

Codes in this directory build two kinds of programs:

1) The first kind, implemented in mmg.c, calculates a k-mmg based on an
   input sequence set.

2) The second kind, implemented in eval_pat.c, evaluates a k-mmg w.r.t a
   positive and a negative input set.

Programs are built differently depending on the object file (in directory
"op/reg*") they are linked with. The file name describes the functionality
of the operator.

If you are lost, first read the following papers on the MMG algorithm (in
chronological order --- most likely, you will need to read only (2)).
If you are unable to find any of these papers, please do not hesitate to
contact us for them.

1. Finding minimal generalizations for unions of pattern
      languages and its application to inductive inference from positive data
   H. Arimura and T. Shinohara and S. Otsuki 
   STACS'94

2. Protein motif discovery from positive examples
       by Minimal Multiple Generalization over regular patterns
   H. Arimura and R. Fujino and T. Shinohara and S. Arikawa
   Genome Informatics Workshop 1994

3. Finding Minimal Multiple Generalization over Regular Patterns with
      Alphabet Indexing
   M. Yamaguchi and S. Shimozono and T. Shinohara
   Seventh Workshop on Genome Informatics 1996

4. Knowledge Discovery in Biosequences Using Sort Regular Patterns
   T. Takae and T. Kasai and H. Arimura and T. Shinohara
   Workshop on Applied Learning Theory 1998

5. Measuring Over-generalization in the Minimal Multiple Generalizations
      of Biosequences
   Y. K. Ng, H. Ono and T. Shinohara
   Discovery Science 2005

6. Finding Consensus Patterns in Very Scarce Biosequence Samples from Their
   Minimal Multiple Generalizations
   Y. K. Ng and T. Shinohara
   PAKDD 2006
 
If you need help, contact shino@ai.kyutech.ac.jp
If you found bugs, contact kalngyk@daisy.ai.kyutech.ac.jp
