Changelog
Version 0.4.5 (Current Stable)
- Upgrade Flink dependency from
1.7.0to1.7.2 - Integration of Cypher on Apache Flink (CAPF) to support cypher queries
- Improvement of CSVDataSource
- Added Graph ETL operators to module
gradoop-data-integration - Added Degree Centrality operator
- Added Random Walk operator
- Added Verify operator to remove dangling edges
- Refactored PropertyValue class according to strategy pattern
- Added overwrite support for HBaseDataSink
Version 0.4.4
- Upgrade Flink dependency from
1.6.0to1.7.0 - Generalized Subgraph, Transformation and Property Transformation operators
- Generalized MetaData construct
- Refactored GradoopId to no longer use MongoDB
- Added RollUp example
- Added support to construct GraphCollection from mutliple LogicalGraphs
- Added simple importers for CSV and JSON files
Version 0.4.3
- Upgrade Flink dependency from
1.5.0to1.6.0 -
gradoop-data-integrationmodule added - Fixed bugs in IndexedCSVDataSink
- Sampling Benchmarks added
- Datagen-package removed from Gradoop
- Clustering coefficient computation addded
- Triangle counting added
- Rollup-Operator added
- Unified aggregate functions now applicable to both aggregate and grouping operator
Location of LogicalGraph and GraphCollection changed! Through a refactoring, the location of the classes LogicalGraph and GraphCollection has changed from package org.gradoop.flink.model.api.epgm to org.gradoop.flink.model.impl.epgm.
Version 0.4.2
- PageRank sampling added
- Mockito added for testing support
- Multiple aggregation functions in a single aggregation operator
- GraphCollection support for CSV Source/Sink
- LGTM code quality badge
- PropertyTransformation operator added
- Weakly Connected Components runner added
- CSV delimiter escaping
CSV format changed! Through the newly added graph collection support for CSVDataSource/Sink and IndexedCSVDataSource/Sink, existing graphs created by these sinks from a previous release have to be transformed into the new representation. A GradoopFormatConverter is available in the gradoop-examples package to do this conversion. Use parameter lgcsv (for graphs created by the CSVDataSink) or lgindexed (for graphs created by the IndexedCSVDataSink) as input format and csv or indexed as output format. The outdated sources and sinks are renamed as LogicalGraphCSVDataSource and LogicalGraphCSVDataSink and will be removed in version 0.5.0.
Version 0.4.1
- Cypher, Subgraph and SNA Benchmarks added
- Predicate pushdown for HBase available
- Split property type and value in HBase table layout
-
NULLandSetsupport in PropertyValues - Code coverage report
- Support graphs without edges
-
.print()added forLogicalGraphandGraphCollection - Aggregation example added
- Sampling algorithms extended incl. Graph density computation
Version 0.4.0
- Upgrade Flink from 1.3.1 to 1.5.0
- JSONDataSource and JSONDataSink are now marked as deprecated
- CSVDataSource and CSVDataSink supporting all available property value types
- IndexedCSVDataSink added
- Added Gradoop-Store module with Apache Accumulo and HBase support
- Added Store Api with manual predicate pushdown (for Accumulo DataSource)
- First Sampling Algorithms
--
--
Bug fixes and support for more Gelly algorithms.
Added Pattern Matching and Frequent Subgraph Mining algorithms.
Major refactoring of internal EPGM representation (e.g. ID and property handling), Equality Operators, GDL-based unit testing
Apache Flink replaces MapReduce and Giraph as operator implementation layer and distributed execution engine.
Added support for HBase as distributed graph storage.
First prototype using Hadoop MapReduce and Apache Giraph for operator processing.
