Natural Language Processing and Sentiment Analysis of Customer Reviews using RapidMiner

Project Description

Amway operates in more than 100 countries and is ranked 29th among the largest private companies in the United States.  We have been collecting customer reviews and survey data that has been transcribed from phone conversations in different languages.  We would like a data scientists well-versed in natural langage processing (NLP) to engage in sentiment analysis of this data.

Data Format

The customer review data is in unstructured format and contains approximately 2,000 records.  The initial data set is in the Chinese language and surveys from additional languages will be added after the success of the first phase of this project.  A data sample is attached.

Technology Stack

The data will be analyzed by a RapidMiner-certified data scientist and the analysis results will be exported to Tableau.

Data Analysis

We would like to look at the analysis by the slices listed below. For example can we see the sentiment between months, or in different regions, or by different SKUs, etc.

Be able to slice analysis by:

• Month
• SKU No.
• Complaint Code
• Region
• Province (?)

Text analysis:

• Frequency
• Di/Trigrams
• Sentiment
• Word Associations/Correlations/Clusters

The above represents some of our ideas but we encourage data scientists to suggest other approaches.  Please look at the sample data and provide your approach to the analysis and the kind of insights that can be drawn from it.  Please also provide a ball-park estimate of hours this work may take.