Introduction to VKGs

What is a VKG?

VKG stands for Vector Knowledge Graph. VKGs are a way of accessing and storing information in a way to enable the most efficient communication between Lazarus products.

Information is submitted as human-readable text, but stored using vectors, which are essentially multi-dimensional arrays. These vectors are produced by transforming the submitted information through a process called embedding. The vectors provide efficient storage, are a better medium for programs and models to process, and inherently provide a basis of semantic understanding.

Through the VKG UI, after creating and adding data, you will see a 3D interactive space that contains various spheres. Each sphere is called a node. Each node contains a single piece of information that has been added and is stored. Nodes that are closer together are most similar, while nodes that are farther apart are least similar. Nodes may form clusters in which you can be generally confident that share many similarities, whether that be topic, tone of voice, medium, industry vertical, etc.

Structuring a knowledge base as a VKG makes finding, storing, and updating information quick and easy, empowering your enterprise to do more with your data.

Our Edge

The VKG framework developed at Lazarus is among the most advanced knowledgement management systems available on the market today.

We took it a step further by integrating our language model, RikY-Citations, into the VKG framework. This allows users to have natural language conversations with an intelligent AI that also has access to all of the information available in a VKG.

Combining RikY-Citations and VKG technologies allows businesses to consult virtual subject matter experts that can be revolutionary for your business needs.

Use Cases

Domain Specificity

To enhance the relevance of search results, the system is designed to limit outputs to a specific set of documents. This approach effectively filters out irrelevant content, ensuring more precise and controlled search results.

Efficient Dataset Querying

The VKG supports advanced querying of large datasets by leveraging metadata filters. This functionality helps to narrow down search results and eliminate unwanted or extraneous data, thereby streamlining the retrieval process and improving result accuracy.

Browsing and Visualizing Unstructured Data

Navigating through unstructured data can be challenging. Our VKG employs state-of-the-art UMAP embeddings for data visualization, faciliating easier manipulation and interpretation of complex datasets.