Using recent advancements in NLP, Modulai developed a model for clustering customer feedback into topics, making it possible to monitor sentiment across these topics and detect potential negative responses in real time so that companies can take immediate action.
Challenge
The client is a Stockholm-based startup founded in 2015 that helps businesses grow by the voice of their happy customers, providing a cloud-based solution for customer referrals, responses, recommendations, rewards, reviews, retargeting, and retention. Staying alert to customer opinion is key, since reviews and testimonials are an important part of many companies' marketing strategies.
Solution
The team was given a large dataset of customer feedback in raw text form, with subjects varying greatly and very little contextual information. Modern NLP models that can identify semantic similarity even when the wording differs were used to cluster the feedback into distinct topics. The solution then lets the client monitor these topics, and potentially new ones as they emerge, over time, alerting them to a clear shift in sentiment.
Tools
The team developed a model using a state-of-the-art Transformer architecture to detect semantic similarities in customer feedback, making it possible to identify clusters or topics. The approach leverages recent advancements in NLP for handling noisy, unstructured text data.
Value created
The model groups unstructured customer feedback into topics on its own, then tracks sentiment within each one over time. Instead of reading through raw feedback to spot problems, the client can be alerted the moment sentiment turns negative on a given topic, catching a brewing issue while there is still time to act, which matters for a business built around customer reviews and testimonials.
Learn more
