Elamur Research⌕
☰
home / papers / 2026 / Quantifying Social Sentiment in Hostels Using A Domain-Specific Transformer Pipeline

Quantifying Social Sentiment in Hostels Using A Domain-Specific Transformer Pipeline

Paper · 2026 · Ian W. McMurry

Summary

Social atmosphere matters a great deal to hostel travelers, but it is hard to measure. Booking platforms such as Hostelworld offer an "atmosphere" score. Under expectation–disconfirmation theory, that score mainly reflects whether expectations were met, so calm and party hostels can earn similar ratings. Generic sentiment and topic models also miss implicit social cues in reviews.

The author defines "socialness" as a distinct semantic construct and builds a three-stage pipeline:

  • a cross-encoder trained on a manually annotated seed set,
  • large-scale pseudo-labeling of a review corpus from Hostelworld and Booking.com,
  • distillation of those labels into a sentence-transformer bi-encoder, whose embedding space orders reviews by level of guest interaction rather than general positivity.

A linear classifier on these embeddings beats generic sentence embeddings and zero-shot GPT-4o on held-out human labels. Embedding-space separation of social versus non-social reviews also improves sharply. Aggregated to the property level, scores show that very social hostels are uncommon. The work suggests a reusable recipe for mining large text collections for experiential qualities that reviewers convey only indirectly.

Key points

  • Data: 4,994 hand-labeled reviews train the cross-encoder, which then pseudo-labels 162,840 more. Roughly a third of the corpus is machine-translated, and the seed set was labeled by one annotator.
  • Classification: F1 reaches 0.826 with domain-adapted embeddings, compared with 0.671 for generic embeddings and 0.774 for zero-shot GPT-4o.
  • Embedding separation: the gap between intra- and inter-class cosine similarity rises from 0.018 to 0.704, about a 40-fold increase.
  • Clustering: unsupervised k-means (k = 2) reaches purity 0.909 versus 0.748 for the generic baseline.
  • Property level: hostel-level socialness approximately follows an exponential distribution, with highly social properties being rare.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksPsychology · 1,552 worksSocial Psychology · 306 works

🧭 Topics 1Research subjects and languages the work is about.Sentiment & emotion · 972 works

Cite

Cite as Ian W. McMurry. 2026. Quantifying Social Sentiment in Hostels Using A Domain-Specific Transformer Pipeline. The Proceedings for the 15th Workshop on Computational Approaches to Subjectivity, Sentiment Social Media Analysis (WASSA 2026).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.wassa-1.3

@inproceedings{mcmurry2026quantifying,
  title = {{Quantifying Social Sentiment in Hostels Using A Domain-Specific Transformer Pipeline}},
  author = {Ian W. McMurry},
  booktitle = {The Proceedings for the 15th Workshop on Computational Approaches to Subjectivity, Sentiment Social Media Analysis (WASSA 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.wassa-1.3},
  url = {https://doi.org/10.18653/v1/2026.wassa-1.3}
}

References

13 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Zheng Xiang, Zvi Schwartz, John H. Gerdes, and Muzaffer Uysal. 2015. What can big data and text analytics tell us about hotel guest experience and satisfaction? International Journal of Hospitality Management, 44:120–130. find
  2. Olivier Chapelle, Bernhard Scholkopf, and Alexander Zien. 2009. Semi-supervised learning (chapelle, o. et al., eds.; 2006)[book reviews]. IEEE Transactions on Neural Networks, 20(3):542–542. find
  3. Hostelworld. 2024. State of solo travel report 2024. Accessed: 2025-12-13. find
  4. Hostelworld. 2025. State of solo travel report 2025. Accessed: 2025-12-13. find
  5. Allen H Huang, Hui Wang, and Yi Yang. 2023. Finbert: A large language model for extracting information from financial text. Contemporary Accounting Research, 40(2):806–841. find
  6. Evgeny Kim and Roman Klinger. 2018. Who feels what and why? annotation of a literature corpus with semantic roles of emotions. In Proceedings of the 27th International Conference on Computational Linguistics, pages 1345–1359, Santa Fe, New Mexico, USA. Association for Computational Linguistics. find
  7. Jay Krishnan, Biplab Bhattacharjee, Maheshwar Pratap, Janardan Krishna Yadav, and Moinak Maiti. 2024. Survival strategies for family-run homestays: Analyzing user reviews through text mining. Data Science and Management, 7(3):228–237. find
  8. Dong-Hyun Lee and 1 others. 2013. Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks. In Workshop on challenges in representation learning, ICML, volume 3, page 896. Atlanta. find
  9. Jinhyuk Lee, Wonjin Yoon, Sungdong Kim, Donghyeon Kim, Sunkyu Kim, Chan Ho So, and Jaewoo Kang. 2020. Biobert: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics, 36(4):1234–1240.
  10. Gang Li, Rob Law, Huy Quan Vu, Jia Rong, and Xinyuan Zhao. 2015. Identifying emerging hotel preferences using emerging pattern mining technique. Tourism Management, 46:311–321. find
  11. Ana Oliveira-Brochado and Carla Gameiro. 2013. Toward a better understanding of backpackers’ motivations. Tékhne, 11(2):92–99. find
  12. Richard L. Oliver. 1980. A cognitive model of the antecedents and consequences of satisfaction decisions. Journal of Marketing Research, 17(4):460–469. find
  13. Abraham Pizam and Ady Milman. 1993. Predicting satisfaction among first-time visitors to a destination by using the expectancy disconfirmation theory. International Journal of Hospitality Management, 12(2):197–209. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/mcmurry-2026-quantifying-social-sentiment-hostels-domain/