Sentiment Analysis using Google Generative AI

This repository contains code to perform sentiment analysis on text data using the Google Generative AI API. The code processes text data, adds sentiment analysis results, and generates a sentiment distribution graph.

Installation

1. Install the required Python packages using pip:

   $ pip install google-generativeai
   $ pip install json
   $ pip install pandas
   $ pip install matplotlib
   $ pip install time
   $ pip install dotenv
   
   

2. Create a .env file in the root directory and add your Google Generative AI API key as follows:

   API_KEY=your_api_key_here
   

   Load API_KEY from .env file

   load_dotenv()
   

How it works

1. Google Generative AI Palm API Key

   palm.configure(api_key=os.environ['API_KEY'])
   

2. Google Generative AI Default Settings

   defaults = {
       # ... (default settings)
   }
   

3. Process each record

   def process_and_append_added_sentiment_and_score_results_to_csv(single_record):
       # ... (processing logic)
   

4. Generate sentiment analysis results and save them to a CSV file.

   4.1 Read the data into a DataFrame

   df_no_sentiment_no_score = pd.read_csv("food_review_data_no_sentiment_no_score.csv")
   df_no_sentiment_no_score_filtered = df_no_sentiment_no_score[['Id', 'Summary']]
   

   4.2 Set the chunk size (Do Not Change)

   chunk_size = 1 
   

   4.3 Iterate over chunks

   for i in range(0, len(df_no_sentiment_no_score_filtered), chunk_size):
       chunk = df_no_sentiment_no_score_filtered.iloc[i:i+chunk_size]
       # Save Added Sentiment and Score Results to a CSV file
       process_and_append_added_sentiment_and_score_results_to_csv(chunk)
       # Wait for 5 seconds To Avoid Hitting API Quota
       time.sleep(5)
   

Generates Sentiment Distribution Graph

``` 
def create_sentiment_distribution_graph():
    # ... (graph generation logic)

create_sentiment_distribution_graph()
``` 

Notes

This code assumes you have access to the Google Generative AI API and have obtained an API key.

Be mindful of API quotas and usage limits when running the code.

Make sure to customize the file paths and column names to match your data. Also after each run make sure to clear the contents or delete the csv output file.

The script create_sentiment_distribution_graph generates a bar chart of sentiment distribution, but ensure that matplotlib is properly configured to display the chart.

The code provided processes data in chunks to manage API requests. It currently only supports one record at a time feel but free to Adjust the time.sleep duration as needed.

The dataset used is a partial set of records for food reviews that were obatined from the Stanford Network Analysis Project hosted here: https://www.kaggle.com/datasets/snap/amazon-fine-food-reviews