How to create a knowledge base?

This guide details the steps involved in creating a knowledge base within your ZBrain account.

Getting started

• To begin, log in to your ZBrain account.

• Once you have successfully logged in, click 'Knowledge'.

• Click the 'Create' button to initiate the process of setting up a new knowledge base.

Uploading document

• You will be directed to a screen where you can either upload a document from your device or import it from a data source.

Data source configuration

This page allows users to configure the foundational elements of a new knowledge base.

• After uploading or importing the data, you will be prompted to provide a name and description for the chosen file/data.

• To upload additional documents to the knowledge base, click the ‘Add More’ button located below the uploaded documents. Select the documents from your device to add them.

Additional features

Document summarization

• Enable document summarization by toggling the dedicated switch.

• Select an appropriate large language model to perform the summarization process.

• This feature creates concise overviews of lengthy documents for easier comprehension.

Automated reasoning policy

• Create an automated reasoning policy by activating the feature toggle.

• An automated reasoning policy consists of predefined rules, conditions, and variables that guide the system's reasoning process when responding to queries.

• It extracts structured data from the knowledge base, applies logical reasoning, and ensures responses are accurate and consistent.

• This policy governs how the system interprets information, processes queries, and delivers answers based on established knowledge and logic.

Improve efficiency using Flow

• Enable the ‘Improve Efficiency Using Flow’ option to streamline and enhance the process of transforming documents into refined knowledge bases.

• This feature leverages predefined or custom flows from the Flow Library to automate data extraction and analysis, converting raw documents into structured, actionable insights.

• It is essential for users seeking to create efficient knowledge bases by applying standardized data processing techniques such as text extraction, image analysis, and language model-based summarization.

• By incorporating flows, you optimize the data refinement process, transforming input data into a well-organized, accessible knowledge base. This enhances the solution’s ability to handle various data formats, including text, images, and structured content, ultimately improving operational efficiency and decision-making.

• When you activate the toggle to enable this feature, a button labeled ‘Add a Flow from the Flow Library’ will appear. Clicking this button will open the ‘Add a Flow’ panel.

Types of flows

There are two types of flows available:

ZBrain Flows

ZBrain Flows offer predefined automation solutions for common data processing tasks. Users can choose from the following options:

• OCR (Optical Character Recognition)

  • Purpose: Recognizes and extracts text content from images or documents. This is particularly useful for digitizing physical documents or documents containing non-editable text (e.g., scanned PDFs).

  • Functionality:

    • Extracts text from images or scanned documents.

    • Enables further processing or analysis, such as searching, summarization, or automated reasoning.

• Analyze each page as an image using an LLM

  • Purpose: Treats each document page as an image and processes both the visual and textual content for detailed analysis.

  • Functionality:

    • Converts document pages into a digital format using OCR.

    • Extracted text is analyzed using a Large Language Model (LLM) to:

      • Derive insights.

      • Generate summaries.

      • Classify content based on predefined criteria.

• Extract images from the document and evaluate them using an LLM

  • Purpose: Designed for documents containing images that need to be analyzed.

  • Functionality:

    • Extracts images from the document.

    • Applies an LLM to analyze the images for:

      • Content recognition.

      • Pattern or object identification.

      • Useful for image-heavy documents requiring deeper content understanding.

Custom flows

The custom flows option allows users to create a flow specifically for data extraction, enabling tailored and advanced automation based on unique workflows and processing needs. Users can click on this option to add a custom flow.

Complete all the required fields and click the ‘Next’ button to proceed to the text data refinement page.