What is a Classifier?
Text is unstructured data, meaning it is information that has no pre-set data structure or schema. Therefore it can't be stored, related, and measured in the same way structured data can.
However, if analyzed correctly, it can provide a wealth of insights that other data can’t. Fortunately, Natural Language Processing and Machine Learning can help you transform it into a structure so you can measure, quantify, and ultimately act on it.
This is what Classifiers do in Caravel.
A classifier can be a topic, expressed intent, emotion, sentiment, category, etc. that you want to predict from your text. Caravel then classifies your unstructured text using the classifiers that you've defined to give your text structure and quantify it.
In Caravel there are two types of classifiers: Pre-built Classifiers and Custom Classifiers.
Pre-built classifiers require zero training or setup and can be instantly used to classify your text. They classify common topics like usability, security, and bugs, and common customer intents like upgrading or canceling an account, making a complaint or suggesting a feature.
Learn how to use pre-built classifiers here.
Custom classifiers are trainable and you can give them feedback to learn to predict custom topics or intents on your data. Our custom classifiers are meant to provide accuracy with small sets of training data, meaning you can get good results without spending much time training.
Learn how to set up and train custom classifiers here.
