We have enhanced our login page with additional information for those who prefer not to sign up immediately. The updated starting page layout now includes an informative video and a 3D demo of our knowledge graph, offering a glimpse into the powerful features of PrivateAI.com. This allows curious visitors to explore and understand the platform's capabilities before committing to registration. Check out these new elements to see how PrivateAI.com can benefit you.
In addition to Sign in with Google, we added an option to sign in with Twitter/X.
In a 2D space, visualizing densely connected nodes and edges can lead to clutter and overlapping elements, making it difficult to discern relationships and patterns. A 3D representation allows for a clearer separation of nodes and edges, reducing visual clutter and enabling a more intuitive understanding of the graph’s structure.
Interactive exploration is greatly enhanced in a 3D environment. Users can rotate, pan, and zoom the graph, exploring different perspectives and focusing on areas of interest. This interactivity makes it easier to identify important nodes, central connections, and hidden patterns that might be missed in a static 2D view.
You can now easily export (download) the knowledge graph structure dataset in CSV/XLSX format.
Exporting the knowledge graph structure CSV dataset has a number of applications in the context of machine learning. Graph structure data can be utilized to create features representing nodes and edges, which capture relationships and dependencies in the data. Additionally, advanced graph-based features such as node centrality, clustering coefficients, and path lengths can be derived, enriching the feature set for predictive models.
Graph embedding is another important application. Exported graph data can be used to generate embeddings for nodes using algorithms like Node2Vec or GraphSAGE, which then serve as input features for machine learning models. Graph Neural Networks (GNNs) can also process the structured data to learn representations that encapsulate the graph’s structure, enabling tasks like node classification, link prediction, and graph classification.
We have introduced a graph comparison tool to enhance your data analysis experience, especially beneficial for those working with large datasets. Simply select the knowledge graphs you want to compare or upload two exported files to see the results
The comparison tool provides a semantic similarity analysis, with the following output categories:
This tool is particularly helpful for evaluating datasets and avoiding duplicate data when using datasets for training large language models (LLMs). By identifying the degree of similarity between datasets, you can ensure diverse and unique data inputs, enhancing the quality and effectiveness of your LLM training process.
Now you can open and review each arXiv article right from PrivateAI dashboard.
We have fully redesigned our search feature to give you enhanced control over your search results. The search engine has been upgraded for increased precision and now includes filters for a more tailored search experience. You can now search for data from PrivateAI as well as external sources like arXiv.org, with filter options available to select your preferred source. Enjoy a more efficient and customizable search process with these new improvements. The search is now also available for mobile devices.
The updated graph history screen is now in a separate tab for ease of navigation.
Hide the menu sidebar for a more comfortable workflow. Maximize your screen space by hiding the sidebar, allowing you to utilize most of your screen for your tasks. Enjoy an uncluttered and spacious workspace.
