Topic classification of electric vehicle consumer experiences with transformer-based deep learning

Abstract

Summary The transportation sector is a major contributor to greenhouse gas (GHG) emissions and is a driver of adverse health effects globally. Increasingly, government policies have promoted the adoption of electric vehicles (EVs) as a solution to mitigate GHG emissions. However, government analysts have failed to fully utilize consumer data in decisions related to charging infrastructure. This is because a large share of EV data is unstructured text, which presents challenges for data discovery. In this article, we deploy advances in transformer-based deep learning to discover topics of attention in a nationally representative sample of user reviews. We report classification accuracies greater than 91% (F1 scores of 0.83), outperforming previously leading algorithms in this domain. We describe applications of these deep learning models for public policy analysis and large-scale implementation. This capability can boost intelligence for the EV charging market, which is expected to grow to US$27.6 billion by 2027.

Widespread use of National Academies consensus reports by the American public

“…Specifically, comments were classified using Bidirectional Encoder Representations from Transformers (BERT), a transformer-based natural language machine learning classification algorithm with outstanding performance on subtle classification tasks because it encodes both semantics and the rich latent structure of sentences ( 5 , 6 ). The superiority of BERT over other machine learning natural language classification models has been repeatedly established in varied real-world social science datasets ( 7 – 12 ).…”

A field experiment on workplace norms and electric vehicle charging etiquette

“…However, given the high capital costs to install EV charging points, it is likely that many small or medium size employers may not be able to deploy charging infrastructure, leaving employees to incur the cost of installing costly home charging and potentially propagating misperceptions about energy costs (Asensio, 2019). Prior work using machine learning and artificial intelligence has shown the prevalence of negative consumer experiences in public EV charging infrastructure, with issues such as lack of station availability and functionality, particularly in smaller urban communities (Asensio et al., 2020; Ha et al., 2021). In such cases where there are investments gaps by small and medium businesses, targeted state and local policies can serve an important function to promote equitable charging access, particularly for communities who may be under‐served in access to public charging points.…”

Making sense of electrical vehicle discussions using sentiment analysis on closely related news and user comments

“…While the field of NLP-based EV research is relatively young, there has been some attempt to use topic-analysis/topic-modelling and sentiment analysis/classification of user reviews. Ha et al (Ha et al, 2021) have performed a user-experience-focused topic analysis related to EVs using language transformer models and supervised topic classification. However, whereas Ha et al (Ha et al, 2021) focused mainly on consumer reviews of EV charging stations, we will instead focus more generally on a comparison between consumer and media sentiment towards EVs.…”

“…Ha et al (Ha et al, 2021) have performed a user-experience-focused topic analysis related to EVs using language transformer models and supervised topic classification. However, whereas Ha et al (Ha et al, 2021) focused mainly on consumer reviews of EV charging stations, we will instead focus more generally on a comparison between consumer and media sentiment towards EVs. Ha et al found that frequent topics included charger functionality, range-anxiety, charging cost, and dealership experiences (Ha et al, 2021).…”

“…However, whereas Ha et al (Ha et al, 2021) focused mainly on consumer reviews of EV charging stations, we will instead focus more generally on a comparison between consumer and media sentiment towards EVs. Ha et al found that frequent topics included charger functionality, range-anxiety, charging cost, and dealership experiences (Ha et al, 2021). Given this, we believe that topic-modeling can be an effective way of understanding consumer data.…”