This repo is for tutorials related to deciphering and exploring consumer drug reviews or other survey texts
Companies rely on surveys to collect invaluable feedback about how their services and products are working for people. Open-ended responses offer crucial insight, allowing customers to describe their experience in their own words. Yet what they say can be difficult to analyze systematically. Using natural language processing, you’ll learn how to categorize opinions, identify themes and emotions (e.g., is the respondent upset or delighted?), detect unusual patterns, retrieve pieces of information, summarize key points, and even handle emojis!
Key techniques to learn in this tutorial:
● Recognize unstructured feedback and be able to confidently analyze free-form such responses ● Use lexical resources (e.g., NRC) to attribute emotion, sentiment and intensity to illustrate the consumer’s experience with a product or service ● Quickly pre-process and cluster consumer responses, extract and visualize keywords with ● Build a model that incorporates the structured data types (e.g, review rating score and time posted), unstructured (review headline and content), and engineered features to predict if review will be useful to other readers