An overall star rating or a single positive or negative score does not show which parts of a product customers like or dislike, because one review often praises one feature and criticizes another. A company that wanted to understand the strengths and weaknesses of products sold online needed sentiment tied to specific product characteristics, across hundreds of thousands of reviews.
The client chose to work with Coherent Solutions to engineer a natural language processing (NLP) solution based on aspect-based sentiment analysis, which measures sentiment toward individual product characteristics instead of rating a review as a whole.
Coherent Solutions provided AI and machine learning engineering, data processing, and cloud engineering. The solution identifies the sentiment attached to each product characteristic and highlights product strengths and weaknesses across customer reviews.
Client: A company seeking product- and attribute-level insights from online customer reviews
Industry: Technology, e-commerce
Challenge: Overall sentiment scores do not show which product characteristics shape customer opinion, and hundreds of thousands of reviews had to be processed at scale
Solution: An NLP solution using aspect-based sentiment analysis
Outcome: Sentiment for specific product characteristics, product strengths and weaknesses identified across reviews, and customer feedback processed at scale
Coherent's role: Engineering partner for the NLP solution, cloud engineering, and scalable data processing
The client needed to gain a deeper understanding of its more than 200K customer reviews for products sold online. It wanted to know the strengths and weaknesses of individual products and use that information to make its marketing more relevant.