Amazon Food Reviews - NLP

Data Source : Data

Machine-Learning Objective

NLP is a field in machine learning with the ability of a computer to understand, analyze, manipulate, and potentially generate human language.

This project mainly focuses on Sentiment Analysis of Amazon Food Reviews. Various methods ranging from wordvectors to complex deep learning methods are used in this project.

The Amazon Fine Food Reviews dataset consists of reviews of fine foods from Amazon.

• Number of reviews: 568,454
• Number of users: 256,059
• Number of products: 74,258
• Timespan: Oct 1999 - Oct 2012
• Number of Attributes/Columns in data: 10

Attribute Information:

• Id
• ProductId - unique identifier for the product
• UserId - unqiue identifier for the user
• ProfileName
• HelpfulnessNumerator - number of users who found the review helpful
• HelpfulnessDenominator - number of users who indicated whether they found the review helpful or not
• Score - rating between 1 and 5
• Time - timestamp for the review
• Summary - brief summary of the review
• Text - text of the review

Approach

Preprocessing

• Data Cleaning - Deduplications
• Removing html tags , punctuations and set of special characters like , or . or # etc .
• Checking if the word is made up of english letters and not alpha numeric characters .
• Converting to lowercase .
• Removal of stopwords .
• Stemming of words .

Featurization

It is necessary to convert the words in vector format for NLP tasks . So used following methods to convert words in vector format .

• Bag of Words (BoW)
• Bi-Grams and N-Grams
• TF-IDF (Term Frequency–Inverse Document Frequency)
• Average Word2Vec
• TF-IDF weighted Word2vec

• Word2Vec is the most popular technique to learn word embeddings . Embeddings are the vector representation of words and it is capable of capturing context of a word in a document , semantic and syntatic similarity , relation with other words , etc .

Implementations

• Dimensionality Reduction using t-SNE , PCA
• Naive Bayes
• Logistic Regression
• SGD for linear Regression
• Support Vector Machines
• Decision Trees
• Random Forests and GBDT
• KMeans , Agglomerative and Hierarchical Clustering methods (Unsupervised)
• TruncatedSVD (Co-occurrence matrix) - Matrix Factorization method (Used in recommendation systems)

Examples

• Truncated SVD (Based on occurrence similar words) :

• Dimensionality reduction using UMAP on Average W2V to visualize seperation of positive and negative reviews :

• Top 20 important words that are responsible for classification of reviews :