AA-CDNB: adaptive autoregressive CAVIAR-dragonfly optimization with Naive Bayes for reason identification

• Published in: Evolutionary intelligence Vol. 11; no. 1-2; pp. 3 - 17
• Main Authors: Patil, Harshali P., Atique, Mohammad
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.10.2018; Springer Nature B.V
• Subjects: Algorithms; Applications of Mathematics; Artificial Intelligence; Bayesian analysis; Bioinformatics; Classification; Computer simulation; Control; Data mining; Engineering; Experimentation; Mathematical and Computational Engineering; Mathematical models; Mechatronics; Optimization; Prediction models; Robotics; Sentiment analysis; Special Issue; Statistical Physics and Dynamical Systems; Text editing; TWMA; Naive Bayes; Reason identification; Dragonfly algorithm; Prediction
• ISSN: 1864-5909; 1864-5917
• DOI: 10.1007/s12065-018-0155-3

Sentiment analysis is the critical process, which generates the subjective information from the text documents that are available online. Literature presents various kinds of task, like sentiment classification, affect classification, reason identification, and predictive analysis and so on, for sentiment analysis. This work brings the reason identification system from the classified sentiment and the affect classes through the automation of the optimization techniques. The proposed adaptive autoregressive conditional autoregressive value at risk-dragonfly optimization with Naive Bayes (AA-CDNB) algorithm finds the reasons behind the sentiments present in the tweets by joining the dragonfly algorithm and the Naive Bayes classifier. Also, the proposed model utilizes the tangential weighted moving average (TWMA) model, for predicting the sentiment reasons to appear shortly. The experimentation of the proposed work utilizes the BITS PILANI tweets database for the simulation and further, the results are compared with various models. The proposed AA-CDNB model has outclassed other models with the values of 1, 0.888, and 0.920, for the sensitivity, specificity, and accuracy metrics, respectively. Also, the results of the TWMA prediction model is compared with the other models based on the error performance, and it is proved that the TWMA model has improved results.