Introduction
Graphwise is happy to announce the immediate availability of GraphDB 10.8, which includes the next-generation Talk to Your Graph functionality. It integrates large language models (LLMs) with vector-based retrieval of relevant enterprise information and precise SPARQL querying of knowledge graphs that hold trusted factual data and domain knowledge. This new feature enables non-technical users to derive real-time insights and retrieve and explore complex, multi-faceted data through natural language.
GraphDB 10.8 also brings seamless, high-availability cluster deployments across multiple regions, ensuring zero downtime and data consistency without compromising performance, enhanced SPARQL endpoint integration, and Google Cloud Storage backup compatibility.
Talk to Your Graph 2.0
By leveraging knowledge graphs for retrieval augmented generation (RAG), organizations can enhance answer quality and augment their proprietary information with machine-interpretable domain knowledge. Graphs help connect the dots across diverse data sources, surface knowledge, and derive competitive insights. This may be why Gartner is putting knowledge graphs at the epicenter of their 2024 Impact Radar right next to Generative Artificial Intelligence (GenAI).
“Graphwise’s latest version of its GraphDB engine enables us to experiment, prototype, and showcase the potential of GraphRAG to deliver accurate, explainable, and replicable research retrieval and insights,” said Gary Leicester, Content Metadata Controller at CABI – an international, inter-governmental, not-for-profit organization that provides information and applying scientific expertise to solve problems in agriculture and the environment. “Leveraging Talk-to-Your-Graph 2.0 technology allows us to demonstrate this potential rapidly and intuitively, paving the way for a production-ready solution.”
Following closely on the heels of the formation of Graphwise – the result of the merger between Semantic Web Company and Ontotext – the latest features make accessing complex datasets easier, by improving user interactions with knowledge graphs that enable natural, context-rich conversations through customizable, intelligent agents. Non-technical users can now self-service their data retrieval and analysis tasks instantly, removing the delays and the overheads that occur when relying on data management staff. Users can also easily ask for explanations, evidence, and clarifications to check supporting information and gain confidence in the answers provided.
GraphDB 10.8 reduces the R&D time for GenAI applications by offering a no-code framework based on GenAI-powered agents that intelligently combine multiple retrieval methods to deliver context-rich conversations and reduce non-determinism. To help AI developers fine-tune conversational agents (chatbots), it automatically heals retrieval query errors and provides quick access to the underlying method invocations, results, and error messages.
This new version of GraphDB was designed not only for data scientists, knowledge engineers, and enterprise users working with large knowledge graphs, but also for decision-makers in data-intensive industries such as financial services, manufacturing, and life sciences who rely on sophisticated data insights and need intuitive, conversational access to information. Key features include:
- Knowledge graph-driven conversational AI through intelligent agents: Combining the latest in RAG technology, the solution enables agents to retrieve data in real time and deliver precise and context-rich responses, all within a conversational, AI-driven format.
- Diverse query methods feeding flexible retrieval workflows: Each agent leverages a full range of query methods—SPARQL for structured data, graph embedding-based vector similarity search for focused, open-ended questions, and full-text search for broader open-ended inquiries. This versatility enables agents to interpret and respond dynamically across a wide spectrum of inquiries, from pinpointing related concepts to analyzing extensive datasets.
- Multi-agent personalization with memory: Users can set up multiple agents, each tailored to their specific data and domain-specific needs. With unique instructions and memory capabilities, this allows for seamless adaptation to various workflows and data interactions.
“This release of our graph database engine is particularly important because it removes technical barriers and allows users to interact conversationally with data without needing query-building expertise. While we launched an early version of the Talk-to-Your-Graph tool a year ago, the new version offers much more comprehensive query-answering and increases the range of questions that can be answered. What’s even more important, GraphDB 10.8 will massively reduce the time data scientists need to configure and fine-tune a chatbot,” said Atanas Kiryakov, President of Graphwise. “By accelerating access to data, this release lets users conduct advanced searches quickly and accurately across the knowledge graph. As a result, enterprises can scale data interactions across teams while maintaining customizability to meet specific workflow and business requirements.”
Check out this video to see GraphDB Product Owner Pavel Mihaylov walking you through the new Talk to Your Graph capabilities!
Other exciting features
Multi-region cluster support – GraphDB 10.8 supports high-availability cluster deployments across multiple regions, ensuring zero downtime and data consistency without compromising performance. This enhancement, designed for DevOps teams, uses an optimized architecture and refined cluster protocols to enable efficient read and write operations in multi-region environments.
Enhanced SPARQL endpoint integration – the new release also introduces improved SPARQL endpoint integration, allowing data engineers to incorporate internal data sources into knowledge graphs. This provides seamless use of Ontopic Studio’s SPARQL endpoint in GraphDB via federation with secure JWT token authentication.
Google Cloud Storage backup – GraphDB 10.8 now supports Google Cloud Storage, enabling full compatibility with all major cloud providers (Google Cloud, AWS, and Azure) for cloud backups. This streamlines data storage and backup management in multi-cloud environments, offering greater flexibility and improved disaster recovery options.
Availability on Google Cloud Marketplace – Ontotext’s GraphDB now offers flexible Bring Your Own License (BYOL) and SaaS (Managed Service) options. This lets Google Cloud customers leverage GraphDB’s semantic capabilities through a managed service for streamlined, hands-free deployments or by bringing their licenses for tailored control.
Now that customers can get GraphDB from Google Cloud Marketplace enables them to rapidly deploy a robust knowledge graph solution directly within Google’s environment. The two deployment options let organizations optimize based on data needs, budgets, and technical requirements:
- BYOL (Bring Your Own License): Ideal for enterprises seeking direct control over licensing and resource allocation, enabling GraphDB’s powerful semantic database features within their existing Google Cloud infrastructure.
- SaaS (Managed Service): A fully managed environment where Ontotext handles deployment, scaling, and maintenance, allowing your teams to focus on data insights rather than infrastructure.
Companies across Finance, Healthcare, Manufacturing, and Publishing sectors rely on GraphDB thanks to its performance, compliance with W3C standards, and compatibility with other data systems like Elasticsearch, Kafka, and Lucene. The availability of GraphDB on Google Cloud, in addition to its availability on AWS and Azure, helps enterprises unify and analyze data with efficiency, supporting mission-critical AI and machine learning initiatives with ease.
Several significant bugs related to transaction processing, SHACL rules, cluster replication, and cloud backup consistency have also been resolved, improving both stability and performance.
The development of the features in this release was partially funded by the EnrichMyData (GA nu: 101070284) Horizon Europe research project.
Download GraphDB and start building knowledge graphs for your data management practices!
