We are delighted to launch Raphtory v0.15.0! This release introduces "Perspectives," a powerful new way to work with your graph data, alongside major upgrades to our Python API for advanced analytics and a shift to the high-performance Parquet format for disk operations.

This update is designed to give you more powerful, flexible, and efficient ways to model and analyze complex, evolving data. Let's get into the details.

🚀 Headline Feature: Unlocking Deeper Insights with Perspectives

The headline feature of v0.15.0 is the introduction of Perspectives. A Perspective allows you to create multiple, distinct "views" of your graph from the same underlying dataset.

You can now model different types of relationships or entities without altering the source data. For example, in a social network, you could create a "user" perspective and a "post" perspective, each with its own properties and connections, all derived from the same stream of events. This enables more sophisticated and organized analysis of multifaceted datasets.

🐍 Major Python API Upgrades: group_by and agg

We've supercharged our Python API with powerful new aggregation capabilities, bringing a familiar, dataframe-like experience to your graph analysis.

You can now use group_by() and agg() on algorithm results to perform complex aggregations with ease. This significantly simplifies post-processing and allows you to roll up data, calculate group statistics, and generate insights directly within your Raphtory workflow. We've also added a convenient to_df() function to the main graph object for easy data extraction.

💥 Breaking Changes and New On-Disk Format: Hello, Parquet!

This release brings important changes to how Raphtory stores data on disk, enabling better performance and future-proofing the platform.

• New On-Disk Format: We have changed the on-disk representation of a Raphtory graph to use the industry-standard Parquet format. This results in significantly faster load times and reduced memory usage. The new functions are save_to_disk() and load_from_disk().
• API Updates: save_to_zip() and save_to_file() are now deprecated. We've also clarified our Python API for time windows, renaming Window.window() to Window.tumbling() and Range.window() to Range.rolling() for better clarity.

Other Notable Improvements

This release is packed with many other enhancements:

• New Algorithms: We've added two new similarity algorithms: Jaccard Similarity and Overlap Coefficient, expanding your toolkit for node and neighborhood comparisons.
• Core Engine: The on-disk format improvements have also reduced the memory footprint of loaded graphs.
• UI Enhancements: The Raphtory UI now includes a timeline view for exploring the temporal aspects of your graph.
• Bug Fixes: This release includes various bug fixes, including a critical fix for our connected components algorithm (Component.undirected()).

Get Started with v0.15.0 Today!

The introduction of Perspectives and powerful new aggregation functions makes v0.15.0 a landmark release for Raphtory. We're continuing our mission to provide the most powerful and flexible temporal graph platform available.

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

We can't wait to see the new, complex analyses you'll perform with these features. As always, your feedback is crucial, so please reach out on GitHub or in our community channels. Happy analyzing!