What do riders say and where? The detection and analysis of eyewitness transit tweets

Kabbani, O.; Klumpenhouwer, Willem; El-Diraby, Tamer E.; Shalaby, Amer

doi:10.1080/15472450.2022.2026773

Cited by 3 publications

(4 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Traditional assessment of sentiment or customer satisfaction via social media employs a generic sentiment lexicon in the transit domain. Several authors have acknowledged that these types of generic lexicons are context dependent and can have a severe impact on the accuracy of the sentiment analysis if not used for a suitable topic (3,5,15). To improve the accuracy of the transit sentiment analysis, we used a hybrid approach to develop a transit-specific sentiment lexicon, adopting different techniques from recently conducted research on domain-specific sentiment lexicons (24)(25)(26)(27).…”

Section: Customer-oriented Analysismentioning

confidence: 99%

“…On-Time Performance Calculation. Trips were separated according to the group they belonged to, resulting in a total of 6,040,111 stop-level details for control routes (3,196,090 before the service change versus 2,844,021 after) and 4,465,801 stop-level details for treatment routes (2,015,586 before the service change versus 2,450,215 after). On-time performance was calculated for each group and period based on the proportion of schedule deviations that were within Calgary Transit's time window of 1 min early and 5 min late.…”

Section: Operational-oriented Analysismentioning

confidence: 99%

“…Hosseini et al ( 14 ) drew on industry reports and surveys on customer satisfaction to develop a large set of transit-specific terms that were linked with themes such as spatial availability, safety and security, and service delivery. This lexicon was further extended by Kabbani et al ( 3 ) in their spatial, temporal, semantic, and sentiment analysis of a real-time stream of tweets in Calgary, Canada.…”

Section: Literature Reviewmentioning

confidence: 99%

“…A new complementary source for assessing customer opinion that has been gaining popularity in the research field is Twitter, the widely used social media platform ( 2 ). Users post tweets frequently, often in real time ( 3 ), and are willing to express their feelings about their interactions and experience with transit systems. Tweets convey very concise messages because of the imposed character limit, and they can be easily obtained.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Using Twitter to Gauge Customer Satisfaction Response to a Major Transit Service Change in Calgary, Canada

Al-Sahar,

Klumpenhouwer,

Shalaby

et al. 2023

Transportation Research Record: Journal of the Transportation R

Self Cite

View full text Add to dashboard Cite

Measuring public opinion about the quality of transit services is a key factor in understanding and addressing customer dissatisfaction and increasing customer loyalty and ridership. Sentiment analysis using social media—in particular Twitter—is a relatively cheap and potentially powerful complement to traditional survey methods, which are expensive and limited in sample size. This study aims to evaluate customer response to the introduction of Calgary Transit’s MAX routes. We compared customer satisfaction expressed on Twitter with measured service reliability in the form of on-time performance. We also employed a qualitative research approach using content analysis from Twitter to gauge rider satisfaction over several service attributes before and after the service change. A transit-specific sentiment lexicon was developed to support this study using a hybrid approach. This lexicon outperformed generic sentiment lexicons traditionally used in transit studies with regard to both accuracy (18.4%) and F1-score (7.1%). We found that the overall perception of on-time performance from riders using Twitter was similar to the actual performance in the field. This was also observed for one individual route on which stops with poor schedule adherence were linked with negative feedback. This study concludes that combining customer-oriented measures from Twitter with operational-oriented ones would enable transit agencies to make better-informed decisions for planning and operational purposes.

show abstract

Section: Customer-oriented Analysismentioning

confidence: 99%

Section: Operational-oriented Analysismentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Using Twitter to Gauge Customer Satisfaction Response to a Major Transit Service Change in Calgary, Canada

Al-Sahar,

Klumpenhouwer,

Shalaby

et al. 2023

Transportation Research Record: Journal of the Transportation R

Self Cite

View full text Add to dashboard Cite

show abstract

On the Relationship Between Crowdsourced Sentiments and Mobility Trends During COVID-19: A Case Study of Kyoto

Sun,

Kobayashi,

Nakao

et al. 2023

Data Sci. Transp.

View full text Add to dashboard Cite

MetRoBERTa: Leveraging Traditional Customer Relationship Management Data to Develop a Transit-Topic-Aware Language Model

Leong,

Abdelhalim,

et al. 2024

Transportation Research Record: Journal of the Transportation R

View full text Add to dashboard Cite

Transit riders’ feedback provided in ridership surveys, customer relationship management (CRM) channels, and, in more recent times, through social media, is key for transit agencies to better gauge the efficacy of their services and initiatives. Getting a holistic understanding of riders’ experience through the feedback shared in those instruments is often challenging, mostly because of the open-ended, unstructured nature of text feedback. In this paper, we propose leveraging traditional transit CRM feedback to develop and deploy a transit-topic-aware large language model (LLM) capable of classifying open-ended text feedback to relevant transit-specific topics. First, we utilize semi-supervised learning to engineer a training dataset of 11 broad transit topics detected in a corpus of 6 years of customer feedback provided to the Washington Metropolitan Area Transit Authority. We then use this dataset to train and thoroughly evaluate a language model based on the RoBERTa architecture. We compare our LLM—MetRoBERTa—with classical machine learning approaches utilizing keyword-based and lexicon representations. Our model outperforms those methods across all evaluation metrics, providing an average topic classification accuracy of 90%. Finally, we provide a value proposition of this work demonstrating how the language model, alongside additional text processing tools, can be applied to add structure to open-ended text sources of feedback such as Twitter. The framework and results we present provide a pathway for an automated, generalizable approach for ingesting, visualizing, and reporting transit riders’ feedback at scale, enabling agencies to better understand and improve customer experience.

show abstract

What do riders say and where? The detection and analysis of eyewitness transit tweets

Cited by 3 publications

References 19 publications

Using Twitter to Gauge Customer Satisfaction Response to a Major Transit Service Change in Calgary, Canada

Using Twitter to Gauge Customer Satisfaction Response to a Major Transit Service Change in Calgary, Canada

On the Relationship Between Crowdsourced Sentiments and Mobility Trends During COVID-19: A Case Study of Kyoto

MetRoBERTa: Leveraging Traditional Customer Relationship Management Data to Develop a Transit-Topic-Aware Language Model

Contact Info

Product

Resources

About