Netnography Study: End‑Users’ Satisfaction with Public Transport in UPPER Cities

Academic abstract

The UPPER Horizon Europe project aims to accelerate a public‑transport revolution across ten European cities, supporting the EU Cities Mission for climate‑neutral mobility by 2030. As part of its user‑research component, this paper presents a netnography study that examined citizen satisfaction with six transport modes (shared bike, bus, tram/subway, taxi, shared LEV, shared car) in five Living‑Lab cities: Valencia, Île‑de‑France, Rome, Oslo and Mannheim. Over 15 000 online comments and reviews were harvested from TripAdvisor, Google Reviews and Twitter (Jan – Feb 2023) and processed with natural‑language‑processing algorithms to extract sentiment polarity, emotions (anger, joy, sadness) and hate levels. Results are aggregated per mode and city, revealing distinct perception patterns and gender‑based differences in satisfaction.

Supporting evidence

The methodology combined web‑scraping, gender‑detection tools (ScrapeHero, Gender API) and NLP pipelines (Amazon Comprehend for sentiment, pysentimiento for emotion/hate, WordStream Maker for term frequency). Key steps included:

1. Data acquisition – systematic extraction of user‑generated content from the three platforms.
2. Metadata enrichment – tagging comments by gender, residency (tourist vs. local) and transport mode.
3. Sentiment & emotion analysis – classification into positive, negative, neutral, mixed and quantification of anger, joy, sadness, and aggression levels.
4. Semantic & verbatim analysis – identification of recurring topics (e.g., ticket issues, station design, vehicle condition) and extraction of illustrative excerpts.

The aggregated dataset enabled comparative dashboards across modes and cities, forming the evidential basis for the findings below.

Figure.1

Key findings

• Mode‑level sentiment – Tram/Subway and Taxi received the highest proportion of positive comments, whereas Bus and Shared Bike showed a lower positive‑to‑negative ratio.
• Rating vs. comment ratio – Although Tram/Subway achieved the best average rating, Taxi exhibited the highest positive/negative comment ratio (3 : 1 vs. 2.5 : 1 for Tram/Subway), indicating stronger alignment with user expectations.
• Issue clusters – Negative remarks frequently referenced ticket‑purchase problems, ageing rolling stock, overcrowding (tram/subway), poor vehicle maintenance and cleanliness (bus), and dock/e‑bike reliability (shared bike).
• Gender dynamics – Approximately 68 % of comments on shared‑mobility modes originated from male users; women tended to favour buses, taxis and subways and expressed fewer criticisms overall.
• Hate & aggression hotspots – The most common triggers for hostile language were ticket‑related frustrations, security concerns on subways, and unsafe driving behaviours on buses.

The comments classification shows that Subway/Tram, Taxi, Shared LEV and Shared car have obtained more positive comments than negative comments, while for Bus and Shared bike this ratio changes. 

On the other hand, the best rating (Subway/Tram) were not obtained by the transport mode with more positive comments, but that with the best ratio positive/negative comments (3 for Subway/Tram vs. 2.5 for Taxi). So according to this ratio, and considering that positive comments and negative comments are related to fulfilling users’ expectations, we get another transport mode classification where Subway/Tram and Taxi are transport modes that cover reasonably user’s expectations and Shared LEV, Shared car, Shared bike and Bus does not. 

Figure.2

In order to have a deeper understanding about what is wrong with those transport modes that are part of the second group, we can explore the terms that users are employing when they make positive and negative comments (verbatim analysis). The following figure presents the semantic analysis of the comments collected in the five living labs for the Shared Bike.

The bubbles graph included in next figure presents the terms related to positive comments and negative comments for the Shared bike. The terms customer, terminal, broken and electric used exclusively in negative comments, jointly with bicycle, service, station and application (employed in both, positive and negative comments), suggest that users consider bikes and docks are not properly maintained, e-bikes could be an interesting alternative, and the customer service should improve.  

Figure.3

To monitor levels of hate and aggression is critical to identify the triggering topics in user comments. Among the most frequently mentioned concerns in comments containing hate were Ticket issues (e.g. problems related to ticket purchase, including long queues or malfunctioning machines), subway challenges (e.g. concerns about security, aging of train carriages, or overcrowding), bus problems (e.g. complaints about old, poorly maintained, and unclean buses, or aggressive and unsafe driving), and Station-related issues (e.g. poorly designed and maintained stations, inaccessibility with long corridors, or difficulties for carrying luggage). 

In the context of gender-specific analysis, several notable trends have emerged. Firstly, a gender bias becomes apparent in the usage of shared transportation modes (including bicycles, light electric vehicles, and cars), with approximately 67.7% of comments originating from male users. Secondly, women tend to favour buses, taxis, and subways more frequently, while demonstrating lower utilization of shared transportation options. Lastly, men exhibit a greater propensity for criticism towards public transportation, as evidenced by a higher incidence of negative feedback when compared to women.