In-Close4 CONCEPT MINING AND TREE BUILDING PROGRAM - 64 bit version - Windows, Linux, MacOS and Cross-platform Versions now included.
Download free and open source at: https://sourceforge.net/projects/inclose/
Copyright (c) 2017 Simon Andrews, Sheffield Hallam University, UK, Conceptual Structures Research Group, s.andrews@shu.ac.uk
In-Close4 mines .cxt (Formal Context) and FIMI .dat files for Formal Concepts (http://www.upriss.org.uk/fca/fca.html)
In-Close4 builds and outputs concept trees in JSON format for the D3.js Collapsible Tree Format
To build and view a tree, output the tree using In-Close4 with output option 4. Open the tree visualiser in Chrome web browser (or other browser, but not IE) at: http://homepages.shu.ac.uk/~aceslh/fca/fcaTree.html
Then upload the output concepts.json file.
*NEW FEATURES*
In-Close4 outputs a concept tree in JSON format that can be visualised at: http://homepages.shu.ac.uk/~aceslh/fca/fcaTree.html
In-Close4 makes use of 64 bit architecture to be faster (64-bitwise operation) and to cope with larger contexts/number of concepts (more memory available).
In-Close4 can now input FIMI .dat files.
Concepts are now output in JSON format or csv.
BUG-FIX: issue with slow pre-processing (sorting) when file contains many repeated objects (see Kirchberg et al 2014: Formal Concept
Discovery in Semantic Web Data) now fixed by using a better version of Quick Sort.
There are also the following options:
-Output the cxt file after sorting (sorts columns in order of density and sorts rows as if they were binary numbers). Note this option also means that FIMI .dat files can be converted into .cxt format.
-Specify minimum size of intent (min no of attributes in a concept).
-Specify minimum size of extent (min no of objects in a concept).
-Output concepts to a file called concepts.json in JSON format. Three options:
1) concepts as lists of index numbers of objects and attributes.
2) concepts as lists of names of objects and attributes.
3) concepts with arrays for many-valued attributes, e.g. if the objects share the attributes "location-Sheffield" and
"location-London" the corresponding JSON construction will be "location" : ["Sheffield","London"].
4) output a Concept Tree (concepts.JSON) where child concepts are within parent concepts. Can be visualised at http://homepages.shu.ac.uk/~aceslh/fca/fcaTree.html
5) output a Concept Tree (concepts.JSON) where a concept includes the index number of its parent.
6) output concepts as a csv file.
In-Close4 outputs the count of concepts (that satisfy the minimum support specified).
In-Close4 outputs the time taken for pre-processing (sorting) and the time taken for concept mining.
In-Close4 outputs the breakdown of concept count by size in two files:
-noConsByBsize.txt: List of <number of attributes> - <number of concepts having that number of attributes>
-noConsByAsize.txt: List of <number of objects> - <number of concepts having that number of objects>
In-Close4 outputs the reduced context (cxt) file that results from the minimum sizes of intent and extent specified by the user, also removing objects (rows) and attributes (columns) that are not involved in concepts that satisfy the minimum support.
Thus it is possible to produce a context that only contains the 'largest' concepts, making it possible to visualise the resulting lattice using tools such as 'Concept Explorer' (http://sourceforge.net/projects/conexp/).
In-Close4 is released under the Free and Open Source Software, MIT License. However, the author politely requests that users of the Software please inform him (by email: s.andrews@shu.ac.uk) what use they have made or are making of the Software. This is purely to help the author record what impact the Software has had, whether in private use, commercial application, internal/non-commercial application, research or public benefit. Thank you.
