SNAPAmazon

Offering a broad and deep view into consumer sentiments and opinions.

Dataset Overview

This dataset compiles 34,686,770 Amazon reviews from 6,643,669 users across 2,441,053 products, offering deep insights into consumer sentiments.

Data Origin

Derived from the Stanford Network Analysis Project (SNAP), this dataset spans an 18-year period up to March 2013, providing a broad historical perspective on consumer opinions.

Sentiment Classification

Reviews are categorized into negative and positive sentiments based on their ratings, with scores of 1-2 considered negative and 4-5 positive, excluding neutral reviews.

Dataset Structure

The dataset is meticulously structured, featuring 1,800,000 training samples and 200,000 testing samples for each sentiment category, designed for polarity analysis.

File Format

Provided in CSV format, each file includes three columns: polarity, title, and text, representing the sentiment classification, review heading, and body of the review, respectively.

Usage Guidelines

Researchers and developers can utilize this dataset to analyze trends, perform sentiment analysis, and develop machine learning models to better understand consumer behavior.

TOKEN SHOWCASE

List of tokens people are building with Solana

🙏 Please add your token

BTC