Announcing Raphtory v0.14.0: Faster Queries, Powerful Filtering, and New Algorithms

Announcing Raphtory v0.14.0: Faster Queries, Powerful Filtering, and New Algorithms

We are excited to announce the release of Raphtory v0.14.0, an update focused on delivering significant performance boosts, more expressive query capabilities, and a host of new features to expand your analytical toolkit.

This release introduces a powerful caching mechanism to accelerate your algorithms, adds granular filtering for nodes and edges, and brings new, cutting-edge algorithms to the platform. Let's explore the major changes.

🚀 Headline Feature: Blazing-Fast Queries with Cached Views

The most significant new feature is the .cache_view() function. This creates a lightweight index of the nodes and edges in your current view, dramatically speeding up global algorithms and complex analytical pipelines.

If you are running multiple algorithms or queries on the same subset of your graph, using .cache_view() first will provide a substantial performance improvement.

🔬 More Powerful Queries: Advanced Filtering and New Functions

We've supercharged the query engine with more powerful and flexible filtering capabilities.

• Node and Edge Property Filtering: You can now filter nodes and edges based on their property values. This includes checking for property existence, using comparison operators (e.g., less than, greater than, equal to), and filtering based on inclusion in a list.
• New create_node Function: We've added a create_node function that, unlike add_node, will fail if the node already exists. This is especially useful for more robust integrations, such as in GraphQL.
• Convenient import_as Functions: New import_as functions allow you to rename nodes and edges when importing them from one graph to another, simplifying data integration workflows.

🧠 New Algorithms for Deeper Insights

This release expands our library of built-in algorithms, enabling new types of analysis:

• FastRP Embedding: We've implemented FastRP, a cutting-edge algorithm for generating network embeddings via very sparse random projection. This is a fast and accurate way to create vector representations of your graph for machine learning tasks.
• Maximum-Weighted Matching: You can now find the maximum-weighted matching in your graphs, a classic algorithm with applications in everything from logistics to bioinformatics.

💥 Breaking Changes

To improve our API, we've introduced a few breaking changes:

• The function Graph.add_property has been renamed to Graph.add_properties to better reflect its capability to add multiple properties at once.
• Direct creation of a RemoteGraph has been removed. Remote graphs can now only be created through the client, ensuring a more consistent and secure connection.

Other Notable Improvements

This release is packed with many other enhancements:

• Python API: Numerical properties now return NumPy arrays for better performance and memory usage. The secondary time index is now exposed, and we've added wrappers to construct vectors from any Python iterable.
• UI Updates: The UI now includes a "Saved graphs" page to easily open, view, and get top-level statistics for graphs stored on the server.
• GraphQL Enhancements: We've added new functions like edge_id, explode, and namespace to give you more power and flexibility when querying your graph via GraphQL.
• Core Engine: We've made numerous under-the-hood improvements, including faster subgraph creation, faster node counting, and exposing more low-level APIs for easier extension development.

Get Started with v0.14.0 Today!

This release is a major step forward in performance, flexibility, and analytical power. The new caching, filtering, and algorithms unlock exciting new possibilities for your graph projects.

For a complete list of all changes, check out the full release notes on GitHub.

We're excited to see what you build. As always, we welcome your feedback on GitHub or in our community channels. Happy analyzing!

‍