About 80% of business-relevant information originates in unstructured form – primarily text. Feedback contained in places like social media posts, online reviews, emails, and survey verbatims contain insights that go beyond a score or 5-star rating. They provide the texture and insight a business needs to understand “Why” a customer likes or dislikes any part of the customer experience.
Text Analytics allows companies to uncover countless issues, opinions, and opportunities that would traditionally be buried in pages and pages of text data. Text Analytics allow to uncover emerging issue trends before they balloon into giant problems and capitalize on opportunities that could otherwise be missed.
“Precision” and “Recall” are the two high level and common metrics used to measure text analytics software performance.
Precision is the proportion of comments that were correctly categorized into a given topic. For example, if a topic analysis system identifies 100 references to the topic “staff attitude” and 90 of the identifications are correct, then precision for this topic is 90%.
Recall measures the completeness of your Text Analytics system. If there are actually 120 true references to a topic “staff attitude,” for example, then the system recall for this topic is 75% (90/120.)
Read full post here
Text Analytics allows companies to uncover countless issues, opinions, and opportunities that would traditionally be buried in pages and pages of text data. Text Analytics allow to uncover emerging issue trends before they balloon into giant problems and capitalize on opportunities that could otherwise be missed.
“Precision” and “Recall” are the two high level and common metrics used to measure text analytics software performance.
Precision is the proportion of comments that were correctly categorized into a given topic. For example, if a topic analysis system identifies 100 references to the topic “staff attitude” and 90 of the identifications are correct, then precision for this topic is 90%.
Recall measures the completeness of your Text Analytics system. If there are actually 120 true references to a topic “staff attitude,” for example, then the system recall for this topic is 75% (90/120.)
Read full post here
RSS Feed