| Description
|
Given a training set encoded as vectors of cue (or feature) occurrences in weka format, this web service computes P(cuei|class): the probability of seeing each cue as a member or non-member of the class using MLE approach (counts frequencies of appearance of each cue in each class).
Inputs:
- weka_signatures: classified instances encoded as cue vectors. Each slot of the vector contains the number of times each feature has been observed for that instance. Also, we add special slots: total number of occurrences in the first slot and correct class (1 or 0) and lemma (or any identifier of the instance) in the two last slots. The vectors should be encoded in a weka file, in UTF.8. The cue counts must be encoded as integers, this is, no relative frequency needs to be given but the number of times each cue has been seen. For example, for the class of eventive nouns in English we would have some examples of eventive nouns and some of non-eventive nouns:
@relation eventive.arff'
@attribute [total_occurences] numeric
@attribute [cue_1] numeric
@attribute [cue_2] numeric
…
@attribute [cue_n] numeric
@attribute [eventive] {0,1}
@attribute [lemma] string
@data
5,0,0,…,3,0,visa
386,0,1,…,162,0,characteristic
23,1,0,… ,0,1,ceremony
270,0,2,…,0,1,assembly
Outputs:
The output is a comma separated file with the frequencies each cue has been observed with members and non-members of the class. Also, information about the number of tokens in each class is given. Example:
#cue;data size class;data p class;data size no class;data p no class; cue_1;1301732;0.000883438372876;1516522;0.00137419701132; cue_2;1301732;0.000520076329075;1516522;0.000600716639785;
...
cue_n;1301732;0.0243222107162;1516522;0.0177992801951;
|
IULA relevant information for this Service:
