Creating Ensembles

With ensembles you can combine a number of different classification techniques to get optimal results with zero training.
Ensembles are incredibly flexible and quick to setup.
There are four techniques you can combine within ensembles: Zero-shot Classification, Character and Keyword Matching, Sentiment Classification, and Prebuilt Model Classification.
Zero-shot Classification
To put it simply, Zero-shot Classification is classification on-the-fly: It enables custom classifications that work for your data, for any set of categories you can come up with without the requirement of annotated data.
Zero-shot uses a different approach when making predictions. Instead of requiring you to train it, it starts with just the label and makes predictions based on the information it understands from the label name.
Zero-shot Classification is used when you input a value into the Natural Language Definition field of the labels in your ensemble. Learn more about using zero-shot and creating Natural Language Definitions here.
Character and Keyword Conditions
With Keyword matching you can search keywords in your text using a number of queries. Caravel provides the ability to search for characters using Contains conditions, matching on approximate keywords using Fuzzy Matches conditions, and exact keyword matching using Search Matches conditions.
With both Fuzzy Matches and Search Matches conditions you can assemble search queries to match prefixes, search for phrases within a specified distance, and use and / or operators.
Read more about your options here.
Sentiment Conditions
You can use sentiment classification in your Rules & Conditions in ensembles to further refine your ensemble's classification.
Prebuilt Classifier Conditions
You can use prebuilt models in your Rules & Conditions in ensembles to further refine your ensemble's classification.