Skip to main content

BLU - Blip Language Understanding

Blip Help
  • Updated

Index:

  1. Demonstration video
  2. Introduction
  3. How it works
  4. Installation and Configuration
  5. Support

 

Demonstration video

Introduction 

 

BLU (Blip Language Understanding) is Blip's native AI provider, developed to enhance the comprehension and processing capabilities of chatbots in natural language. As an NLU (Natural Language Understanding) provider, BLU can perform text classification based on intents and entities.

 

With BLU, developers can create smarter and more efficient chatbots that can understand and respond accurately to user requests. By utilizing advanced natural language processing techniques, BLU empowers chatbots to accurately interpret the underlying intents in user text and identify relevant information through entities.

 

Example:

Let's consider a scenario where a user is interacting with a technical support chatbot from an electronics company. The user sends the following message: "My phone is not charging." Using BLU, the chatbot is able to understand the intention behind that message, which is "Problem with phone charging." Additionally, BLU is able to identify the relevant entity "phone" mentioned in the user's message.

 

Based on the identified intention and entity, the chatbot can provide an appropriate and targeted response to resolve the user's issue. For example, the chatbot can offer a series of solutions such as checking the charging cable connection, restarting the device, or contacting the company's technical support.

 

How it works

BLU is compatible with all channels connected to the Blip platform, allowing you to leverage its functionalities in a wide range of chatbot environments. Now, let's explore the functionalities offered by BLU:

 

  • Training/Model Creation: 

The BLU Plugin offers the ability to train and create custom language models for your chatbots. With this functionality, you can feed BLU with relevant training data for your chatbot's context, enabling it to be more accurate and efficient in understanding intentions and entities. This functionality is essential to make use of the others.

 

  • Classification Testing: 

BLU allows you to evaluate the quality of your text classification model. With individual tests, you can provide example phrases and verify how the BLU model classifies those phrases into intentions and entities. Additionally, you can also perform batch tests (available in beta) by providing a set of phrases for BLU to classify and evaluate the overall performance of the model.

 

  • Usage and Performance Reports: 

The BLU Plugin provides detailed reports on the usage and performance of your language model. These reports offer valuable insights into the accuracy of classifications, the distribution of identified intentions and entities, as well as information about resource usage. These reports help you monitor and optimize the performance of your chatbot over time. Additionally, they will soon also help suggest modifications to further improve your bot.



Installation and Configuration

 

When activating the extension through the Blip Store, it should be installed in the same bot where the Knowledge Base is located (in the AI tab of Blip), which is the bot where NLP calls will be made. Follow the steps below to start using BLU in your chatbot.

 

  • The Plugin will use the registered Knowledge Base in Blip to train a new BLU model, which will be used for text classification.

  • Once the model is trained, access the Details section of the model you have created.

 

  • Here you can also perform tests with your model by selecting the "Test" or "Batch Test" options.

 

  • Click on the "Connect Model" option. Important information for model integration will be displayed.

 

  • Save the following information displayed: HTTP call URL, request header, and request body. This data will be needed to configure the HTTP call in the Builder.

 

Now, you need to register the information obtained earlier in an HTTP Call action in the Builder. Follow these steps:

 

  • Open the Builder and access the desired flow.
  • Add an HTTP Call action.
  • Fill in the necessary fields with the information you saved: HTTP call URL, request header, and request body, and save the response in a variable.
  • This variable will contain the identification information of the BLU model and can be used later in the chatbot for decision-making based on the identified Intents and Entities.

 

Support 

 

If you have any questions or encounter any issues with the extension, please contact us at blipaisuite@blip.ai.

 

For more information, visit the discussion on the subject in our community or the videos on our channel. 😃

Related articles