Know Thy Context: Parsing Contextual Information from User Reviews for Recommendation Purposes

2022 | Information Systems Research | Citations: 1

Authors: Bauman, Konstantin; Tuzhilin, Alexander

Abstract: In this paper, we study an important problem of parsing contextual information f ... Expand

Abstract: In this paper, we study an important problem of parsing contextual information from user reviews for recommendation purposes. First, we study the ways contextual information is expressed in user reviews and obtain novel insights about it. Among other things, we demonstrate that such type of information tends to appear at the beginning of the review, in longer sentences, in the sentences written in the past tense or using gerund form, and in the sentences referring to some points in time. Second, we propose a novel context parsing method for systematically extracting contextual information from user-generated reviews that relies on the insights obtained in our study. We apply the proposed method to three different Yelp applications (restaurants, hotels, and beauty & spas) and demonstrate that it works well and leads to better recommendation performance than the baseline approaches. Our method systematically extracts more comprehensive sets of relevant contextual variables and corresponding phrases than the baselines. Our analysis also shows the importance of the newly discovered contextual information for recommendation purposes. The obtained results and the proposed method can help to get more comprehensive knowledge about contextual variables in a given application that leads to better recommendations. Collapse

Topics: online review recommender system logistics management user experience user-generated content

Methods: word2vec hierarchical clustering machine learning product review experiment

Cultural differences in hospitality service evaluations: mining insights of user generated content

2022 | Electronic Markets | Citations: 3

Authors: Guan, Chong; Hung, Yu-Chen; Liu, Wenting

Abstract: Online peer-to-peer platforms have been widely utilized to facilitate sharing va ... Expand

Abstract: Online peer-to-peer platforms have been widely utilized to facilitate sharing vast amounts of user-generated content (UGC) in the hospitality industry. However, research on the cultural impacts of the broad spectrum of UGC remains nascent and limited. This study therefore elaborates how the hospitality service dimensions that are reflected in UGC both predict service evaluation and are contingent on a reviewer’s prevailing culture. We analyzed a usable sample of 9,257 hotel reviews from 148 countries with latent Dirichlet allocation and aspect-based sentiment analysis algorithms to identify the top service features and corresponding polarity scores. Machine learning identified three dimensions of hospitality service quality: adaptability, reliable delivery, and tangibles. These three dimensions predict customers’ overall hotel ratings, and the magnitude and direction of their effects depend on alignments of cultural orientations. Following Hofstede’s cultural dimensions, reliable delivery and tangibles align with individualism and uncertainty avoidance cultures. Adaptability aligns with power distance, masculinity, long-term orientation, and indulgence cultures. Collapse

Topics: service quality national culture open source Python online review

Methods: structural equation modeling BERT sentiment analysis computational algorithm machine learning

Theories: cultural dimensions theory SERVQUAL

An Ontology of Emotion Process to Support Sentiment Analysis

2022 | Journal of the Association for Information Systems | Citations: 1

Authors: Storey, Veda; Park, Eun Hee

Abstract: Sentiment analysis is used to mine text data from many sources, including blogs ... Expand

Abstract: Sentiment analysis is used to mine text data from many sources, including blogs, support forums, and social media, in order to extract customers' opinions and attitudes. The results can be used to make important assessments about a customer's attitude toward a company and if and how a company should respond. However, much research on sentiment analysis uses simple classification, where the polarity of a text that is mined is classified as positive, negative, or neutral. This research creates an ontology of emotion process to support sentiment analysis, with an emphasis on obtaining a more fine-grained assessment of sentiment than polarity. The ontology is grounded in a theory of emotion process and consists of concepts that capture the generation of emotion all the way from the occurrence of an event to the resulting behaviors of the person expressing the sentiment. It includes two lexicons: one for affect and one for appraisal. The ontology is applied to posts obtained from customer support forums of large companies to show its applicability in a multilevel evaluation. Doing so provides an example of a complete ontology assessment effort. Collapse

Topics: knowledge representation online review Bing evaluation criteria website

Methods: ontological modelling design artifact sentiment analysis experiment aspect-based sentiment analysis

A Sentiment Analysis of Star-rating: a Cross-Cultural Perspective

2022 | HICSS | Citations: 0

Authors: Wan, Yun; Nakayama, Makoto

Abstract: Consumer-generated reviews and ratings are critical for the tourism industry. Th ... Expand

Abstract: Consumer-generated reviews and ratings are critical for the tourism industry. The star rating distribution of services can significantly influence a consumer’s decision-making and choice of services. We analyze the star rating distribution of restaurant reviews from three nations (Japan, China, and the U.S.) and find two distribution patterns: bimodal and unimodal. Then, we analyze the sentiment correlation with each star rating across the three cultures. We find the inconsistency of positive sentiment correlation with 5- and 4-star ratings generated by Japanese consumers. Possible contributing factors, including biases, national culture, and socioeconomic conditions, are discussed. Collapse

Topics: online review user-generated content national culture website

Methods: product review correlation analysis sentiment analysis aspect-based sentiment analysis

Theories: cognitive dissonance theory

Supporting Product Development by a Trend Analysis Tool Applying Aspect-Based Sentiment Detection

2022 | DESRIST | Citations: 0

Authors: Wörner, Janik; Konadl, Daniel; Schmid, Isabel; Leist, Susanne

Abstract: Incorporating product trends into innovation processes is imperative for compani ... Expand

Abstract: Incorporating product trends into innovation processes is imperative for companies to meet customers’ expectations and to stay competitive in fiercely opposing markets. Currently, aspect-based sentiment analysis has proven an effective approach for investigating and tracking towards products and corresponding features from social media. However, existing trend analysis tools on the market that offer aspect-based sentiment analysis capabilities, do not meet the requirements regarding the use case Product Development. Therefore, based on these requirements, we implemented an artifact by following the design science research. We applied our tool to real-world social media data (37,638 Yelp reviews) from one major fast-food restaurant in the US, and thereby demonstrated that our tool is capable of identifying remarkable and fine-grained product trends. Collapse

Topics: new product development marketing management spreadsheet social media innovation management

Methods: design principle design theory aspect-based sentiment analysis theoretical contribution design artifact

Consumer preference analysis based on text comments and ratings: A multi-attribute decision-making perspective

2022 | Information & Management | Citations: 0

Authors: Zhu, Bin; Guo, Dingfei; Ren, Long

Abstract: With the popularity of social media, extracting consumer preferences from online ... Expand

Abstract: With the popularity of social media, extracting consumer preferences from online consumer-generated content is of vital importance for product/service providers to develop tailored marketing strategies. However, existing approaches face difficulties analyzing consumer preferences over different attributes of alternatives (restaurants, hotels, etc.), which hinders product/service providers from comprehensively understanding consumer choice decisions. To address this issue, we solve for the consumer preferences over the attributes represented by attribute weights based on consumers’ historical data, including text comments and overall ratings. Specifically, for each comment and a corresponding rating, we first employ sentiment analysis to calculate values of the attributes, and then develop a quadratic programming model to solve for the weights. Based on a stream of a consumer’s text comments and overall ratings, we can correspondingly obtain a stream of weights indexed by the comment time. We then model this stream of weights as hesitant judgments and employ a hesitant multiplicative programming method to solve for the final weights that fit the consumer’s preferences over attributes at the highest satisficing level. In the application of recommendation, our approach not only provides insights into the consumer’s preferences but also has higher prediction power compared with some state-of-the-art methods. Collapse

Topics: recommender system artificial intelligence

Methods: sentiment analysis optimization model word embedding non linear optimization historical review

Theories: theory of bounded rationality

Unveiling the Hidden Truth of Drug Addiction: A Social Media Approach Using Similarity Network-Based Deep Learning

2021 | Journal of Management Information Systems | Citations: 0

Authors: Xie, Jiaheng; Zhang, Zhu; Liu, Xiao; Zeng, Daniel

Abstract: Opioid use disorder (OUD) is an epidemic that costs the U.S. healthcare systems ... Expand

Abstract: Opioid use disorder (OUD) is an epidemic that costs the U.S. healthcare systems $504 billion annually and poses grave mortality risks. Existing studies investigated OUD treatment barriers via surveys as a means to mitigate this opioid crisis. However, the response rate of these surveys is low due to social stigma around opioids. We explore user-generated content in social media as a new data source to study OUD. We design a novel IT system, SImilarity Network-based DEep Learning (SINDEL), to discover OUD treatment barriers from patient narratives and address the challenge of morphs. SINDEL significantly outperforms state-of-the-art NLP models, reaching an F1 score of 76.79 percent. Thirteen types of treatment barriers were identified and verified by domain experts. This work contributes to information systems with a novel deep-learning-based approach for text analytics and generalized design principles for social media analytics methods. We also unveil the hurdles patients endure during the opioid epidemic. Collapse

Topics: social media societal evaluation criteria social media analytics electronic health decision support system

Methods: deep learning word embedding machine learning word2vec cluster analysis

Increasing Consumer Engagement in Social Media: The Moderating Role of Product Lifecycle

2020 | European Conference On Information Systems | Citations: 0

Authors: Eslami, Pouyan; Ghasemaghaei, Maryam; Hassanein, Khaled

Abstract: Social media usage has been proliferated among different groups of individuals. ... Expand

Abstract: Social media usage has been proliferated among different groups of individuals. This proliferation has provided an opportunity for firms to get connected with their potential consumers, to introduce and promote their products, and to enhance their consumer engagement in social media. In this study, a research model has been proposed which investigates the impact of influencer popularity, discourse logic, and argument frame on consumer engagement in social media. Moreover, this study investigates the moderating role of product lifecycle and product involvement on the above relationships. The proposed research model will be examined using secondary data from three different social media platforms of Facebook, Instagram and Twitter on different groups of products (i.e., product lifecycle stages and product involvements) and social media users (i.e., influencers and ordinary users). Finally, a series of post-hoc analyses will be conducted to understand if social media platforms vary in terms of consumer engagements. Collapse

Topics: social media Facebook Instagram Twitter social influence

Methods: sentiment analysis partial least squares regression aspect-based sentiment analysis structural equation modeling

Theories: social influence theory

A Weakly Supervised WordNet-Guided Deep Learning Approach to Extracting Aspect Terms from Online Reviews

2020 | ACM Transactions on Management Information Systems | Citations: 0

Authors: Tao, Jie; Zhou, Lina

Abstract: The unstructured nature of online reviews makes it inefficient and inconvenient ... Expand

Abstract: The unstructured nature of online reviews makes it inefficient and inconvenient for prospective consumers to research and use in support of purchase decision making. The aspects of products provide a fine-grained meaningful perspective for understanding and organizing review texts. Traditional aspect term extraction approaches rely on discrete language models that treat words in isolation. Despite that continuous-space language models have demonstrated promise in addressing a wide range of problems, their application in aspect term extraction faces significant challenges. For instance, existing continuous-space language models typically require large collections of labeled data, which remain difficult to obtain in many domains. More importantly, previous methods are largely data driven but overlook the role of human knowledge in guiding model development. To address these limitations, this study designs and develops weakly supervised WordNet-guided deep learning to aspect term extraction. The approach draws on deep-level semantic information from WordNet to guide not only the selection representative seed terms but also the pruning of aspect candidate terms. The weak supervision is provided by a very small set of labeled data. We conduct a comprehensive evaluation of the proposed method using both direct and indirect methods. The evaluation results with Yelp restaurant reviews demonstrate that our proposed method consistently outperforms all baseline methods including discrete models and the state-of-the-art continuous-space language models for aspect term extraction across both direct and indirect evaluations. The research findings have broad research, technical, and practical implications for various stakeholders of online reviews. Collapse

Topics: knowledge representation online review organizational productivity user-generated content decision making

Methods: language model WordNet structural equation modeling word embedding deep learning

Social media analytics: Extracting and visualizing Hilton hotel ratings and reviews from TripAdvisor

2019 | International Journal of Information Management | Citations: 8

Authors: Chang, Yung-Chun; Ku, Chih-Hao; Chen, Chun-Hung

Abstract: Analyzing and extracting insights from user-generated data has become a topic of ... Expand

Abstract: Analyzing and extracting insights from user-generated data has become a topic of interest among businesses and research groups because such data contains valuable information, e.g., consumers’ opinions, ratings, and recommendations of products and services. However, the true value of social media data is rarely discovered due to overloaded information. Existing literature in analyzing online hotel reviews mainly focuses on a single data resource, lexicon, and analysis method and rarely provides marketing insights and decision-making information to improve business’ service and quality of products. We propose an integrated framework which includes a data crawler, data preprocessing, sentiment-sensitive tree construction, convolution tree kernel classification, aspect extraction and category detection, and visual analytics to gain insights into hotel ratings and reviews. The empirical findings show that our proposed approach outperforms baseline algorithms as well as well-known sentiment classification methods, and achieves high precision (0.95) and recall (0.96). The visual analytics results reveal that Business travelers tend to give lower ratings, while Couples tend to give higher ratings. In general, users tend to rate lowest in July and highest in December. The Business travelers more frequently use negative keywords, such as “rude,” “terrible,” “horrible,” “broken,” and “dirty,” to express their dissatisfied emotions toward their hotel stays in July. Collapse

Topics: online review visual analytics Google+ cloud computing Tableau software

Methods: product review exploratory data analysis support vector machine sentiment analysis machine learning