There’s a certain limitation that one’s can explain consumer’s behavior and every unique variation of it, this lead to a different approach on examining consumer’s behavior, the one that focus on the situational influences on behavior[2]. Although the difficulty of obtaining and processing those information is such a monumental task, which in turn affect the accuracy and confident level of the result. The main cause of this problem is that there isn’t any information gathering strandard that allow us to observer and documented every consumer’s opinion on certain issues, not to mention to also record the place and the time when the consumer gave their opinion and it became imposible to do if we wanted to monitored the consumers in large numbers. This research propose a new way of observing, gathering and processing consumer behavior and the situation that they currently in using social media post, especially Twitter’s Tweets by using Lexicon-based sentiment analysis for word detection and hourglass of emotion for emotion detection.
We also understand that emotions is a broad, complex and multi-dimentional terms [3][6]. That’s why this research also propose a new way of detecting and categorizing emotion, by using the lowest level possible for emotion namely its polarity[3]. Using polarity it became easier to categorize each of the sentiments, and also the parameter is easily defined.
Lexicon-based sentiment analysis is a method to seperate senteces down to its building blocks, words. Analyzing the emotion polarity from each of the words is done by using a dictionary called lexicon dictionary that contains negative and positive words with its emotional intensity[12]. The lexicon dictionary in this reseach is based on previous research[5], the collection of Tweets or this research test data is also came from previos research[10]. The situational variable analysis, is done by extracting the situational variable data from each of the tweet. In this research only two situational variable extracted they were location and time. Only those two variables is avaliable inside the Tweet. Analyzing the situational variable is done by counting the mean location and time for each emotion polarities. This method shown that whether those situational variable has a significant effect on one’s emotions or not. This research hope that the result could be used as a references for companies to efficiently promote their products on social media platform based on human emotion. For example we can predict smartphone sales[13] and target certain time of the day to boost that sale base on the user’s emotion. Also combined with state of the art machine learning based consumer buyer pattern[11] we can further improve the accuracy by integrating human emotion.
[1] A. Reda, et al, “Emotion and Sentiment Analysis from Twitter Text”, in Computational Science, 2019.
[2] B. Russel, “Situational Variables and Consumer Behavior”, in Journal of Consumer Research : Volume 2, 1975, pp. 157 – 164.
[3] C. Erick, et al, “The Hourglass of Emotion”, in Cognitive Behavioural System 2011, 2012, pp. 144-157.
[4] C. Lea, and M. Particio, “Emotion Detection from Text : A Survey”, University of Alicante : Spain, 2014.
[5] C. Yanqing, and S. Steven, “Building Sentiment Lexicons for All Major Languages”, in ACL : Volume 2, 2014, pp. 383-389.
[6] F. Johny R.J, et al, “The World of Emotions is not Two Dimensional”, in Psychological Science 2007 : Volume 18, 2007, pp. 1050-1057.
[7] F. Gordon R, and Y. Marie, “Situational Influences on Consumers Attitudes and Behavior”, in Journal of Management : Volume 20, Number 10, 2013, pp. 1 - 31.
[8] G. Jenifer P, et al, “Personality Consistency and Situational Influences on Behaviour”, in Journal of Business Research : Volume 58, 2018, pp. 518 - 525.
[9] S. Kashifa, et al, “Emotion Detection from Text and Speech : A Survey”, in Social Network Analysis and Mining, 2018.
[10] S. Mei, M. Rahmad, and A. Mirna, “Emotion Classification on Indonesian Twitter Dataset”, in Proceeding of International Conference on Asian Language Processing 2018.
[11] S. Saavi, and A. Ognjen, “Machine learning based prediction of consumer purchasing decisions: the evidence and its significance”, in Proceedings AI and Marketing Science workshop AAAI-2018
[12] T. Maite, et al, “Lexicon-Based Methods for Sentiment Analysis”, in Computational Linguistic : Volume 34, Number 2, 2011.
[13] T. Suppawong, amd T. Conrad, “Fad or Here to Stay: Predicting Product Market Adoption and Longevity Using Large Scale”, in Social Media Data, Proceedings of ASME International Design Engineering Technical Conferences & Computers and Information in Engineering Conference 2013.
[14] W. Scott, et al, “The Influence of Situational Varaiables on Brand Personality Choice”, in International Journal of Marketing Studies : Volume 4, Number 5, 2012, pp. 103 – 115.
[15] Y. Ali, et al, “Current State of Text Sentiment Analysis from Opinion to Emotion Mining”, in ACM Computing Surveys : Volume 50, No. 2, Article 25, 2017.