Excluding false values with association rule mining in Weka

Question

I am using Weka 3.6 to do Association Rule mining. In our data set, each transaction is a word, and each letter in the word is an item. The rules that we are mining would be in the format of {a set of letters} -> {another set of letters}.

So far, I have formatted six transactions by representing the existence of a letter with '1' and the absence of a letter with '0', but this is giving me some unwanted rules. Specifically, I am getting rules where the absence of a letter implies the absence of another letter, e.g. C='0' ==> U='0'.

How do I filter out rules that represent the absence of items?

---

For what it's worth, here is my .arff file:

@relation transactions
@attribute A {'1','0'}
@attribute C {'1','0'}
@attribute D {'1','0'}
@attribute E {'1','0'}
@attribute I {'1','0'}
@attribute K {'1','0'}
@attribute M {'1','0'}
@attribute N {'1','0'}
@attribute O {'1','0'}
@attribute U {'1','0'}
@attribute Y {'1','0'}
@data
'0','0','0','1','0','1','1','1','1','0','1' %monkey
'0','0','1','1','0','1','0','1','1','0','1' %donkey
'1','0','0','1','0','1','1','0','0','0','0' %make
'0','1','0','0','0','1','1','0','0','1','1' %mucky
'0','1','0','1','1','1','0','1','1','0','0' %conkie
'1','0','0','0','0','1','1','1','1','0','1' %mankoy

Answer 1

you can try this:

   @RELATION ITEMS
   @ATTRIBUTE itemM {?,1}
   @ATTRIBUTE itemO {?,1}
   @ATTRIBUTE itemN {?,1}
   @ATTRIBUTE itemK {?,1}
   @ATTRIBUTE itemE {?,1}
   @ATTRIBUTE itemY {?,1}
   @ATTRIBUTE itemU {?,1}
   @ATTRIBUTE itemD {?,1}
   @ATTRIBUTE itemA {?,1}
   @ATTRIBUTE itemC {?,1}
   @ATTRIBUTE itemI {?,1}

   @DATA

1,1,1,1,1,1,?,?,?,?,?
?,1,1,1,1,1,?,1,?,?,?
1,?,?,1,1,?,?,?,1,?,?
1,?,?,1,?,1,1,?,?,1,?
?,1,1,1,1,?,?,?,?,1,1
1,1,1,1,?,1,?,?,1,?,? 

or it works too:

   @RELATION ITEMS
   @ATTRIBUTE itemM {M} 
   @ATTRIBUTE itemO {O}
   @ATTRIBUTE itemN {N}
   @ATTRIBUTE itemK {K}
   @ATTRIBUTE itemE {E}
   @ATTRIBUTE itemY {Y}
   @ATTRIBUTE itemU {U}
   @ATTRIBUTE itemD {D}
   @ATTRIBUTE itemA {A}
   @ATTRIBUTE itemC {C}
   @ATTRIBUTE itemI {I}

   @DATA

M,O,N,K,E,Y,?,?,?,?,?
?,O,N,K,E,Y,?,D,?,?,?
M,?,?,K,E,?,?,?,A,?,?
M,?,?,K,?,Y,U,?,?,C,?
?,O,N,K,E,?,?,?,?,C,I
M,O,N,K,?,Y,?,?,A,?,? 

Answer 2

I'm not sure if there is a more formal way of doing this, but replacing the '0' values with ? works for me.

Using the following .arff file works for me:

@relation transactions
@attribute A {'1','0'}
@attribute C {'1','0'}
@attribute D {'1','0'}
@attribute E {'1','0'}
@attribute I {'1','0'}
@attribute K {'1','0'}
@attribute M {'1','0'}
@attribute N {'1','0'}
@attribute O {'1','0'}
@attribute U {'1','0'}
@attribute Y {'1','0'}
@data
?,?,?,'1',?,'1','1','1','1',?,'1' %monkey
?,?,'1','1',?,'1',?,'1','1',?,'1' %donkey
'1',?,?,'1',?,'1','1',?,?,?,? %make
?,'1',?,?,?,'1','1',?,?,'1','1' %mucky
?,'1',?,'1','1','1',?,'1','1',?,? %conkie
'1',?,?,?,?,'1','1','1','1',?,'1' %mankoy