private void saveSemanticTypesInformation(
Worksheet worksheet, VWorkspace vWorkspace, Collection<SemanticType> semanticTypes)
throws JSONException {
JSONArray typesArray = new JSONArray();
// Add the vworksheet information
JSONObject vwIDJObj = new JSONObject();
vwIDJObj.put(ClientJsonKeys.name.name(), ParameterType.vWorksheetId.name());
vwIDJObj.put(ClientJsonKeys.type.name(), ParameterType.vWorksheetId.name());
vwIDJObj.put(ClientJsonKeys.value.name(), vWorksheetId);
typesArray.put(vwIDJObj);
// Add the check history information
JSONObject chIDJObj = new JSONObject();
chIDJObj.put(ClientJsonKeys.name.name(), ParameterType.checkHistory.name());
chIDJObj.put(ClientJsonKeys.type.name(), ParameterType.other.name());
chIDJObj.put(ClientJsonKeys.value.name(), false);
typesArray.put(chIDJObj);
for (SemanticType type : semanticTypes) {
// Add the hNode information
JSONObject hNodeJObj = new JSONObject();
hNodeJObj.put(ClientJsonKeys.name.name(), ParameterType.hNodeId.name());
hNodeJObj.put(ClientJsonKeys.type.name(), ParameterType.hNodeId.name());
hNodeJObj.put(ClientJsonKeys.value.name(), type.getHNodeId());
typesArray.put(hNodeJObj);
// Add the semantic type information
JSONObject typeJObj = new JSONObject();
typeJObj.put(ClientJsonKeys.name.name(), ClientJsonKeys.SemanticType.name());
typeJObj.put(ClientJsonKeys.type.name(), ParameterType.other.name());
typeJObj.put(ClientJsonKeys.value.name(), type.getJSONArrayRepresentation());
typesArray.put(typeJObj);
}
setInputParameterJson(typesArray.toString(4));
}
@Override
public void generateJson(String prefix, PrintWriter pw, VWorkspace vWorkspace) {
Workspace workspace = vWorkspace.getWorkspace();
alignment = AlignmentManager.Instance().getAlignment(workspace.getId(), worksheetId);
SemanticTypes types = worksheet.getSemanticTypes();
Map<String, ColumnNode> hNodeIdTocolumnNodeMap = createColumnNodeMap();
Map<String, SemanticTypeNode> hNodeIdToDomainNodeMap = createDomainNodeMap();
JSONStringer jsonStr = new JSONStringer();
try {
JSONWriter writer = jsonStr.object();
writer.key("worksheetId").value(worksheetId).key("updateType").value("SemanticTypesUpdate");
writer.key(JsonKeys.Types.name());
writer.array();
// Iterate through all the columns
for (HNodePath path : worksheet.getHeaders().getAllPaths()) {
HNode node = path.getLeaf();
String nodeId = node.getId();
writer.object();
// Check if a semantic type exists for the HNode
SemanticType type = types.getSemanticTypeForHNodeId(nodeId);
if (type != null && type.getConfidenceLevel() != SemanticType.ConfidenceLevel.Low) {
writer
.key(JsonKeys.HNodeId.name())
.value(type.getHNodeId())
.key(JsonKeys.SemanticTypesArray.name())
.array();
ColumnNode alignmentColumnNode = hNodeIdTocolumnNodeMap.get(type.getHNodeId());
SemanticTypeNode domainNode = hNodeIdToDomainNodeMap.get(type.getHNodeId());
if (alignmentColumnNode == null || domainNode == null) {
logger.error(
"Column node or domain node not found in alignment."
+ " (This should not happen conceptually!):"
+ type);
continue;
}
// Add the primary semantic type
writer
.object()
.key(JsonKeys.Origin.name())
.value(type.getOrigin().name())
.key(JsonKeys.ConfidenceLevel.name())
.value(type.getConfidenceLevel().name())
.key(JsonKeys.isPrimary.name())
.value(true);
// Add the RDF literal type to show in the text box
String rdfLiteralType =
alignmentColumnNode.getRdfLiteralType() == null
? ""
: alignmentColumnNode.getRdfLiteralType().getDisplayName();
String language =
alignmentColumnNode.getLanguage() == null ? "" : alignmentColumnNode.getLanguage();
writer.key(JsonKeys.rdfLiteralType.name()).value(rdfLiteralType);
writer.key(JsonKeys.language.name()).value(language);
// String domainDisplayLabel = (domainNode.getLabel().getPrefix() != null &&
// (!domainNode.getLabel().getPrefix().equals(""))) ?
// (domainNode.getLabel().getPrefix() + ":" + domainNode.getLocalId()) :
// domainNode.getLocalId();
if (!type.isClass()) {
writer
.key(JsonKeys.FullType.name())
.value(type.getType().getUri())
.key(JsonKeys.DisplayLabel.name())
.value(type.getType().getDisplayName())
.key(JsonKeys.DisplayRDFSLabel.name())
.value(type.getType().getRdfsLabel())
.key(JsonKeys.DisplayRDFSComment.name())
.value(type.getType().getRdfsComment())
.key(JsonKeys.DomainId.name())
.value(domainNode.getId())
.key(JsonKeys.DomainUri.name())
.value(domainNode.getUri())
.key(JsonKeys.DisplayDomainLabel.name())
.value(domainNode.getDisplayId())
.key(JsonKeys.DomainRDFSLabel.name())
.value(domainNode.getRdfsLabel())
.key(JsonKeys.DomainRDFSComment.name())
.value(domainNode.getRdfsComment());
} else {
writer
.key(JsonKeys.FullType.name())
.value(domainNode.getId())
.key(JsonKeys.DisplayLabel.name())
.value(domainNode.getDisplayId())
.key(JsonKeys.DisplayRDFSLabel.name())
.value(domainNode.getRdfsLabel())
.key(JsonKeys.DisplayRDFSComment.name())
.value(domainNode.getRdfsComment())
.key(JsonKeys.DomainId.name())
.value("")
.key(JsonKeys.DomainUri.name())
.value("")
.key(JsonKeys.DisplayDomainLabel.name())
.value("")
.key(JsonKeys.DomainRDFSLabel.name())
.value("")
.key(JsonKeys.DomainRDFSComment.name())
.value("");
}
// Mark the special properties
writer
.key(JsonKeys.isMetaProperty.name())
.value(isMetaProperty(type.getType(), alignmentColumnNode));
writer.endObject();
// Iterate through the synonym semantic types
SynonymSemanticTypes synTypes = types.getSynonymTypesForHNodeId(nodeId);
if (synTypes != null) {
for (SemanticType synType : synTypes.getSynonyms()) {
writer
.object()
.key(JsonKeys.HNodeId.name())
.value(synType.getHNodeId())
.key(JsonKeys.FullType.name())
.value(synType.getType().getUri())
.key(JsonKeys.Origin.name())
.value(synType.getOrigin().name())
.key(JsonKeys.ConfidenceLevel.name())
.value(synType.getConfidenceLevel().name())
.key(JsonKeys.DisplayLabel.name())
.value(synType.getType().getDisplayName())
.key(JsonKeys.DisplayRDFSLabel.name())
.value(synType.getType().getRdfsLabel())
.key(JsonKeys.DisplayRDFSComment.name())
.value(synType.getType().getRdfsComment())
.key(JsonKeys.isPrimary.name())
.value(false);
if (!synType.isClass()) {
writer
.key(JsonKeys.DomainUri.name())
.value(synType.getDomain().getUri())
.key(JsonKeys.DomainId.name())
.value("")
.key(JsonKeys.DisplayDomainLabel.name())
.value(synType.getDomain().getDisplayName())
.key(JsonKeys.DomainRDFSLabel.name())
.value(synType.getDomain().getRdfsLabel())
.key(JsonKeys.DomainRDFSComment.name())
.value(synType.getDomain().getRdfsComment());
} else {
writer
.key(JsonKeys.DomainId.name())
.value("")
.key(JsonKeys.DomainUri.name())
.value("")
.key(JsonKeys.DisplayDomainLabel.name())
.value("")
.key(JsonKeys.DomainRDFSLabel.name())
.value("")
.key(JsonKeys.DomainRDFSComment.name())
.value("");
}
writer.endObject();
}
}
writer.endArray();
} else {
writer.key(JsonKeys.HNodeId.name()).value(nodeId);
writer.key(JsonKeys.SemanticTypesArray.name()).array().endArray();
}
writer.endObject();
}
writer.endArray();
writer.endObject();
pw.print(writer.toString());
} catch (JSONException e) {
logger.error("Error occured while writing to JSON!", e);
}
}
public void run() {
long start = System.currentTimeMillis();
// Find the corresponding hNodePath. Used to find examples for training the CRF Model.
HNodePath currentColumnPath = null;
List<HNodePath> paths = worksheet.getHeaders().getAllPaths();
for (HNodePath path : paths) {
if (path.getLeaf().getId().equals(newType.getHNodeId())) {
currentColumnPath = path;
break;
}
}
Map<ColumnFeature, Collection<String>> columnFeatures =
new HashMap<ColumnFeature, Collection<String>>();
// Prepare the column name for training
String columnName = currentColumnPath.getLeaf().getColumnName();
Collection<String> columnNameList = new ArrayList<String>();
columnNameList.add(columnName);
columnFeatures.put(ColumnFeature.ColumnHeaderName, columnNameList);
// Train the model with the new type
ArrayList<String> trainingExamples =
SemanticTypeUtil.getTrainingExamples(worksheet, currentColumnPath);
boolean trainingResult = false;
String newTypeString =
(newType.getDomain() == null)
? newType.getType().getUri()
: newType.getDomain().getUri() + "|" + newType.getType().getUri();
trainingResult =
crfModelHandler.addOrUpdateLabel(newTypeString, trainingExamples, columnFeatures);
if (!trainingResult) {
logger.error("Error occured while training CRF Model.");
}
// logger.debug("Using type:" + newType.getDomain().getUri() + "|" +
// newType.getType().getUri());
// Add the new CRF column model for this column
ArrayList<String> labels = new ArrayList<String>();
ArrayList<Double> scores = new ArrayList<Double>();
trainingResult =
crfModelHandler.predictLabelForExamples(
trainingExamples, 4, labels, scores, null, columnFeatures);
if (!trainingResult) {
logger.error("Error occured while predicting labels");
}
CRFColumnModel newModel = new CRFColumnModel(labels, scores);
worksheet.getCrfModel().addColumnModel(newType.getHNodeId(), newModel);
long elapsedTimeMillis = System.currentTimeMillis() - start;
float elapsedTimeSec = elapsedTimeMillis / 1000F;
logger.info("Time required for training the semantic type: " + elapsedTimeSec);
// long t2 = System.currentTimeMillis();
// Identify the outliers for the column
// SemanticTypeUtil.identifyOutliers(worksheet, newTypeString,currentColumnPath,
// vWorkspace.getWorkspace().getTagsContainer()
// .getTag(TagName.Outlier), columnFeatures, crfModelHandler);
// long t3 = System.currentTimeMillis();
// logger.info("Identify outliers: "+ (t3-t2));
}
