This project implements a sentiment analysis model for evaluating restaurant reviews. The model identifies aspects such as food, drinks, and service, and extracts related sentiment expressions from the reviews.
- Python 3.x
- Gensim
- SpaCy
- SciPy
- Clone the repository:
git clone https://github.com/yourusername/restaurant-sentiment-analysis.git
cd restaurant-sentiment-analysis
- Install required libraries:
pip install gensim spacy scipy
- Download the SpaCy French model:
python -m spacy download fr_core_news_md
- Download the word embeddings file frWac_non_lem_no_postag_no_phrase_200_skip_cut100.bin and place it in the project directory.
Run the analysis on a text file with reviews:
python sentiment_analysis.py reviews.txt
Results are saved in resultats.json.
- sentiment_analysis.py: Main script.
- frWac_non_lem_no_postag_no_phrase_200_skip_cut100.bin: Word embeddings (not included).
- reviews.txt: Input file.
- resultats.json: Output file.
- Loads word embeddings and SpaCy model.
- Defines aspect keywords for food, drinks, and service.
- Calculates similarity between words and aspect keywords.
- Identifies aspect terms and associated sentiments in reviews.
- Saves results in JSON format.