Graphistry 2.23.4: HyperNetX, data bridge, fuzzy matching, and APIs

Posted by Graphistry Team on Aug 16, 2019

Version 2.23.4 brings a bunch to enterprise analysts and developers! Read on to learn more about those, and see full release notes at our new release notes page

 

Graphistry 2.23.4 is another big one.

Enterprise users will be happy to hear about the data bridge, while  fuzzy matching and new URL parameters should make life better for scripters.  This release also provides an example of working with PNNL's new HyperNetX Python package.

 

 LAUNCH IN YOUR AWS 

 

1. Reveal fuzzy links

63158061-4f3b5480-bfcd-11e9-9693-1c5bfcd26b3d

Dirty data is real: when correlating results between steps, or trying to make sense of a data dump, the same entity may show up with multiple conflicting values. For example, when expanding on the history for user "fred", a search result might hit on activity by account "Fred Astaire". Graphistry already helps successfully bring in  search results from across data sets... but it may not be clear why a certain entity is in your graph.

 

To reveal the link between "fred" and case-insensitive substring match "Fred Astaire", just toggle on new "Connect fuzzy matches" control in a pivot. The new options are:

  • None (default): Only link exact matches, maintaining today's behavior
  • All: When a pivot adds a node, check if any nodes in the current or earlier pivots are a case-insensitive contiguous substring match with it. If so, add a fuzzy link.
  • Expansions:  For "expand" pivots, you can instead choose to only check for fuzzy matches on the entities being expanded upon. 

The Splunk, HTTP, and manual data pivots support fuzzy matching. When creating automations for other users, to avoid surprises, we recommend keeping fuzzy matches off by default. Instead,  only enable them for steps you found to be important, such as the above name expansion pivot. We look forward to your ideas around this feature!

 

2. Graphistry Data Bridge

SPLUNK_USE_PROXY=true

Want to run Graphistry in the cloud, but connect to restricted on-prem APIs? Standard Graphistry connectors already support that. But, in cases such as when an on-prem firewall restricts traffic to outbound-only or HTTPS-only, the new data bridge can help. Shipping as a simple Docker container -- so minimal server requirements and footprint -- you can quickly install a data bridge in your restricted environment and switch a connector to bridged instead of direct. 

The data bridge is available to all Graphistry cloud and enterprise users and is easy to setup. It is initially available for Splunk users, and we welcome feedback on which connectors to prioritize next.

 

3: HyperNetX Integration Demo

hg.hypernetx_to_graphistry_bipartite(H.dual()).plot()

hg.hypernetx_to_graphistry_nodes(H).plot()

 

PNNL's HyperNetX is similar to the Python NetworkX API, except for hypergraphs: graphs where an edge may cover more than 2 nodes.

PyGraphistry now includes a sample integration for exploring HyperNetX graph in two ways:

  • hypernetx_to_graphistry_bipartite(hn):
    Convert a HyperNetX graph to Graphistry/Pandas node/edge property table dataframes representing a bipartite graph between 
    Hypernodes and Hyperedges

  • hypernetx_to_graphistry_nodes(hn): 
    Convert a HyperNetX graph to Graphistry/Pandas node/edge property table dataframes representing a graph of Hypernodes that are connected whenever a Hyperedge includes them, effectively treating each hyperedge as a strongly connected component over its hypernodes

You can see a demo notebook on the PyGraphistry github and try it live with Graphistry Core distributions, including on AW Marketplace.

 

4. New URL API parameters

=>

 

Graphistry now supports even more developer control through 15 new URL API parameters. These correspond to settings users can manipulate in-tool:

 

scene: The URL API can now control scene panel options : pointSize, pointOpacity, edgeCurvature, edgeOpacity, showArrows . The names match those in the React API and the controls in the scene settings panel . The URL API arguments cascade during pageload, acting between uploaded dataset configuration settings  and session workbook settings.

histograms: In addition, you can now set showHistograms.

For example, to load the page that shows the graph and hides UI chrome, set `&menu=false&showHistograms=false`

labels: You can now also control label settings from the URL API: labelOpacity, labelColor, labelBackground, showLabels, showLabelOnHover, showLabelPropertiesOnHover, showPointsOfInterest, showPointsOfInterestLabel, pointsOfInterestMax

For example, you disable points of interest on load: `&showPointsOfInterest=false`.

 

5. More 2.0 stability and goodness

We've been cranking away on 2.0, and if you see anything, let us know: each release keeps getting better and better!


  • Stability:
    • Fixes to in-tool docs links

    • Docs for new embedding URL controls

 

Read the release notes for further details.

 

Curious to turn your graph DB data into GPU-accelerated visual graph analytics sessions & dashboards? One-click launch on AWS now! 

LAUNCH IN YOUR AWS 

 

 

Topics: Release, HyperNetX, Data bridge, Fuzzy matching, API