Restaurant Industry Revival or Death: How Do Reviews Change During COVID-19?

2023 | Americas Conference on Information Systems | Citations: 0

Authors: Koohikamali, Mehrdad

Abstract: Yelp is an online social network service in which consumers provide reviews on ... Expand

Abstract: Yelp is an online social network service in which consumers provide reviews on their local business restaurants. Because our data is in the form of human review text, this study utilizes natural language processing (NLP) and machine learning (ML) to vectorize the text reviews and apply sentiment analysis through logistic regression. After optimizing the model parameters, sentiments of popular restaurants in Los Angeles before and during COVID-19 are analyzed. Latent Semantic Analysis (LSA) and topic modeling used to gather the topics of reviews and logistic regression is used for feature importance. Data is extracted from available Yelp reviews of restaurants and includes total of 10,390 reviews. Results explain a difference in the number of reviews, sentiments, and the content before and during the COVID-19. This study provides insights for both theory and practice and shed light into the restaurant industry challenges during the global pandemic. Collapse

Topics: pandemic

Methods: natural language processing logistic regression sentiment analysis Yelp review machine learning

sDTM: A Supervised Bayesian Deep Topic Model for Text Analytics

2023 | Information Systems Research | Citations: 0

Authors: Yang, Yi; Zhang, Kunpeng; Fan, Yangyang

Abstract: Topic modeling methods such as latent Dirichlet allocation (LDA) are powerful to ... Expand

Abstract: Topic modeling methods such as latent Dirichlet allocation (LDA) are powerful tools for analyzing massive amounts of textual data. They have been used extensively in information systems (IS) and business discipline research to identify latent topics for data exploration and as a feature engineering mechanism to derive new variables for analyses. However, existing topic modeling approaches are mostly unsupervised and only leverage textual data, while ignoring additional useful metadata often associated with text, such as star ratings in customer reviews or categories of posts in online forums. As a result, the identified topics and variables derived based on the learned topic model may not be accurate, which could lead to incorrect estimations that affect subsequent empirical analysis and to inferior performance on predictive tasks. In this study, we propose a novel supervised deep topic modeling approach called sDTM, which combines a neural variational autoencoder model and a recurrent neural network. sDTM leverages the auxiliary data associated with text to enhance the topic modeling capability. We conduct empirical case studies and predictive analytics on an online consumer review data set and an online knowledge community data set. Experimental results show that in comparison with benchmark methods, sDTM can enhance both the empirical estimation and predictive performance. sDTM makes methodological contributions to the IS literature and has direct relevance for research using text analytics.History: Alok Gupta, Senior Editor; Huimin Zhao, Associate Editor.Supplemental Material: The online appendix is available at https://doi.org/10.1287/isre.2022.1124. Collapse

Topics: online review open source user-generated content Wikipedia accounting

Methods: topic model deep learning machine learning BERT data transformation

Shared Prosperity (or Lack Thereof) in the Sharing Economy

2022 | Information Systems Research | Citations: 13

Authors: Alyakoob, Mohammed; Rahman, Mohammad S.

Abstract: This paper examines the potential economic spillover effects of a home-sharing p ... Expand

Abstract: This paper examines the potential economic spillover effects of a home-sharing platform—Airbnb—on the growth of a complimentary local service—restaurants. By circumventing traditional land-use regulations and providing access to underutilized inventory, Airbnb attracts visitors to locales that are not traditional tourist destinations. Although visitors generally bring significant spending power, it is unclear whether visitors use Airbnb primarily for lodging and thus do not contribute to the adjacent economy. To evaluate this, we focus on the impact of Airbnb on the restaurant employment growth across locales in New York City (NYC). Specifically, we focus on areas in NYC that did not attract a significant tourist volume prior to the emergence of a home-sharing service. Our results indicate a salient and economically significant positive spillover effect on restaurant job growth in an average NYC locality. A one-percentage-point increase in the intensity of Airbnb activity (Airbnb reviews per household) leads to approximately 1.7% restaurant employment growth. Since home-sharing visitors are lodging in areas that are not accustomed to tourists, we also investigate the demographic and market-structure-related heterogeneity of our results. Notably, restaurants in areas with a relatively high number of White residents disproportionately benefit from the economic spillover of Airbnb activity, whereas the impact in majority Black areas is not statistically significant. We validate the underlying mechanism behind the results by evaluating the impact of Airbnb on Yelp visitor reviews, revealing that areas with increased Airbnb activity experience a surge in their share of NYC visitor reviews. This result is further validated by evaluating the impact of a unique Airbnb neighborhood-level exogenous policy recently implemented in New Orleans. Collapse

Topics: Airbnb sharing platform criminality online review open data

Methods: product review Yelp review longitudinal research propensity score matching propensity score method

Theories: theory of economic growth theory of economic development

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

Fraudulent review detection model focusing on emotional expressions and explicit aspects: investigating the potential of feature engineering

2022 | Decision Support Systems | Citations: 5

Authors: Kumar, Ajay; Gopal, Ram D.; Shankar, Ravi; Tan, Kim Hua

Abstract: Reading customer reviews before purchasing items online has become a common prac ... Expand

Abstract: Reading customer reviews before purchasing items online has become a common practice; however, some companies use machine learning (ML) algorithms to generate false reviews in order to create positive brand images of their own products and negative images of competitors' offerings. Existing techniques use review content to identify fraudulent reviewers; however, spammers become more intelligent, started to learn from their mistakes, and changed their tactics in order to avoid detection techniques. Thus, investigating fraudulent ac­ counts' behaviour of generating fake negative or positive reviews for competitors or themselves and the necessity of ML classifiers to identify fraudulent reviews, is more important than ever. In this research, we present a novel feature engineering approach in which we (1) extract several “review-centric” and “reviewer-centric” features from a dataset; (2) combine the cumulative effects of features distributions into a unified model that represents overall behavior of the fraudulent reviewers; (3) investigate the role of effective data pre-processing to improve detection accuracy; and (4) develop a probabilistic approach to detect fraudulent reviewers by learning a novel M-SMOTE model over a derived balanced dataset and feature distributions, which outperforms other ML models. Our study contributes to the literature on digital platforms and fraudulent review detection with significant managerial and theoretical implications through these novel findings. Collapse

Topics: online review electronic commerce website marketing management logistics management

Methods: feature engineering machine learning computational algorithm XGBoost product review

Healthy reviews!- Online physician ratings reduce healthcare interruptions

2022 | International Conference on Information Systems | Citations: 0

Authors: Kummer, Michael; Laitenberger, Ulrich; Hughes, Danny; Rich, Cyrus; Ayer, Turgay

Abstract: We show that review platforms reduce healthcare interruptions for patients look ... Expand

Abstract: We show that review platforms reduce healthcare interruptions for patients looking for a new physician. We employ a difference-in-differences strategy using physician retirements as a "disruptive shock" that forces patients to find a new physician. We combine insurance claims data with web-scraped physician reviews and highlight a substantial care-gap resulting from a physician's retirement. We then show that online physician reviews reduce this gap and help patients find a new physician faster. Our results are robust to including a variety of controls and various instruments for the availability of physician reviews, but are not found for patients of nonretiring physicians. By reducing interruptions in care, reviews can improve clinical outcomes and lower costs. Collapse

Topics: online review technology adoption database system information system use health information system

Methods: experimental group robustness check Yelp review instrumental variables estimation descriptive statistic

Improving Imbalanced Machine Learning with Neighborhood-Informed Synthetic Sample Placement

2022 | Journal of Management Information Systems | Citations: 0

Authors: Nasir, Murtaza; Dag, Ali; Simsek, Serhat; Ivanov, Anton; Oztekin, Asil

Abstract: Machine learning is widely used in information systems design. Yet, training alg ... Expand

Abstract: Machine learning is widely used in information systems design. Yet, training algorithms on imbalanced datasets may severely affect performance on unseen data. For example, in some cases in healthcare, fintech, or cybersecurity contexts, certain subclasses are difficult to learn because they are underrepresented in training data. Our study offers a flexible and efficient solution based on a new synthetic average neighborhood sampling algorithm (SANSA), which, in contrast to other solutions, introduces a novel "placement" parameter that can be tuned to adapt to each dataset's unique manifestation of the imbalance. This package can be downloaded for R1. We tested SANSA against seven existing sampling methods used in conjunction with the four most frequently used machine learning models trained on 14 benchmark datasets. Our results provide suggestive evidence that SANSA offers a feasible solution to the imbalance problem for most datasets. Our findings provide practical recommendations for how SANSA can be effectively implemented while reducing the complexity level of an imbalanced learning pipeline. Collapse

Topics: marketing management customer service online review logistics management credit card fraud

Methods: computational algorithm machine learning support vector machine artificial neural network descriptive statistic

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

SumExp: A Summarization-Based Approach for Explaining NLP Models

2021 | International Conference on Information Systems | Citations: 0

Authors: Hristova, Diana; Khan, Mohd

Abstract: State-of-the-art NLP models such as BERT, RoBERTa or the Longformer show impres ... Expand

Abstract: State-of-the-art NLP models such as BERT, RoBERTa or the Longformer show impressive results, but are also black boxes with their millions of parameters. As a result, decision makers abstain from using those models as decision support tool, because they do not understand their output. In this paper, we address this topic by additionally closing an important research gab in the XAI literature. We develop SumExp, which is a summarization-based approach for generating explanations for NLP models. In particular, SumExp provides a short, comprehensible summary with the most important information that led the model to a certain prediction. As such, this explanation is not only comprehensible, but also fulfils fidelity (i.e., similar prediction) and stability. We evaluate our approach on three publicly available datasets by first deriving state-of-theart NLP models on them and then applying SumExp to the validation sets. The results confirm the fidelity, stability and comprehensibility of the methodology. Collapse

Topics: fake news explainable artificial intelligence Facebook cloud computing shapley additive explanation

Methods: natural language processing RoBERTa BERT experiment Yelp review

Assessing the Unacquainted: Inferred Reviewer Personality and Review Helpfulness

2021 | Management Information Systems Quarterly | Citations: 0

Authors: Liu, Angela Xia; Li, Yilin; Xu, Sean Xin

Abstract: This work examines the question of who is more likely to provide future helpful ... Expand

Abstract: This work examines the question of who is more likely to provide future helpful reviews in the context of online product reviews by synergistically using personality theories and data analytics. It trains a deep learning model to infer a reviewer's personality traits. This enables analyses to reveal the role of personality traits in review helpfulness among a large population of reviewers. We develop hypotheses on how personality traits are associated with review helpfulness, followed by hypotheses testing that confirms that higher review helpfulness is related to higher openness, conscientiousness, extraversion, and agreeableness and to lower emotional stability. These results suggest the appropriateness of using these five personality traits as inputs for developing a model for predicting future review helpfulness. Based on an ensemble model using supervised classification algorithms, we develop a predictive model and demonstrate its superior performance. Theoretical and practical implications are discussed. Collapse

Topics: personality online review knowledge sharing responsible artificial intelligence marketing management

Methods: support vector machine natural language processing computational algorithm deep learning machine learning

Theories: big five model personality theory