Formal concept analysis

LATTICE

builds a concept lattice. ADD_EXTENT and ADD_INTENT algorithms build the complete concept lattice. ICEBERG builds a lattice with only the top-most concepts of the concept lattice. A bottom concept is added that groups remaining attributes (not introduced by a frequent concept extent) and transforms the semi-lattice into a lattice. Euclidian division is used in FCA4J: e.g. In Iceberg50, the concepts have an extent whose cardinality is >= total object count * 50 /100, where / is Euclidian division

Usage

java -jar fca4j-cli.jar LATTICE <input> [<output-file>] [options]

Options

Option	Description
`-a <ALGO>`	choose AUTO to select dynamically the best algorithm. Supported algorithms are: `AUTO (default)` `ADD_EXTENT` `ADD_INTENT` `ICEBERG`
`-p <PERCENT>`	for ICEBERG: percentage (of extent) to keep the top-most concepts
`-i <INPUT-FORMAT>`	supported formats are: `CXT (Burmeisters ConImp)` `SLF (HTK Standard Latice Format)` `XML (Galicia v3)` `CEX (ConExp)` `CSV (Comma separated values)`
`-o <OUTPUT-FORMAT>`	supported formats are: `XML (default)` `JSON`
`-g <DOTFILE>`	.DOT output file for GraphViz
`-d <DISPLAY-MODE>`	display format of the concepts for GraphViz. Available formats are: `FULL` `SIMPLIFIED (default)` `MINIMAL`
`-sta`	compute concept stability for dot file
`-m <IMPL>`	supported implementations are `BITSET (default)` `ROARING_BITMAP` `SPARSE_BITSET` `HASHSET` `TREESET` `INT_ARRAY` `ARRAYLIST` `BOOL_ARRAY`
`-cd <PATH>`	folder to generate a datalog file by concept for results
`-cdu <DLGPFILE>`	generate a unique datalog file with all concepts descriptions
`-nds`	limit datalog concept descriptions to immediate siblings (it reduces production of negative constraints)
`-s <SEPARATOR>`	separator (CSV format only): `COMMA (default)` `SEMICOLON` `TAB`
`-timeout <SECONDS>`	set timeout for algorithm execution, in seconds
`-v, --verbose`	print a final report of the algorithm execution