public MLInstance(Instance instance, CodeBook codebook) {
messageID = instance.getMessageId();
instID = String.valueOf(instance.getId());
sentence = instance;
tokens = Arrays.asList(instance.getTokens());
posTags = Arrays.asList(sentence.getPOSTags());
// what is this for unlabeled
// check for null or empty list
classLabels = instance.getCodes();
// check if classLabels is null
// get codebook from docID
humanLabels = new ArrayList<String>();
String humanLabel;
if (classLabels != null) {
for (String label : classLabels) {
humanLabel = codebook.getLabel(label);
humanLabels.add(humanLabel);
}
}
// document level data
docSender = instance.getMessageFrom();
// silver instance information
if (instance.getInstanceType() == InstanceType.SILVER) {
isSilverCollection = true;
silverCode = instance.getSilverCode();
// check these labels, is one unverified?
silverVerificationLabel = instance.getVerificationType();
}
}
/*
* This constructor for use with gold documents from repository with tokens and labels
* (possibly null) sender
*/
public MLInstance(
String messageID,
String sentID,
Instance sentence,
List<String> toks,
List<String> labels,
List<String> humanLabels,
String docSender) {
this.messageID = messageID;
instID = sentID;
this.sentence = sentence;
tokens = toks;
// assume that the Instance class returns the tokens and pos tags in order
this.posTags = Arrays.asList(sentence.getPOSTags());
classLabels = labels;
this.humanLabels = humanLabels;
this.docSender = docSender;
isSilverCollection = false;
}
/*
* This constructor for use with silver instances from repository with tokens and verified codes
* (possibly null) sender
*/
public MLInstance(
String messageID,
String sentID,
Instance sentence,
List<String> toks,
String code,
VerificationType label,
String docSender) {
this.messageID = messageID;
instID = sentID;
this.sentence = sentence;
tokens = toks;
// assume that the Instance class returns the tokens and pos tags in order
this.posTags = Arrays.asList(sentence.getPOSTags());
silverVerificationLabel = label;
silverCode = code;
this.docSender = docSender;
// class label list is empty
classLabels = new ArrayList<String>();
isSilverCollection = true;
}
