Opening Up the Conversation: Topic Modeling for Automated Text Analysis in Travel Surveys

Vishnu Baburajan, Joao De Abreu e Silva, Francisco Camara Pereira

    Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

    Abstract

    The difficulties associated with the collection and analysis of open-ended questions in surveys (e.g. “Tell us your opinion about … ”), have encouraged the widespread use of closed-ended responses (e.g. “Your opinion according to a scale of 1–5 … ”). These, in some circumstances, are very restrictive, curbing the recognition of nuances in the survey population, hence limiting our models. Latent Dirichlet Allocation (LDA), now opens avenues for extraction of open-ended responses. In this paper, we analyze the potential of LDA in open-ended responses, by comparing with closed-ended counterpart options. A questionnaire designed based on Theory of Planned Behavior (TPB), is used to collect information on the intentions to use shared autonomous mobility. Two versions of the questionnaire were used, that alternately allow for open- and closed-ended versions of the same questions. Factor analysis was used to construct factors from the Likert scale questions and LDA was used to extract information from open-ended questions. Ordered Probit models were estimated to predict the intention to use shared autonomous mobility services. Attitudes, socio-demographics and current travel behavior of individuals were used in the model. The questionnaire instrument played an important decision in an individual's decision to use the proposed service. Comparable results were obtained for the two versions of the questionnaire, which emphasizes the need for further research on use of open-ended questions in measuring attitudes.
    Original languageEnglish
    Title of host publication2018 21st International Conference on Intelligent Transportation Systems (ITSC)
    PublisherIEEE
    Publication date2018
    Pages3657-3661
    DOIs
    Publication statusPublished - 2018
    EventITSC 2018: 21st International IEEE Conference on Intelligent Transportation Systems - Maui, Maui, United States
    Duration: 4 Nov 20187 Nov 2018
    Conference number: 21
    https://www.ieee-itss.org/itsc

    Conference

    ConferenceITSC 2018: 21st International IEEE Conference on Intelligent Transportation Systems
    Number21
    LocationMaui
    CountryUnited States
    CityMaui
    Period04/11/201807/11/2018
    Internet address

    Keywords

    • Autonomous Mobility
    • Theory of Planned Behavior
    • Likert Scales
    • Open-ended Responses
    • Topic Modelling
    • Latent Dirichlet Allocation

    Cite this

    Baburajan, V., De Abreu e Silva, J., & Pereira, F. C. (2018). Opening Up the Conversation: Topic Modeling for Automated Text Analysis in Travel Surveys. In 2018 21st International Conference on Intelligent Transportation Systems (ITSC) (pp. 3657-3661). IEEE. https://doi.org/10.1109/ITSC.2018.8569380