How can topic-modelling of user-reviews reshape market surveys? Exploring factors influencing usage intention of e-learning services through a novel multi-method approach

Arghya Ray, Pradip Kumar Bala, Rashmi Jain

Research output: Contribution to journalReview articlepeer-review

Abstract

Online user-generated-content is not only a critical performance parameter for service-providers but also serve as a vital information source for prospective customers. For understanding the factors influencing the adoption or continuance of an e-service, traditional approaches required a qualitative or quantitative-based analysis. In this study, we propose a novel approach to generate and analyse path model in real-time by combining the qualitative, quantitative and natural language processing (NLP)-based approaches. We undertook an emic approach using semi-structured interview schedule (ten participants) and an etic approach using topic modelling on extant literature (3,570 articles) for exploring factors influencing motives behind use of e-learning services. We tested the path-model using traditional quantitative-based (542 respondents) and the proposed NLP-based approaches (3,227 online reviews) through structural equation modelling (SEM). Results of this study revealed content gratification as the most important predictor of usage intention. This study concludes with the implications, limitations and future research directions.

Original languageEnglish
Pages (from-to)259-284
Number of pages26
JournalInternational Journal of Business Information Systems
Volume40
Issue number2
DOIs
StatePublished - 2022

Keywords

  • consumer perspectives
  • e-learning
  • latent Dirichlet allocation
  • LDA
  • multi-method approach
  • NLP-based approach
  • path analysis
  • qualitative analysis
  • quantitative analysis
  • sentiment analysis
  • text mining

Fingerprint

Dive into the research topics of 'How can topic-modelling of user-reviews reshape market surveys? Exploring factors influencing usage intention of e-learning services through a novel multi-method approach'. Together they form a unique fingerprint.

Cite this