Social Media provides an opportunity for people to share their idea about different aspects of life. Traveling is one of the essential aspects of life. In this paper, we use Bidirectional Encoder Representations from Transformers(BERT) for sentiment classification of online reviews to implement a Convolutional Neural Network model that can recommend airline tickets. Different sentiment classification models are compared based on people’s online reviews from six online social platforms. The new model can classify airline tickets as an economical choice or not economical choice ticket based on various aspects of the trips, including airline customer satisfaction, travel destination, hotel information, restaurants, and tourist attractions. This work contribution is two-fold: first, it examines the importance of choosing a transfer learning model for sentiment classification of online social platforms in the recommender system’s prediction accuracy. Second, the implementation of the Convolutional Neural Network model, which can classify airline tickets based on data generated by multiple online social platforms. Using the natural language processing approach to use the transfer learning model to improve the CNN model’s prediction accuracy is a new approach to use online social platforms to recommend economical airline tickets to consumers.

Using Transfer Learning Approach to Implement Convolutional Neural Network model to Recommend Airline Tickets by Using Online Reviews

Pervasive sensing of people's opinions is becoming critical in strategic decision processes, as it may be helpful in identifying problems and strengthening strategies. A recent research trend is to understand users' opinions through a sentiment analysis of contents published in the Twitter platform. This approach involves two challenges: the large volume of available data and the large variety of used languages combined with the brevity of texts. The former makes manual analysis unreasonable, whereas the latter complicates any type of automatic analysis. Since sentiment analysis is a difficult process for computers, but it is quite simple for humans, in this article, we transform the sentiment analysis process into a game. Indeed, we consider the game with a purpose approach and we propose a game that involves users in classifying the polarity (e.g., positive, negative, neutral) and the sentiment (e.g., joy, surprise, sadness) of tweets. To evaluate the proposal, we used a dataset of 52,877 tweets, we developed a Web-based game, we invited people to play the game, and we validated the results through two different methods: ground-truth and manual assessment. The obtained results showed that the game approach is effective in measuring people' sentiments and also highlighted that participants liked to play the game.

Sentiment analysis and Twitter: a game proposal

Social media popularity and importance is on the increase due to people using it for various types of social interaction across multiple channels. This systematic review focuses on the evolving research area of Social Opinion Mining, tasked with the identification of multiple opinion dimensions, such as subjectivity, sentiment polarity, emotion, affect, sarcasm and irony, from user-generated content represented across multiple social media platforms and in various media formats, like text, image, video and audio. Through Social Opinion Mining, natural language can be understood in terms of the different opinion dimensions, as expressed by humans. This contributes towards the evolution of Artificial Intelligence which in turn helps the advancement of several real-world use cases, such as customer service and decision making. A thorough systematic review was carried out on Social Opinion Mining research which totals 485 published studies and spans a period of twelve years between 2007 and 2018. The in-depth analysis focuses on the social media platforms, techniques, social datasets, language, modality, tools and technologies, and other aspects derived. Social Opinion Mining can be utilised in many application areas, ranging from marketing, advertising and sales for product/service management, and in multiple domains and industries, such as politics, technology, finance, healthcare, sports and government. The latest developments in Social Opinion Mining beyond 2018 are also presented together with future research directions, with the aim of leaving a wider academic and societal impact in several real-world applications.

/pdf/over-a-decade-of-social-opinion-mining-a-systematic-review-341o02yd7b.pdf

Over a Decade of Social Opinion Mining: A Systematic Review

Social networking sites, blogs and many more informal texting sites have taken a center stage in our lives, and in this informal messages we tend to show a lot of emotions towards a certain social cause or towards friends. Quite impressive work is done in emotion analysis or emotion classification area, but we require algorithms to identify emotion strength/intensity from informal and short text messages. Emotion intensity from a short text can help us in identifying the degree of emotion in a sentence belonging to the same emotion category. This paper proposes a novel idea for solving the problem of identification of emotion intensity from social media data.

Emotion intensity detection for social media data

Nowadays, social media have become one of the most important methods of communication that provide a real-time and rich source of information, including sentiments. Understanding the population sentiment is a key goal for organisations and governments. In recent years, quite a lot of research has been done on sentiment analysis from social media. However, all the work in the state of the art is focused on a specific pre-defined subset of tweets, e.g. sentiment analysis via keywords search from tweets for relevant brands, products, services, events and so forth. Monitoring the general sentiment at national level through the whole social media stream is not done due to the challenges of filtering sentiment-irrelevant information, diversity of vocabulary usage in general tweets across topics causing low accuracy and the need for bilingual or multilingual models. This paper proposes a system for general population sentiment monitoring from a social media stream (Twitter), through comprehensive multi-level filters, and our proposed improved latent Dirichlet allocation (LDA) (Wang et al. in ACM Trans Internet Technol 18(1):1–23, 2017; Wang and Al-Rubaie in Appl Soft Comput 33:250–262, 2015; https://patents.google.com/patent/US20170293597A1/en) method for sentiment classification. Experiments show that our proposed improved LDA for sentiment analysis yields the best results, and also validate our proposed system for national sentiment monitoring in Abu Dhabi using twitter.

/pdf/national-happiness-index-monitoring-using-twitter-for-5cm4eto2x3.pdf

National happiness index monitoring using Twitter for bilanguages

In this paper we present a novel approach to Twitter Sentiment Analysis. The approach adopted is to analyse the lexicon features of the tweets for classifying its sentiment (positive, negative and neutral). The training data is made more exhaustive by including various manually labelled tweets, in addition to the existing word stock to keep up with the changing micro logging trends. For Data Preprocessing, a novel spell checking algorithm is introduced, an operation for disjoining compound words such as "high hopes" is implemented and emoticons are replaced by suitable emotion words like happy or sad. After this initial preprocessing, the machine learning algorithms are (Support vector machines and Maximum entropy) are applied. We also propose an avant-garde sentiment scoring mechanism to estimate the degree of the sentiment. Our approach is able to assign sentiments to tweets with an accuracy of 80%.

Twitter Sentiment Analysis -- A More Enhanced Way of Classification and Scoring

MULTI-WINDOW TIME-AWARE PERSONALIZED RECOMMENDATION SYSTEM A multi-window time-aware personalized recommendation system and method is provided. The system includes a memory having computer-readable instructions stored therein. The system further includes a processor configured to access purchase history of one or more users and to select a pair of items purchased by the one or more users based on the purchase history. The processor is further configured to determine a contributory affinity value between a first item and a second item of the selected pair of items for each of the one or more users. The contributory affinity value is determined based on a first time-decay function and a second time-decay function associated with purchase times of the first and second items. In addition, the processor is configured to combine the contributory affinity values for each of the one or more users to estimate the affinity value between the first and second item across multiple time-windows. (Figure 1)

Sanket Sahu

Papers

Twitter Sentiment Analysis -- A More Enhanced Way of Classification and Scoring

Multi-window time-aware personalized recommendation system