• PDF / 2,076,193 Bytes
• 29 Pages / 595.276 x 790.866 pts Page_size
• 43 Downloads / 235 Views

DOWNLOAD

REPORT

---

(0123456789().,-volV)(0123456789().,-volV)

Swarm optimization clustering methods for opinion mining Ellen Souza1,2 • Diego Santos1,2 • Gustavo Oliveira1 • Alisson Silva1 • Adriano L. I. Oliveira1

Ó Springer Science+Business Media B.V., part of Springer Nature 2018

Abstract Supervised machine learning and opinion lexicon are the most frequent approaches for opinion mining, but they require considerable effort to prepare the training data and to build the opinion lexicon, respectively. In this paper, a novel unsupervised clustering approach is proposed for opinion mining. Three swarm algorithms based on Particle Swarm Optimization are evaluated using three corpora with different levels of complexity with respect to size, number of opinions, domains, languages, and class balancing. K-means and Agglomerative clustering algorithms, as well as, the Artificial Bee Colony and Cuckoo Search swarm-based algorithms were selected for comparison. The proposed swarm-based algorithms achieved better accuracy using the word bigram feature model as the pre-processing technique, the Global Silhouette as optimization function, and on datasets with two classes: positive and negative. Although the swarm-based algorithms obtained lower result for datasets with three classes, they are still competitive considering that neither labeled data, nor opinion lexicons are required for the opinion clustering approach. Keywords Opinion mining  Opinion clustering  Text clustering  Swarm optimization  Twitter

1 Introduction The growth of user-generated text on micro-blogs, social media, and e-commerce websites provides a massive quantity of data that allows discovering the experiences, opinions, and feelings of electors, fans, customers, and others (Marine-Roig and Anton Clave´ 2015). These electronic Word of Mouth statements expressed on the web are prevalent in the business and service industries to enable

& Ellen Souza [email protected]; [email protected] Diego Santos [email protected] Gustavo Oliveira [email protected] Alisson Silva [email protected] Adriano L. I. Oliveira [email protected] 1

Center of Informatics, Federal University of Pernambuco (CIn-UFPE), Recife, Pernambuco, Brazil

2

MiningBR Research Group, Federal Rural University of Pernambuco (UFRPE), Recife, Pernambuco, Brazil

customers to share their point of view (Ravi and Ravi 2015). In order to enhance the acquisition of a product or service and to improve the user satisfaction, most websites provide the opportunity for users to write reviews. On the other hand, customers identify online reviews as having a significant influence on their purchase in various economic sectors: 87% of customers consider those reviews on their purchase in the hotel sector, 84% in the travel sector, 79% for restaurants, 79% in the legal sector, 78% in the automotive sector, 76% in the medical sector, and 73% for home purchasing (ComScore 2016). Since it is a rich source of real-time information, there has been an increasing interest in the scientific community to create systems capable of ex