ProductsAI & Analytics

Unlocking insights trapped in unstructured wine reviews

Technology challenge from Forrester Data Strategy & Insights 2019

At this year’s Forrester Data Strategy & Insights conference in Austin, TX, Forrester’s Boris Evelson presented a live technology challenge to OpenText and another vendor in his session, “Best in Class Platforms.” The challenge was to ingest a data set of more than 150,000 wine reviews live on stage and demonstrate which insights OpenText™ Magellan™ could gather from the data by extracting topics, categories, entities and sentiment. The data included structured fields like price, points and variety and unstructured data like descriptors of flavor, acidity and general sentiment towards the wine.

The challenge was broken up into three steps: ingesting data and performing natural language processing to detect the topics, categories and sentiment from the unstructured reviews; exploring data, building metrics and creating a dashboard; and performing some advanced analytics and a machine learning model to predict points based on reviews.

Take a look at how we tackled each of those steps in the demo video below.

Virgil Dodson

Virgil is the Director of Product Marketing and Developer Community for OpenText AI and Analytics

Related Posts

Back to top button