AI is invaluable for graph analysis, especially in the context of transforming text into graph structures. It enables the extraction of meaningful insights from unstructured textual data and enhances the capabilities of graph-based applications in various domains. Modern LLM-based AI tools are quickly gaining popularity in the graph space but usually, they do not provide a way to efficiently tune the transformation process.

Our Text2graph online tool was created with continuous model learning in mind. Using OpenAI models together with clever prompt engineering for continuous learning is a valuable approach to improve the model's performance and adapt it to specific domains or tasks. No card or API key is required to start. Just sign up for a free Graphlytic Cloud account and you'll get free 50.000 OpenAI tokens to use the tool.

Here's how to use the incremental learning effectively for this purpose:

1. Start with an initial paragraph that covers a typical scenario in your domain. This paragraph can serve as the foundation for your model's understanding of the task.

2. Gradually introduce new paragraphs as part of the graph extraction process. Every paragraph can be up to 500 words long.

3. After every paragraph inspect the generated graph and do these simple actions:

   • Rename entities that have been extracted incorrectly.

   • Merge duplicate nodes and relationships in the graph.

By following these steps and maintaining an iterative and adaptive approach, you can use Text2graph to facilitate continuous learning and enhancement of your text-to-graph extraction capabilities. This approach allows your model to evolve and improve its performance over time, making it more effective in handling diverse and evolving data.

What do you get in return?

1. Browse the graph to see which parts have been extracted from which paragraph. Simply click on a paragraph in the history to highlight parts of the graph. It works also the other way around, click on some node to see in which paragraphs was this node detected.

2. Extracted graphs can be exported as an image, or set of Cypher or Gremlin queries to quickly transfer the graph into your favorite graph database. Use Graphlytic to import this graph into your database.

Try Text2graph online tool for free and let us know what can we improve to make it better or to include more use cases. No card is required and no OpenAI key is required to start using the Text2graph tool. Simply sign up for a free Graphlytic Cloud account and use up to 50.000 OpenAI tokens.