪 公众意见 BERITA KEBAKARAN HUTAN MELALUI KOMPARASI 算法 支持向量机 DAN K 最近邻 BERBASIS 粒子群优化
DOI：
10.33480/pilar.v13i1.153
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
Lilyani Asri Utami
中科院分区：
文献类型：
--
作者：
Lilyani Asri Utami
情感分析是确定基于文本的数据集的内容是积极的还是消极的过程。目前，公众舆论是一个人在寻求解决方案的决定中的重要资源。支持向量机（SVM）和K-最近邻（K-NN）等分类算法被许多研究者提出用于评论意见的情感分析。本研究的问题是选择特征选择来提高支持向量机（SVM）和K-近邻（K-NN）对森林火灾新闻舆情评论的准确率，并比较其最高的准确率。比较算法，SVM产生的准确率为80.83%和AUC 0.947，然后与基于PSO的SVM的准确率为87.11%和AUC 0.922进行比较。K-NN算法的测试结果数据准确率为85.00%，AUC为0.918，然后比较基于k-NN的PSO的准确率为73.06%，AUC为0.500。测试结果表明，粒子群算法可以提高SVM的精度，但不能提高K-NN算法的精度。基于粒子群算法的支持向量机算法为森林火灾新闻评论的分类问题提供了解决方案，使之更加准确和优化。
Sentiment analysis is a process to determine the content of text-based datasets which are positive or negative. At present, public opinion be an important resource in the decision of a person in finding a solution. Classification algorithms such as Support Vector Machine (SVM) and K-Nearest Neighbor (K-NN) is proposed by many researchers to be used in sentiment analysis for review opinion. The problem in this research is the selection of feature selection to improve accuracy values Support Vector Machine (SVM) and K-Nearest Neighbor (K-NN) and compare the highest accuracy for sentiment analysis review public opinion about the news of forest fires. The comparison algorithms, SVM produces an accuracy of 80.83% and AUC 0.947, then compared with SVM based on PSO with an accuracy of 87.11% and AUC 0.922. The test result data for K-NN algorithm accuracy was 85.00% and the AUC 0.918, then compared for accuracy by k-NN-based PSO amounted to 73.06% and the AUC 0.500. The results of the testing of the PSO algorithm can improve the accuracy of SVM, but are not able to improve the accuracy of the algorithm K-NN. SVM algorithm based on PSO proven to provide solutions to the problems of classification review news opinion forest fires in order to more accurately and optimally.