Skip to content

Releases: memgraph/mage

Release 1.3

23 May 15:42
@jmatak jmatak
v1.3
1d0bce3
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
Release 1.3

MAGE goes to GPU - Annoucing MAGE 1.3

After algorithmic update, here comes one scalability and speed update. With the integration of NVIDIA cuGraph - collection of GPU accelerated graph algorithms created from NVIDIA engineers. In this update, Memgraph is brining several supported algorithms with GPU accelerated computations.

Supported algorithms

Integration with NVIDIA cuGraph enables Memgraph to scale fast and provide trivial interface for running the algorithms throught the openCypher query language. Different algorithms are supported in this release, more will come in the future where we'll follow the newest NVIDIA releases.

The list of supported algorithms

The list of supported utility tools:

What's Changed

Full Changelog: v1.2...v1.3

Contributors

gitbuda, jmatak, and 3 other contributors
Assets 2
Loading

Release 1.2

20 Apr 10:56
@jmatak jmatak
v1.2
e6c024e
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
Release 1.2

Happy to present our newest release - MAGE 1.2.

Lots of things are new compared to the previous release 1.1. With this release, MAGE is bringing novely in various different areas:

  • Dynamic Graph Algorithms
    • Temporal Graph Networks (TGN)
    • Dynamic Betweenness Centrality
    • Dynamic Katz Centrality
  • Machine Learning
    • Temporal Graph Networks (TGN)
  • Static Graph Algorithms
    • Louvain Community Detection
    • Maximum Flow
    • Static Katz Centrality
  • Utility
    • Import/Export module (JSON)

Dynamic Graph Algorithms

Temporal Graph Networks (TGN)

This algorithm brings a huge impact on the current graph analytics setting because it combines the best of two worlds - streaming with temporal data and graph machine learning. Use it to predict, classify, and embed nodes and edges in temporal/streaming use-case settings.

Dynamic Betweenness Centrality

Betweenness centrality is powered by the ratio of shortest paths traversing through the node of interest. By isolating the registered changes in the biconnected component, the dynamic algorithm saves the computational time and space when applying updates on a changed graph.

Dynamic Katz Centrality

Katz Centrality sets a higher rank to the nodes with a high amount of close-flowing paths. The more low-order paths we have, the more central the node is. To adapt the scale and calculate fast, the algorithm is approximative, which means that the calculated centralities may differ from the true values, but the rankings are preserved.

Machine Learning

Temporal Graph Networks (TGN)

Temporal Graph Network is the machine learning model developed by Twitter engineers - Temporal Graph Networks for Deep Learning on Dynamic Graphs by Rossi et.al.. The model allows the possibility of handling different temporal events occurring on graphs, such are node features updates, node deletions, edge deletions, and so on, and adapting the under-the-hood model for downstream prediction/classification task. Temporal Graph Networks are powered by Pytorch.

Static Graph Algorithms

Louvain community detection

The method that calculates communities by updating the modularity - measurement of density within and outside communities.

Maximum flow

An algorithm capable of calculating the maximum possible flow through relationships with the weighted capacity. This is a powerful tool when analyzing the flow of any resource in movements - like gas or electricity networks.

Static Katz Centrality

The static implementation of Katz Centrality - measurement proportional to the number of closer paths flowing through the node of interest.

Utility

Import and export module - with just the one Cypher query the whole graph can be exported in the file and then imported when needed again. Importing and exporting currently work with JSON format.

What's Changed

New Contributors

Full Changelog: v1.1.1...v1.2

Contributors

jmatak, brunos252, and 3 other contributors
Assets 2
Loading

Release 1.1.1 - Memgraph 2.2.0 integration

24 Feb 11:35
@jmatak jmatak
v1.1.1
6aa2d3a
Compare
Choose a tag to compare
Release 1.1.1 - Memgraph 2.2.0 integration

What's Changed

  • Version of Memgraph bumped to 2.2.0
  • Importing and exporting JSON data from json_util by @katarinasupe in #98
  • New deployment build for ARM by @jbajic in #122

Full Changelog: v1.1...v1.1.1

Contributors

jbajic and katarinasupe
Assets 2
Loading

Release 1.1.

10 Dec 15:27
@antoniofilipovic antoniofilipovic
v1.1
da0e4d4
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
Release 1.1.

Highlights

We are excited to announce the release of MAGE 1.1.

MAGE has acquired a few new skills since release 1.01 that he is excited to share with you:

  • Approximative streaming PageRank
  • Dynamic community detection
  • Dynamic Node2Vec
  • Node2Vec

Dynamic Node2Vec

Mapping graph entities into vector space makes it easier to process graph data. When the system is organized in such a way that data in the form of connections arrive in a very short period, then Dynamic Node2Vec can serve to map them into vector space in real-time. The local change algorithm solves the problem of scalability and stability that standard methods would have.

Dynamic PageRank

Dynamic PageRank is made as a substitute for its classic version which instead of restarting when a new version of the graph appears, just updates over the changes that have taken place. The algorithm is based on the work of Bahman Bahmani et. al. called "Fast Incremental and Personalized PageRank".

Dynamic Community Detection

Implementation of "LabelRankT: Incremental Community Detection in Dynamic Networks via Label Propagation" is based on a label propagation framework made for detecting communities on the graph. Made in C++ the algorithm shows the benefits of calculating communities on streaming data.

Node2Vec

The node2vec is a semi-supervised algorithmic framework for learning continuous feature representations for nodes in networks. The algorithm generates a mapping of nodes to a low-dimensional space of features that maximizes the likelihood of preserving network neighborhoods of nodes. By using a biased random walk procedure, it enables exploring diverse neighborhoods. In tasks such as multi-label classification and link prediction, node2vec shows great results.

What's Changed

New Contributors

Full Changelog: v1.0...v1.1

Contributors

tonilastre, antaljanosbenjamin, and 5 other contributors
Assets 2
Loading
0 Join discussion

Release 1.01 - Memgraph 2.1 integration

22 Nov 15:41
@jmatak jmatak
v1.01
073c79f
Compare
Choose a tag to compare
Release 1.01 - Memgraph 2.1 integration

Integration of MAGE 1.0 with Memgraph 2.1

Assets 2
Loading

Release 1.0.

28 Oct 12:10
@antoniofilipovic antoniofilipovic
v1.0
002fc0f
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
Release 1.0.

What's Changed

Full Changelog: v0.1...v1.0

MAGE now supports Memgraph 2.0

Contributors

jmatak
Assets 2
Loading
0 Join discussion

Release 0.1.

28 Oct 12:09
@antoniofilipovic antoniofilipovic
v0.1
c976ed4
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
Release 0.1.

What's Changed

New Contributors

Full Changelog: https://github.com/memgraph/mage/commits/v0.1

Contributors

gitbuda, MasterMedo, and 7 other contributors
Assets 2
Loading
0 Join discussion