Updated by Thays Oliveira
As a result of the constant innovation in our Artificial Intelligence studies, a new type of Intelligence is available in our module: the Content Intelligence. With it, it is possible to register informative texts and ask questions to the intelligence based on its content, without having to train it. This is possible due to powerful natural language processing (NLP) engines.
In this article, we'll talk about the main elements of a Content Intelligence and present a brief comparison between Content and Classification Intelligences (also known as Entity and Intent).
Elements of a Content Intelligence
Despite being a new type of Intelligence, several of its elements will be familiar to those who have used the classification modality, such as: name, categories and description, in addition to have support for hundreds of languages and easy integration with Weni Flows.
The biggest differential of a content intelligence, however, are the Knowledge Bases. They are responsible for registering the content and language of an informative text. An intelligence can contain multiple knowledge bases, each containing a subdivision of the main subject of the intelligence and/or a translation of the text of another knowledge base.
An example of using multiple bases is the usage in an educational context. It is possible to create an Intelligence with the subject "High school Chemistry". In it, you can add a base for each topic, such as General, Organic and Inorganic Chemistry.
Another example is related to the day-to-day of a company: an Intelligence of Frequently Asked Questions (FAQ) can be created, storing in each knowledge base a specific subject: billing, sales or service information.
When to use a Content Intelligence
Usually, a Content Intelligence offers good results when fed with texts that bring information about a certain subject.
However, the choice between the two types of intelligences can be analyzed on a case-by-case basis, depending on the needs of your project. To help in the decision process, here is a brief comparison of the two types of intelligence:
Needs to be trained
Ideal for informative texts
Ideal for individual phrases
Information feeding method
Phrases with entities and intents
Confusion Matrix View