English Booking.com Reviews

Dataset Description

This dataset contains 1,486,195 English-language hotel reviews scraped from Booking.com, covering the period between April 1, 2020 and September 9, 2022. Each entry corresponds to a single user-generated review and includes metadata, textual content, and a wide range of linguistic features.

Contents

Review metadata: hotel name, hotel country, reviewer username (anonymized), review and stay dates, review ID, traveler type, room information, number of nights, rating scores, review helpfulness votes, and hotel responses (where available).

Review text: original titles, positive and negative free-text fields, and full review content. Reviews are available in English (either originally written in English or translated).

LIWC-22 variables: the full set of text-analytic features from Linguistic Inquiry and Word Count 2022 (LIWC-22; Boyd, Ashokkumar, Seraj, & Pennebaker, 2022), capturing linguistic style, psychological processes, affect, cognition, social references, and topical domains.

Derived linguistic features (computed with a spaCy-based Python pipeline):

WordCount: number of alphabetic tokens per review

WordLength: mean word length in characters

NominalizationShare: proportion of nouns ending with common nominalization suffixes (e.g., -tion, -ment, -ness)

LexicalDiversity_MATTR: moving-average type–token ratio (MATTR; window size = 50)

Part-of-speech shares: proportion of adjectives, nouns, verbs, and adverbs in each review

Concreteness: semantic projection (Wartena, 2022) anchored to Brysbaert et al. (2014)

Data Structure

The dataset is provided as a CSV file with 148 variables (columns). A full variable list is included in the file header.

Usage Notes

All reviews were collected from publicly available Booking.com pages.

No personally identifying information is included; usernames are pseudonymous.

Portions of the text may have been machine-translated by Booking.com (variable: translation).

LIWC-22 variables are provided for research purposes; LIWC is licensed software that requires purchase.

All analysis code—Python scripts for computing lexical variables (word count, word length, nominalization share, MATTR, and POS shares) and R scripts implementing the semantic projection procedure to obtain average concreteness ratings across reviews—is openly accessible via OSF: https://osf.io/pgbj8/files/osfstorage/68be8f3ac3872097f774ea5b.

Funding

Data scraping and processing were supported by funding from the Forschungskommission (FoKo-805387), University of Lucerne.

Citation

If you use this dataset, please cite:

Boyd, R. L., Ashokkumar, A., Seraj, S., & Pennebaker, J. W. (2022). The development and psychometric properties of LIWC-22. Austin, TX: University of Texas at Austin. http://dx.doi.org/10.13140/RG.2.2.23890.43205

Brysbaert, M., Warriner, A. B., & Kuperman, V. (2014). Concreteness ratings for 40 thousand generally known English word lemmas. Behavior Research Methods, 46(3), 904–911. https://doi.org/10.3758/s13428-013-0403-5

Wartena, C. (2022). On the geometry of concreteness. In Proceedings of the 7th Workshop on Representation Learning for NLP (pp. 204–212). https://doi.org/10.18653/v1/2022.repl4nlp-1.21