Memgraph has released an update to its graph database. The update aims to make it easier to build AI solutions powered by graph technology.

Memgraph is an open source graph database that can be used for real-time streaming. It is compatible with Neo4j, is ACID compliant and uses the Cypher query language for structuring, manipulating and exploring data. Memgraph is built in C/C++ and uses an in-memory first architecture. It allows users to run Python, Rust, and C/C++ code natively, and can ingest data from sources including Kafka, SQL and CSV files.

Memgraph features include support for deep-path traversals, and the ability to use advanced capabilities such as accumulators and path filtering without adding additional application logic. It comes with native support for machine learning, streaming and dynamic algorithms. It supports multi-tenancy use and high availability replication, and offers role-based and label-based access control.

The updated version, Memgraph 3.0 combines vector search and knowledge graphs with LLMs, which the company's CEO, Dominik Tomicevic, says addresses the limitations of LLMs (large language models) such as hallucinations and inability to keep up with business change.

The main improvement to this release is the addition of
Retrieval-Augmented Generation in graph (or GraphRAG), enhance reasoning. This is how the updated version reduces hallucinations, and works securely within an enterprise's specific context and data.

Tomicevic says that most LLMs rely on rigid probabilistic frameworks, meaning that updating the base model with new data is both computationally expensive and impractical, making it a significant limitation for businesses.

Memgraph 3.0's incorporation of RAG in graph mean business users can more easily create knowledge graphs that enhance LLMs, while preventing the accidental exposure of proprietary information.

Knowledge graphs can be used to represent data held within documents and the metadata relating to the documents.
Memgraph 3.0 enables the system to navigate the internal knowledge graph and instead of just referencing similar user-generated questions, it guides the system to pinpoint the network locations where probable answers reside. The use of RAG also means when new data is added, the apps don't require retraining of the LLM.

Tomicevic says the combination makes it easier to build AI applications:

"it's all bundled, off the shelf, and open source. Developers can dive in without hesitation and start creating chatbots and AI agents today."

 As a Memgraph customer, David Meza, Head of Analytics Human Capital at NASA comments:

"At NASA, we are integrating Memgraph in our Human Capital Intelligent Query System to efficiently manage our human capital knowledge graph, enabling faster retrieval of relevant information for employees. Its graph-based approach allows us to keep track of real-time updates, ensuring accurate connections between various policy documents and data sources. By incorporating Memgraph into our RAG process, we enhance our system’s responsiveness and better address NASA’s knowledge extraction without requiring extensive manual data coordination." 

Memgraph 3.0 is available for download now.

More Information

Memgraph Download Page

Memgraph On GitHub

Memgraph Website

Related Articles

Google Launches Spanner Graph

Graph Query Language Gets Official Adoption

Neo4j 5 Adds Autonomous Clustering

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info