Amazon Product Analysis: Customers' Review Scraping and Sentiment Analysis

Built a tool to crawl Amazon reviews and classify review with CNN. This project use 2 methods of vectorizing text:

• Tokenizer, Embedding from Tensorflow
• gensim.Word2Vec.

Folder details:

• crawler:
  • exe tool
  • requirements.txt
  • crawler python script
• data:
  • multiple data files in csv format
  • merge.ipynb: merge all above data files
• model_word2vec:
  • data
  • word2vec model
  • CNN model
• model_tensorflowtokenizer:
  • data
  • CNN model

Technologies:

• Crawl: python BeautifulSoup, Selenium
• NLP: preprocessing, vectorize with Tensorflow.Tokenizer and gensim.Word2Vec
• Neural Network: tensorflow