Back to Projects
Natural Language Processing

Customer Review Topic Modelling and Emotion Analysis

What drives negative reviews for a UK wellness centre chain? Topic modelling, emotion classification and an instruction-tuned LLM applied to 27,585 reviews from two public review platforms.

TOPIC WORDS
excellentrecommendpooroutstandingdisappointingaveragewonderfulwasterelaxingprofessionalrudeokayamazing
Type
Topic Modelling / NLP
Domain
Customer Experience
Data
27,585 reviews, 2 platforms
Origin
Cambridge DS Career Accelerator coursework

The Challenge

Topic modelling of 2,031 angry reviews traced the anger mainly to four issues: cleanliness, staff conduct, equipment and membership fees.

A national wellness centre chain receives thousands of reviews across public platforms. Star ratings show that customers are unhappy, but not why, or where. The task was to turn unstructured review text into specific, location-aware themes that operations teams can act on.

The work combined classical NLP with transformer models and an instruction-tuned LLM, and compared what each approach surfaced.

Approach

01
Preparation and anonymisation
Filtered to English-language reviews, removed duplicates, and stripped personal names and brand terms before analysis. Word frequencies and word clouds compared the two platforms.
02
Topic modelling of negative reviews
Defined negative reviews as ratings below 3, then ran BERTopic (sentence-transformer embeddings) on reviews from locations present on both platforms, and compared the top 20 locations by negative reviews on each platform.
03
Emotion analysis
Classified every negative review with a pre-trained BERT emotion model, kept the reviews where anger was the top emotion, and ran BERTopic again on that subset.
04
LLM topic extraction and LDA comparison
Used Falcon-7B-Instruct to extract three topics per review and suggest actions for each topic cluster, and compared the results with a Gensim LDA model (10 topics).

Results

27,585
English-language reviews analysed (11,857 + 15,728 across two platforms)
2,031
Negative reviews where anger was the dominant emotion
4
Main anger drivers identified: cleanliness, staff, equipment, fees

Negative reviews numbered 2,422 on one platform and 3,396 on the other. Among angry reviews, the largest themes were cleanliness of toilets and changing rooms (334 reviews), staff rudeness (326), equipment (271) and membership fees and cancellation (260). The first topic-modelling run on negative reviews pointed the same way: showers and changing-room cleanliness first, then broken equipment.

Seven locations appeared in both platforms' top-20 lists for negative reviews, giving a short list of sites where problems are consistent rather than platform-specific.

Limitations: the first BERTopic run used the first 1,000 negative reviews rather than a random sample, and the LLM step ran on 50 angry reviews only. The emotion model labelled about 1,700 negative reviews as "joy", a likely misclassification. LLM-generated suggestions are generic and were not validated.

How it was tested

TEST REPORTTEST REPORTCUSTOMER REVIEWSCUSTOMER REVIEWS
Checked
  • Reviews filtered to English, de-duplicated and stripped of personal names and brand terms before analysis.
  • Three approaches compared on the same reviews: BERTopic, Gensim LDA and an instruction-tuned LLM.
Found
  • The emotion model labelled about 1,700 negative reviews as "joy", a likely misclassification.
Limits
  • The first BERTopic run used the first 1,000 negative reviews rather than a random sample.
  • The LLM step ran on 50 angry reviews only, and its suggestions are generic and were not validated.

Technology Stack

Python BERTopic Sentence Transformers Hugging Face Transformers BERT emotion model Falcon-7B-Instruct Gensim LDA NLTK spaCy pandas
Interested in this work or something similar?