Keyphrase Substitution Recommender

Overview

This project aims to recommend substitutions for phrases in a given text based on the similarity to a list of standard phrases. The key components of the project include a custom KeyphraseCountVectorizer for extracting keyphrases, a TextAnalyzer class for computing similarity scores, and a recommendation function (recommend_option) for user interaction.

Dependencies

• keyphrase_vectorizers: A custom module for implementing the KeyphraseCountVectorizer.
• spacy: A natural language processing library for tokenization and vectorization.
• numpy: A library for numerical operations, used for calculating vector means.
• sklearn: A machine learning library, used for cosine similarity computation.

Install dependencies using:

pip install keyphrase_vectorizers spacy numpy scikit-learn

Download the SpaCy English model:

python -m spacy download en_core_web_sm

Usage

Keyphrase Extraction:

• The phrase_seeker function takes input text, utilizes KeyphraseCountVectorizer to extract keyphrases, and returns a list of keyphrases.

Text Analysis and Recommendations:

• The TextAnalyzer class is responsible for analyzing the input text and recommending substitutions.
• Instantiate TextAnalyzer with a list of standard phrases.
• Use the analyze_text method to compute similarity scores and generate substitution suggestions.

User Interaction:

• The recommend_option function serves as a user interface for receiving input text and displaying substitution recommendations.
• Run the function to enter your text, and the system will suggest replacements based on similarity scores.

Customization Adjust the standard_phrases list to include phrases relevant to your domain. Tune the similarity threshold in the analyze_text method for more precise recommendations. Feel free to enhance the keyphrase extraction or similarity computation methods based on your specific requirements.

Run the recommendation function

• recommend_option()