SANSA 0.7.1

Dependency Changes

• Apache Spark 2.4.3 → 2.4.4
• Apache Flink 1.8.0 → 1.9.1
• Apache Jena 3.11.0 → 3.13.1

SANSA ML 0.6.0

Features

• Spark
  • Add coverall integration
  • Further improvement of unit test coverage
  • #12 Refactor vandalism detection package
  • #20 Align with RDF layer

Bug Fixes

• #13 Classes with names that only differ in casing
• #15 Hard-coded spatial partitioning value (DBSCAN)
• #17 geospark scope removed

Dependency Changes

• Apache Spark 2.4.0 → 2.4.3
• Apache Flink 1.7.0 → 1.8.0
• Apache Jena 3.9.0 → 3.11.0

SANSA ML 0.5.0

Features

• Spark
  • Numerical outlier detection
  • RDF Graph Kernels (Alpha)
  • Update: knowledge graph embedding support(pre Alpha)
  • Decision trees (pre Alpha)
  • Update: Clustering
    • Power iteration clustering
    • DBScan (experimental)
    • Unified interface for all clustering algorithms

Dependency Changes

• Apache Spark 2.4.0
• Apache Flink 1.7.0
• Apache Jena 3.9.0

SANSA 0.4.0

Features

• Spark
  • New: Numerical outlier detection(Beta status)
  • New: RDF Graph Kernels (Alpha)
  • Update: knowledge graph embedding support(pre Alpha)
  • Update: Decision trees (pre Alpha)
  • Update: Clustering(Beta status)
  • Update: Semantic similarity measures
  • Update: Vandalism Detection in WikiData(Beta status)(Beta status)

Dependency changes

• Apache Spark 2.3.1
• Apache Flink 1.5.0
• Apache Jena 3.7.0

SANSA-ML 0.3.0

Features

• Updated: Rule mining algorithm for RDF graphs based on AMIE+ further developed (still beta status)
• Updated: semantic similarity measures: They can be defined as a function of common and distinctive features among different entities. We have implemented the following measures:
  • Jaccard similarity,
  • Rodr ́ıguez and Egenhofer similarity
  • Tversky Ratio Model
  • Batet Similarity
• Updated: Clustering algorithms further extended and evaluated (Experimental)
  • Silvia Link Clustering
  • Border Flow (Extended for RDF)
  • Power Iteration Clustering (Extended for RDF)
  • Modularity Clustering
• New : Anomaly detection (beta status)
• New : Vandalism Detection (beta status)
• Knowledge graph embedding approaches integrated into the SANSA core: TransE (beta status), DistMult (beta status)
• In-Progress: Terminological Decision Trees for the classification of concepts

Dependency changes

• Scala 2.11.11
• Apache Spark 2.2.1
• Apache Flink 1.4.0
• Apache Jena 3.5.0

SANSA ML 0.2.0

Features

• Spark
  • Rule mining algorithm for RDF graphs based on AMIE+ further developed (still beta status)
  • Distributed Tensor Factorisation (very experimental, not fully integrated)
  • Several semantic similarity measures implemented (experimental)
  • Power Iteration Clustering with custom similarity measures
  • Border Flow Clustering
• Flink
  • RDF By Modularity Clustering algorithm (introduced by Newman- DOI: https://doi.org/10.1103/PhysRevE.69.066133) (beta status)

Dependency changes

• Spark 2.1.1
• Flink 1.3.0
• JenaAPI 3.1.1

SANSA ML 0.1.0

Features

• Spark
  • An RDF clustering algorithm based on an approach for undirected graphs maximizing a modularity function, which was first introduced by Newman (DOI: https://doi.org/10.1103/PhysRevE.69.066133) (beta status)
  • A rule mining algorithm for RDF graphs based on AMIE+ (beta status)
• Flink is not supported in this release