Uncategorized

STAGES OF DATA SCIENCE IN SENTIMENT ANALYSIS ON BUZZILLIONS.COM

When the data is structured in the form of databases, it is easy for anyone to analyze and predict the results. The going gets tough when we try to analyze the text data and we have to do it as one cannot read all the text in the whole world. Almost all the business domains have this problem but this issue effects the retail industry more.

 

Buzzillions is a product review site that is well trafficked and consumers use them to make buying decisions based on the product reviews posted already for that particular product. This article focuses on how the sentiment analysis aims to determine the attitude of consumers over a particular product.  According to Wikipedia, Opinion mining (sometimes known as sentiment analysis or emotion AI) refers to the use of natural language processing, text analysis, computational linguistics, and biometrics to systematically identify, extract, quantify, and study affective states and subjective information.

 

This is a serious problem faced by the ecommerce companies and can be solved with the application of data science techniques.  There are different stages involved in solving this particular problem and they are listed below:

  • Problem identification
  • Collecting the data
  • Processing the data
  • Model building
  • Arriving at the best possible solution to the problem

Let’s discuss these stages in detail here.

 

PROBLEM IDENTIFICATION:

When the customer buys a product online he gives a review about the product online which helps the other consumers to decide to buy the product which is reviewed or sometimes not to buy the product. So it is crucial for the eCommerce companies to analyze about the product review.

 

Here as stated earlier the data is in the form of text and hence unstructured. It is impossible to deal with this kind of data using the usual methods to analyze the trends behind it. The Natural Language Processing techniques help to process the text and identify what is the overall opinion of the consumers over any product.

 

COLLECTING THE DATA:

This unstructured data is collected by scrapping the ecommerce website by the suitable tools and should be converted to a structured format.

 

PROCESSING THE DATA USING NLP TECHNIQUES:

The data collected should be processed and it involves various tasks like cleaning the data into structured format, treating missing values, identifying outliers, identifying the parts of speech, removing the stop words, identifying the root word, tokenizing, etc. This can be achieved with the help of wordnet. The following steps are being followed in the processing the data using NLP techniques:

  1. Lexical Analysis
  2. Syntactic Analysis (Parsing)
  3. Semantic Analysis
  4. Discourse Integration
  5. Pragmatic Analysis

 

MODEL BUILDING:

After the data is processed, the patterns can be identified in the data through data visualization techniques. Later the suitable models can be built with various algorithms to identify the opinions by connecting to the sentiwordnet library. The models developed can be tuned for optimizing the results.

 

ARRIVING AT THE BEST POSSIBLE SOLUTION TO THE PROBLEM:

After the most optimized algorithm is being developed and checked for the feasibility. Later it can be implemented for the day to day operations of mining the opinions and identifying the sentiments behind the reviews with the live streaming data. This will help the ecommerce companies to identify which products and the suppliers are doing better and what are the major causes for the negative review for a particular product (Descriptive analytics), what will the review for the particular product in the future (Prescriptive analytics) and if the review going to be negative in the future (which is predicted), how can this be solved (Prescriptive analytics) 

Facebooktwittergoogle_pluspinterestlinkedinmail

Leave a Reply

Your email address will not be published. Required fields are marked *