FCA tools, algorithms and datasets

This is a collection of tools, algorithms and datasets, we used in our works on Formal Concept Analysis. Our intention is to make the results of our research as transparent as possible.

Software

• Performance comparison of CbO-based algorithms
  Language: Data and Javascript  Sources: cbo-graphs.tgz  Online: View
• FCA algorithms and tools
  Language: C++  Sources: fca-apps-jg.tgz   License: GPLv3
• LinCbO in the framework made by Bazhanov & Obiedkov (https://github.com/yazevnul/fcai )
  Language: C++  Sources: LinCbO.zip   License: MIT
• Context binarizer
  Language: Java  Sources: ContextBinarizer.zip   License: MIT
• GLinCbO: computation of the Duquenne-Guigues basis of fuzzy formal contexts (graded LinCbO, graded NextClosure baseline, correctness suite, scripts and measured results of the experiments)
  Language: C++  Sources: GLinCbO.zip   License: MIT

Datasets

All provided datasets are in Burmeister format , were binarized from the original dataset given in the last column.

Dataset	#objects	#attributes	#incidences	#intents	#ps.intents	Source
inter10crx	653	139	40,170	10,199,818	20,108
inter10shuttle	43,500	178	3,567,907	38,199,148	936
inter3magic	19,020	52	399,432	1,006,553	4181
inter4magic	19,020	72	589,638	24,826,749	21,058
inter5bike_day	731	93	24,650	3023,326	20,425
inter5crx	653	79	20,543	348,428	3427
inter5shuttle	43,500	88	1,609,510	333,783	346
inter6shuttle	43,500	106	2,002,790	381,636	566
nom10bike_day	731	100	9293	52,697	29,773
nom10crx	653	85	8774	51,078	6240
nom10magic	19,020	102	209,220	583,386	154,090
nom10shuttle	43,500	97	435,000	2931	810
nom15magic	19,020	152	209,220	1,149,717	397,224
nom20magic	19,020	202	209,220	1,376,212	654,028
nom5bike_day	731	65	9293	61,853	16,296
nom5bike_hour	17,379	90	238,292	1,868,205	320,679
nom5crx	653	55	8774	29,697	2162
nom5keg	65,554	144	1,834,566	13,262,627	42,992
nom5shuttle	43,500	52	435,000	1461	319
ord10bike_day	731	93	28,333	664,713	11,795
ord10crx	653	79	37,005	1,547,971	2906
ord10shuttle	43,500	88	1,849,216	97,357	279
ord5bike_day	731	58	14,929	81,277	5202
ord5bike_hour	17,379	83	457,578	2,174,964	99,691
ord5crx	653	49	19,440	139,752	973
ord5magic	19,020	42	535,090	821,796	1267
ord5shuttle	43,500	43	868,894	4068	119
ord6magic	19,020	52	662,177	2,745,877	2735

All above datasets can be downloaded in archive.

Datasets used by Bazhanov & Obiedkov can be found on their GitHub repository: https://github.com/yazevnul/fcai/tree/refactoring/test_data/contexts

Fuzzy datasets

Fuzzy formal contexts used in the experimental evaluation of GLinCbO, obtained by the graded scaling of numeric attributes to a chain of k truth degrees (k = 3, 5, 7; equidistant or quantile cutpoints; each also in an object-clarified version), 132 contexts in total. The archive contains the contexts, the original data, the column specifications, and the scaling tool (Python): fuzzydata.zip . The number of objects after clarification is given for equidistant cutpoints.

Dataset	#objects	#attributes	#objects clarified, k=3	k=5	k=7	Source
iris	150	4	21	46	77
haberman	306	3	18	54	90
seeds	210	7	62	149	177
ecoli	336	7	75	176	248
glass	214	9	61	106	137
wine	178	13	169	178	178
crx	653	46	501	569	614
bike_day	731	39	665	717	723
bike_hour	17,379	41	6,961	11,667	14,032
magic	19,020	10	712	2,680	5,755
kegg	64,608	28	928	4,281	8,999

The data come from the UCI Machine Learning Repository (M. Kelly, R. Longjohn, K. Nottingham: The UCI Machine Learning Repository, https://archive.ics.uci.edu) and are licensed under CC BY 4.0.

Accepted publications

• Radek Janostik, Jan Konecny, Petr Krajca: Pruning techniques in LinCbO for the computation of the Duquenne-Guigues basis.
  Inf. Sci. 616: 182-203 (2022)
  DOI 10.1016/j.ins.2022.10.057 , ISSN: 00200255
• Petr Krajca: On Pruning Techniques in Map-Reduce Style CbO Algorithms.
  Annals of Mathematics and Artificial Intelligence
  
• Radek Janostik, Jan Konecny: LCM from FCA Point of View: A CbO-style Algorithm with Speed-up Features.
  Int. J. Approx. Reason. 142: 64-80 (2022)
  DOI 10.1016/j.ijar.2021.11.005 , ISSN: 0888613X
• Radek Janostik, Jan Konecny: LinCbO: Fast algorithm for computation of the Duquenne-Guigues basis.
  Inf. Sci. 572: 223-240 (2021)
  DOI 10.1016/j.ins.2021.04.104 , ISSN: 00200255
• Jan Konecny, Petr Krajca: Systematic categorization and evaluation of CbO-based algorithms in FCA.
  Inf. Sci. 575: 265-288 (2021)
  DOI 10.1016/j.ins.2021.06.024 , ISSN: 00200255
• Pruning Techniques in LinCbO for Computation of the Duquenne-Guigues Basis.
  ICFCA 2021: 91-106
  DOI 10.1007/978-3-030-77867-5_6
• Petr Krajca: Improving the Performance of Lindig-Style Algorithms with Empty Intersections.
  ICCS 2021: 91-104
  DOI 10.1007/10.1007/978-3-030-86982-3_7
• Petr Krajca, Martin Trnecka: Reducing Negative Impact of Noise in Boolean Matrix Factorization with Association Rules.
  IDA 2021: 365-375
  DOI 10.1007/978-3-030-74251-5_29
• Radek Janostik, Jan Konecny, Petr Krajca: Interface between Logical Analysis of Data and Formal Concept Analysis.
  Eur. J. Oper. Res. 284(2): 792-800 (2020)
  DOI 10.1016/j.ejor.2020.01.015 , ISSN 0377-2217
• Jan Konecny: Attribute implications in L-concept analysis with positive and negative attributes: Validity and properties of models.
  Int. J. Approx. Reason. 120: 203-215 (2020)
  DOI 10.1016/j.ijar.2020.02.009
• Radek Janostik, Jan Konecny: General framework for consistencies in decision contexts.
  Inf. Sci. 530: 180-200 (2020)
  DOI 10.1016/j.ins.2020.02.045 , ISSN: 00200255
• Radek Janostik, Jan Konecny, Petr Krajca: LCM is Well Implemented CbO: Study of LCM from FCA Point of View.
  CLA 2020: 47-58
  
• Jan Konecny, Petr Krajca: Pruning in Map-Reduce Style CbO Algorithms.
  ICCS 2020: 103-116
  DOI 10.1007/978-3-030-57855-8_8
• Jan Konecny, Petr Krajca: On attribute reduction in concept lattices: The polynomial time discernibility matrix-based method becomes the CR-method.
  Inf. Sci. 491: 48-62 (2019)
  DOI 10.1016/j.ins.2019.03.038 , ISSN 0020-0255
• Petr Krajca, Martin Trnecka: Parallelization of the GreConD Algorithm for Boolean Matrix Factorization.
  ICFCA 2019: 208-222
  DOI 10.1007/978-3-030-21462-3_14

Submitted publications

• Jan Konecny: LinCbO for data with grades: computing the Duquenne-Guigues basis of a fuzzy formal context.
  Information Sciences
  

Supported by the grant JG 2019 of Palacký University Olomouc, No. JG_2019_008.