March 2010 Summaries

Google is currently accepting proposals for Summer of Code projects with Neo4j, a graph database. Two proposed projects are already on board: one with uDIG to make Neo4j a spatial database capable of complex algorithms like routing and N-dimensional indexing; the other with Gephi to enable visualization of graphs directly from Neo4j. These projects aim to make Neo4j compatible with OpenGIS and GeoTools standards, respectively, allowing it to be used as an alternative to PostGIS and other GIS backends. The opportunities for students to work on these projects are expected to be popular, so it's recommended to submit proposals early.

The South by Southwest conference hosted a fun 45-minute clip of a NOSQL smackdown between representatives from Cassandra, MongoDB, CouchDB and SimpleDB. The discussion showcased the strengths and weaknesses of each system, with Werner Vogels, CTO of Amazon and expert in large-scale distributed systems, providing insightful comments on graph databases, particularly Neo4j, which he believes excels at handling multiple relationships and connections.

The Neo4j graph database is suitable for modeling hierarchical data, such as categories and products. It stores data as nodes and relationships between them, with key-value properties on both nodes. The Neo4j Python language bindings are used to interact with the database. A common Entity-Relationship Diagram can be applied to model categories and products. The data model consists of entities (categories, products, attribute types) and relationships (between categories and attribute types). The graph structure lends itself well to indexing. Subreference nodes are created to connect multiple entry points to the graph. These nodes are lazily created when requested. Attribute definitions are modeled as relationships between a category and an attribute type node. A traverser is used to fetch all attribute definitions which apply to a product. The Neo4j database can be used to store hierarchical data, such as categories and products, in a more efficient and scalable way than traditional relational databases.

The meetup has been rescheduled from Thursday, March 25th to Wednesday, March 31st due to the speaker's unavailability. The event will take place at Twitter HQ in San Francisco, starting with pizza at 6:30pm and featuring a Neo4j presentation at 7pm, followed by beer, discussions, and code sharing. Interested individuals are invited to register and bring their laptops to learn more about graph databases and explore the possibilities of using graph technologies for their applications.

In 2010, the Neo4j.rb JRuby bindings for the Neo4j graph database underwent two releases, with version 0.4.1 introducing migrations and a batch inserter for large data volumes, improving traversal performance and adding relationship indexing options. The neo4jr-simple wrapper was also released in late 2009, providing a simple interface to Neo4j, and is currently available in version 0.2.1. These developments were made possible by the contributions of many individuals, including Andreas Ronge and Matthew Deiters.