Selected software and data resources developed by DSXAI

Back to the DSXAI team page

Actively developed projects

• Cleverminer.org: system for GUHA association rule mining in Python
  • Lead developer: Petr Máša
    
  • PyPI repository: https://pypi.org/project/cleverminer
  • Web: https://www.cleverminer.org/
  • See also papers:
    • A novel algorithm weighting different importance of classes in enhanced association rules
    • A novel algorithm for mining couples of enhanced association rules based on the number of output couples and its application
    • Enhanced association rules and Python
• EasyMiner: web interface for association rule mining
  • Lead developer: Stanislav Vojíř
  • Web demo: https://easyminer.eu
  • See also paper: EasyMiner.eu: Web framework for interpretable machine learning based on rules and frequent itemsets
• RDFRules: toolbox for association rule mining from graph and relational data
  • Lead developer: Václav Zeman
    
  • Repository: https://github.com/propi/rdfrules
  • See also paper: RDFRules: Making RDF Rule Mining Easier and Even More Efficient
• ActionRules package: recommendation of „best next action“ through Python and GPU optimization
  • Lead developer: Lukáš Sýkora
    
  • PyPI repository: https://pypi.org/project/action-rules/
• qCBA: R package for post-optimization of numerical information in rule models; recovers info lost through numerical attribute binning
  • Lead developer: Tomáš Kliegr
    
  • Repository: https://cran.r-project.org/web/packages/qCBA
• arc: R package for classification on the basis of association rules
  • Lead developer: Tomáš Kliegr
    
  • Repository: https://cran.r-project.org/web/packages/arc/index.html
• UTA: web interface for the UTA method of multi-criterion decision making (supporting non-monotonous utility functions)
  • Lead developer: Tomáš Kliegr
    
  • Demo: https://nlp.vse.cz/uta/
  • See also paper: UTA – NM: Explaining Stated Preferences with Additive Non-Monotonic Utility Functions
  • Demo: https://nlp.vse.cz/uta/
• Suite of tools for using LLMs in bias mitigation and rule learning
  • Lead developer (supervising students’ projects): Tomáš Kliegr
    
  • Traceable LLM-based validation of statements in knowledge graphs,
    • Github: https://github.com/KIZI/LLM-based-validation
  • LLM-based feature generation from text for interpretable machine learning,
    • Github: https://github.com/KIZI/llm-features
  • Meaningless is better: hashing bias-inducing words in LLM prompts improves performance in logical reasoning and statistical learning,
    • Github: https://github.com/KIZI/Hashing_LLM_Debiasing
• Extended AIQ test: tools and datasets associated with the Algorithmic Intelligence Quotient
  • Lead developer: Ondřej Vadinský
    
  • Algorithmic Intelligence Quotient test (Legg and Veness, 2013) is a test of machine intelligence that approximates the Universal Intelligence of an agent (Legg and Hutter, 2007). The extended version addresses several issues of the original test
  • Repository: https://github.com/xvado00
  • Actual extended test: https://github.com/xvado00/AIQ/releases
  • Tool for semantic analysis of environment programs used by the test: https://github.com/xvado00/SemAnEP-tool/releases 
• Fact-checking report summarization tool based on LLM
  • Lead developer: Peter Vajdečka
    
  • https://gitlab.com/FIS-VSE/projekty/cimple/summarization_app
• PatOMat2: tool for pattern-based ontology transformation
  • Lead developer: Ondřej Zamazal
    
  • Demo: https://owl.vse.cz/patomat2/
• PURO Modeler: graphical drafting of ontology skeletons
  • Lead developer: Marek Dudáš
    
  • Repository: https://github.com/marek-dudas/PURO-Modeler,
  • Demo: https://puromodeler.vse.cz/
• OReCaP:Ontology Recommendation via Categorization Power
  • Lead developer: Viet Bach Nguyen
    
  • Repository: https://github.com/nvbach91/orecap,
  • Demo: https://fcp.vse.cz/orecap/
• Code List Analyzer: extraction of embedded codelists from KGs / ontologies
  • Lead developer: Viet Bach Nguyen
    
  • Online: https://fcp.vse.cz/iga-hybrid/

Other functional tools (with development currently suspended)

• LISp-Miner: desktop data mining tool realizing several analytical GUHA procedures of mechanizing hypothesis formation described in the related monograph.  The procedures deal with enhanced association rules, couples of enhanced association rules, conditional histograms, couples of conditional histograms, pairs of categorical attributes, couples of pairs of categorical attributes, and with enhanced action rules (see also here and here).
  • Also features a grid-based mining solution (LM-Grid) and generation of (nugget-containing) artificial datasets suitable for data mining education (ReverseMiner).
  • Contact: Jan Rauch
  • Web: http://lispminer.vse.cz
• InBeat: explainable recommender system using rules and linked data
  • Contact: Tomáš Kliegr
  • In collaboration with the Web Intelligence group at the CTU.
  • Web: https://inbeat.eu/
• EntityClassifier.eu: identifies entities in text and assigns them DBpedia types
  • Contact: Tomáš Kliegr
  • In collaboration with the Web Intelligence group at the CTU
  • Exploits the Linked Hypernym Dataset (LHD) containing millions of entity-type assignments
  • Web: http://entityclassifier.eu/ (tool) and http://ner.vse.cz/datasets/linkedhypernyms/ (dataset).
• PatOMat: suite of tools for pattern-based transformation of ontologies and vocabularies
  • Contact: Ondřej Zamazal
  • Superseded by PatOMat2
  • Includes a graphical plugin for Protégé GUIPOT. Best demo award at EKAW’12.
  • Web: http://patomat.vse.cz/.
• DB-quiz: Quiz game powered by knowledge graphs
  
  • Contact: Václav Zeman
  • Two players strive to connect the three sides of the triangular board by occupying hexagonal fields. A field is occupied through correctly answering a question generated from a fact base: Czech or English DBpedia, or a custom one. The challenging aspects of the game design are question despoilerification and difficulty rating.
  • Web: http://mynarz.net/db-quiz/
• SPARQlab: online exercise book for learning the SPARQL query language.
  • Contact: Vojtěch Svátek
    
  • Currently non-functional due to compatibility issues.
  • Both the application and the dataset documentation is currently in Czech only.
• LODSight: linked dataset summarization tool with special focus on employed vocabularies
  • Contact: Marek Dudáš
    
  • Currently non-functional
  • Aims to give insights into a dataset and vocabularies at the same time, by distinguishing between different vocabularies and providing examples of concrete values from the dataset as annotations to the summary graph.
• OOSP – Online Ontology Set Picker: tool for rapidly building tailored collections of ontologies based on their various metrics
  • Contact: Ondřej Zamazal
    
  • Currently non-functional
  • Collections can then be used as benchmarks for testing ontology management tools. The source collections used are LOV and Bioportal.
• NEST: expert system shell with multiple methods of uncertainty processing and reasoning algorithms.
  • Contact: Petr Berka
    
  • Web-based version: http://nest.vladalas.info/
  • Desktop version currently unavailable

Datasets, knowledge graphs and ontologies

• Forthcoming…