Post by Admin on Jun 1, 2015 17:53:25 GMT
with keywords it depends on how there used. Normally a keyword is
associated to a word and is acted upon when it sees that word. If there
are 2 words in an input that activate a keyword then only one of the
keywords will be used. You can get round this by using the keyword
nnactivatecell to activate another cell. If using say the word activate
you would set up the cell something this:
cell word = activate
keyword = nnactivatecell
tags = <once><activate-cell:1234>
this will activate the cell when it finds activate in the input then it
will activate cell 1234
here the tag <once> is used to stop multiple cell activations. The cell
you active can be any cell in the database if you wanted to use another
keyword then setup cell 1234 something like this:
cell number = 1234
keyword = messagebox
tags = <keyword-ifa><activate-cell:1235>
field 8 = did it work?
you can then use the <activate-cell:> in this cell to keep going
I think if its starting to sound like Ro then this is the learning
routines kicking in. It will be the processes that use the neural
network chewing through the database looking for associations DT
Learning will help to refine these outputs. You could also try adjusting
the RQ values in settings.
To link an image to a word change the operation value of the word to 3
and put the image filename in the file field. The image needs to be in
the image folder e.g.
cell word = look
operation = 3
file = Image.jpg
cell word = see
operation = 3
file = Image.jpg
When the Ai starts it creates a copy of the database this is called
BrainDatabase.bak if the database goes weird then you can delete the
main database BrainDatabase.dat and rename BrainDatabase.bak to
BrainDatabase.dat then when the Ai restarts it will use the backup. You
need to close the Ai before renaming the files.
associated to a word and is acted upon when it sees that word. If there
are 2 words in an input that activate a keyword then only one of the
keywords will be used. You can get round this by using the keyword
nnactivatecell to activate another cell. If using say the word activate
you would set up the cell something this:
cell word = activate
keyword = nnactivatecell
tags = <once><activate-cell:1234>
this will activate the cell when it finds activate in the input then it
will activate cell 1234
here the tag <once> is used to stop multiple cell activations. The cell
you active can be any cell in the database if you wanted to use another
keyword then setup cell 1234 something like this:
cell number = 1234
keyword = messagebox
tags = <keyword-ifa><activate-cell:1235>
field 8 = did it work?
you can then use the <activate-cell:> in this cell to keep going
I think if its starting to sound like Ro then this is the learning
routines kicking in. It will be the processes that use the neural
network chewing through the database looking for associations DT
Learning will help to refine these outputs. You could also try adjusting
the RQ values in settings.
To link an image to a word change the operation value of the word to 3
and put the image filename in the file field. The image needs to be in
the image folder e.g.
cell word = look
operation = 3
file = Image.jpg
cell word = see
operation = 3
file = Image.jpg
When the Ai starts it creates a copy of the database this is called
BrainDatabase.bak if the database goes weird then you can delete the
main database BrainDatabase.dat and rename BrainDatabase.bak to
BrainDatabase.dat then when the Ai restarts it will use the backup. You
need to close the Ai before renaming the files.