@Override
public UpdateContainer doIt(Workspace workspace) {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
SuperSelection selection = getSuperSelection(worksheet);
String worksheetName = worksheet.getTitle();
try {
// preparing model file name
final String modelFileName =
workspace.getCommandPreferencesId() + worksheetId + "-" + worksheetName + "-model.ttl";
final String modelFileLocalPath =
ServletContextParameterMap.getParameterValue(ContextParameter.R2RML_PUBLISH_DIR)
+ modelFileName;
File f = new File(modelFileLocalPath);
// preparing the graphUri where the model is published in the triple store
String graphName =
worksheet
.getMetadataContainer()
.getWorksheetProperties()
.getPropertyValue(Property.graphName);
if (graphName == null || graphName.isEmpty()) {
SimpleDateFormat sdf = new SimpleDateFormat("dd-MMM-yyyy-kkmmssS");
String ts = sdf.format(Calendar.getInstance().getTime());
graphName =
"http://localhost/"
+ workspace.getCommandPreferencesId()
+ "/"
+ worksheetId
+ "/model/"
+ ts;
worksheet
.getMetadataContainer()
.getWorksheetProperties()
.setPropertyValue(Property.graphName, graphName);
}
// If the model is not published, publish it!
if (!f.exists() || !f.isFile()) {
GenerateR2RMLModelCommandFactory factory = new GenerateR2RMLModelCommandFactory();
GenerateR2RMLModelCommand cmd =
(GenerateR2RMLModelCommand)
factory.createCommand(
workspace,
worksheetId,
TripleStoreUtil.defaultModelsRepoUrl,
graphName,
selection.getName());
cmd.doIt(workspace);
} else {
// if the model was published 30 min ago, publish it again, just to be sure
long diff = Calendar.getInstance().getTimeInMillis() - f.lastModified();
if ((diff / 1000L / 60L) > 30) {
f.delete();
GenerateR2RMLModelCommandFactory factory = new GenerateR2RMLModelCommandFactory();
GenerateR2RMLModelCommand cmd =
(GenerateR2RMLModelCommand)
factory.createCommand(
workspace,
worksheetId,
TripleStoreUtil.defaultModelsRepoUrl,
graphName,
selection.getName());
cmd.doIt(workspace);
}
}
// TripleStoreUtil tUtil = new TripleStoreUtil();
StringBuffer query =
new StringBuffer(
"prefix rr: <http://www.w3.org/ns/r2rml#> prefix km-dev: <http://isi.edu/integration/karma/dev#> ");
/* ****** this is the query for the list of columns.
PREFIX km-dev: <http://isi.edu/integration/karma/dev#>
PREFIX rr: <http://www.w3.org/ns/r2rml#>
select distinct ?class where {
{
?x1 rr:subjectMap/km-dev:alignmentNodeId "------- The full url of the column/class --------".
?x1 rr:predicateObjectMap/rr:objectMap/rr:column ?column .
?x1 rr:subjectMap/rr:predicate ?class .
}
UNION
{
?x1 rr:subjectMap/km-dev:alignmentNodeId "------- The full url of the column/class --------".
?x1 (rr:predicateObjectMap/rr:objectMap/rr:parentTriplesMap)* ?x2 .
?x2 rr:predicateObjectMap/rr:objectMap/rr:column ?column .
?x2 rr:predicateObjectMap/rr:predicate ?class .
}
}
* */
query.append("select distinct ?class ?column where { ");
if (graphName != null && !graphName.trim().isEmpty()) {
query.append(" graph <" + graphName + "> { ");
}
query
.append("{ ?x1 rr:subjectMap/km-dev:alignmentNodeId \"")
.append(this.nodeId)
.append(
"\" . ?x1 rr:predicateObjectMap/rr:objectMap/rr:column ?column . ?x1 rr:subjectMap/rr:predicate ?class .")
.append(" } UNION { ")
.append("?x1 rr:subjectMap/km-dev:alignmentNodeId \"")
.append(this.nodeId)
.append("\" . ?x1 (rr:predicateObjectMap/rr:objectMap/rr:parentTriplesMap)* ?x2 .")
.append(" ?x2 rr:predicateObjectMap ?x3 . ")
.append(" ?x3 rr:objectMap/rr:column ?column . ?x3 rr:predicate ?class .")
.append(" } }");
if (graphName != null && !graphName.trim().isEmpty()) {
query.append(" } ");
}
logger.info("Query: " + query.toString());
String sData =
TripleStoreUtil.invokeSparqlQuery(
query.toString(), TripleStoreUtil.defaultModelsRepoUrl, "application/json", null);
if (sData == null | sData.isEmpty()) {
logger.error("Empty response object from query : " + query);
}
HashMap<String, String> cols = new HashMap<String, String>();
try {
JSONObject obj1 = new JSONObject(sData);
JSONArray arr = obj1.getJSONObject("results").getJSONArray("bindings");
for (int i = 0; i < arr.length(); i++) {
String colName = arr.getJSONObject(i).getJSONObject("column").getString("value");
String colValue = arr.getJSONObject(i).getJSONObject("class").getString("value");
if (cols.containsKey(colName)) {
logger.error("Duplicate Column <-> property mapping. " + colName + " <=> " + colValue);
} else {
cols.put(colName, colValue);
}
}
} catch (Exception e2) {
logger.error("Error in parsing json response", e2);
}
logger.info("Total Columns fetched : " + cols.size());
final HashMap<String, String> columns = cols;
return new UpdateContainer(
new AbstractUpdate() {
@Override
public void generateJson(String prefix, PrintWriter pw, VWorkspace vWorkspace) {
JSONObject obj = new JSONObject();
try {
Iterator<String> itr = columns.keySet().iterator();
JSONArray colList = new JSONArray();
while (itr.hasNext()) {
JSONObject o = new JSONObject();
String k = itr.next();
o.put("name", k);
o.put("url", columns.get(k));
colList.put(o);
}
obj.put("updateType", "FetchColumnUpdate");
obj.put("columns", colList);
obj.put("rootId", nodeId);
pw.println(obj.toString());
} catch (JSONException e) {
logger.error("Error occurred while fetching worksheet properties!", e);
}
}
});
} catch (Exception e) {
String msg = "Error occured while fetching columns!";
logger.error(msg, e);
return new UpdateContainer(new ErrorUpdate(msg));
}
}
public static void test() throws Exception {
ServletContextParameterMap contextParameters =
ContextParametersRegistry.getInstance().getDefault();
ModelingConfiguration modelingConfiguration =
ModelingConfigurationRegistry.getInstance()
.getModelingConfiguration(contextParameters.getId());
// String inputPath = Params.INPUT_DIR;
String outputPath = Params.OUTPUT_DIR;
String graphPath = Params.GRAPHS_DIR;
// List<SemanticModel> semanticModels = ModelReader.importSemanticModels(inputPath);
List<SemanticModel> semanticModels =
ModelReader.importSemanticModelsFromJsonFiles(Params.MODEL_DIR, Params.MODEL_MAIN_FILE_EXT);
// ModelEvaluation me2 = semanticModels.get(20).evaluate(semanticModels.get(20));
// System.out.println(me2.getPrecision() + "--" + me2.getRecall());
// if (true)
// return;
List<SemanticModel> trainingData = new ArrayList<SemanticModel>();
OntologyManager ontologyManager = new OntologyManager(contextParameters.getId());
File ff = new File(Params.ONTOLOGY_DIR);
File[] files = ff.listFiles();
for (File f : files) {
ontologyManager.doImport(f, "UTF-8");
}
ontologyManager.updateCache();
// getStatistics1(semanticModels);
// if (true)
// return;
ModelLearningGraph modelLearningGraph = null;
ModelLearner_Old modelLearner;
boolean iterativeEvaluation = false;
boolean useCorrectType = false;
int numberOfCRFCandidates = 4;
int numberOfKnownModels;
String filePath = Params.RESULTS_DIR;
String filename = "results,k=" + numberOfCRFCandidates + ".csv";
PrintWriter resultFile = new PrintWriter(new File(filePath + filename));
StringBuffer[] resultsArray = new StringBuffer[semanticModels.size() + 2];
for (int i = 0; i < resultsArray.length; i++) {
resultsArray[i] = new StringBuffer();
}
for (int i = 0; i < semanticModels.size(); i++) {
// for (int i = 0; i <= 10; i++) {
// int i = 3; {
resultFile.flush();
int newSourceIndex = i;
SemanticModel newSource = semanticModels.get(newSourceIndex);
logger.info("======================================================");
logger.info(newSource.getName() + "(#attributes:" + newSource.getColumnNodes().size() + ")");
System.out.println(
newSource.getName() + "(#attributes:" + newSource.getColumnNodes().size() + ")");
logger.info("======================================================");
if (!iterativeEvaluation) numberOfKnownModels = semanticModels.size() - 1;
else numberOfKnownModels = 0;
if (resultsArray[0].length() > 0) resultsArray[0].append(" \t ");
resultsArray[0].append(
newSource.getName()
+ "("
+ newSource.getColumnNodes().size()
+ ")"
+ "\t"
+ " "
+ "\t"
+ " ");
if (resultsArray[1].length() > 0) resultsArray[1].append(" \t ");
resultsArray[1].append("p \t r \t t");
while (numberOfKnownModels <= semanticModels.size() - 1) {
trainingData.clear();
int j = 0, count = 0;
while (count < numberOfKnownModels) {
if (j != newSourceIndex) {
trainingData.add(semanticModels.get(j));
count++;
}
j++;
}
modelLearningGraph =
(ModelLearningGraphSparse)
ModelLearningGraph.getEmptyInstance(ontologyManager, ModelLearningGraphType.Sparse);
SemanticModel correctModel = newSource;
List<ColumnNode> columnNodes = correctModel.getColumnNodes();
// if (useCorrectType && numberOfCRFCandidates > 1)
// updateCrfSemanticTypesForResearchEvaluation(columnNodes);
modelLearner = new ModelLearner_Old(ontologyManager, columnNodes);
long start = System.currentTimeMillis();
String graphName =
!iterativeEvaluation
? graphPath + semanticModels.get(newSourceIndex).getName() + Params.GRAPH_FILE_EXT
: graphPath
+ semanticModels.get(newSourceIndex).getName()
+ ".knownModels="
+ numberOfKnownModels
+ Params.GRAPH_FILE_EXT;
if (new File(graphName).exists()) {
// read graph from file
try {
logger.info("loading the graph ...");
DirectedWeightedMultigraph<Node, DefaultLink> graph = GraphUtil.importJson(graphName);
modelLearner.graphBuilder = new GraphBuilder(ontologyManager, graph, false);
modelLearner.nodeIdFactory = modelLearner.graphBuilder.getNodeIdFactory();
} catch (Exception e) {
e.printStackTrace();
}
} else {
logger.info("building the graph ...");
for (SemanticModel sm : trainingData) modelLearningGraph.addModel(sm, false);
modelLearner.graphBuilder = modelLearningGraph.getGraphBuilder();
modelLearner.nodeIdFactory = modelLearner.graphBuilder.getNodeIdFactory();
// save graph to file
try {
GraphUtil.exportJson(
modelLearningGraph.getGraphBuilder().getGraph(), graphName, true, true);
} catch (Exception e) {
e.printStackTrace();
}
}
List<SortableSemanticModel_Old> hypothesisList =
modelLearner.hypothesize(useCorrectType, numberOfCRFCandidates);
long elapsedTimeMillis = System.currentTimeMillis() - start;
float elapsedTimeSec = elapsedTimeMillis / 1000F;
List<SortableSemanticModel_Old> topHypotheses = null;
if (hypothesisList != null) {
topHypotheses =
hypothesisList.size() > modelingConfiguration.getNumCandidateMappings()
? hypothesisList.subList(0, modelingConfiguration.getNumCandidateMappings())
: hypothesisList;
}
Map<String, SemanticModel> models = new TreeMap<String, SemanticModel>();
// export to json
// if (topHypotheses != null)
// for (int k = 0; k < topHypotheses.size() && k < 3; k++) {
//
// String fileExt = null;
// if (k == 0) fileExt = Params.MODEL_RANK1_FILE_EXT;
// else if (k == 1) fileExt = Params.MODEL_RANK2_FILE_EXT;
// else if (k == 2) fileExt = Params.MODEL_RANK3_FILE_EXT;
// SortableSemanticModel m = topHypotheses.get(k);
// new SemanticModel(m).writeJson(Params.MODEL_DIR +
// newSource.getName() + fileExt);
//
// }
ModelEvaluation me;
models.put("1-correct model", correctModel);
if (topHypotheses != null)
for (int k = 0; k < topHypotheses.size(); k++) {
SortableSemanticModel_Old m = topHypotheses.get(k);
me = m.evaluate(correctModel);
String label =
"candidate"
+ k
+ m.getSteinerNodes().getScoreDetailsString()
+ "cost:"
+ roundTwoDecimals(m.getCost())
+
// "-distance:" + me.getDistance() +
"-precision:"
+ me.getPrecision()
+ "-recall:"
+ me.getRecall();
models.put(label, m);
if (k == 0) { // first rank model
System.out.println(
"number of known models: "
+ numberOfKnownModels
+ ", precision: "
+ me.getPrecision()
+ ", recall: "
+ me.getRecall()
+ ", time: "
+ elapsedTimeSec);
logger.info(
"number of known models: "
+ numberOfKnownModels
+ ", precision: "
+ me.getPrecision()
+ ", recall: "
+ me.getRecall()
+ ", time: "
+ elapsedTimeSec);
// resultFile.println("number of known models \t precision \t recall");
// resultFile.println(numberOfKnownModels + "\t" + me.getPrecision() + "\t" +
// me.getRecall());
String s = me.getPrecision() + "\t" + me.getRecall() + "\t" + elapsedTimeSec;
if (resultsArray[numberOfKnownModels + 2].length() > 0)
resultsArray[numberOfKnownModels + 2].append(" \t ");
resultsArray[numberOfKnownModels + 2].append(s);
// resultFile.println(me.getPrecision() + "\t" + me.getRecall() + "\t" +
// elapsedTimeSec);
}
}
String outName =
!iterativeEvaluation
? outputPath
+ semanticModels.get(newSourceIndex).getName()
+ Params.GRAPHVIS_OUT_DETAILS_FILE_EXT
: outputPath
+ semanticModels.get(newSourceIndex).getName()
+ ".knownModels="
+ numberOfKnownModels
+ Params.GRAPHVIS_OUT_DETAILS_FILE_EXT;
// if (!iterativeEvaluation) {
GraphVizUtil.exportSemanticModelsToGraphviz(
models,
newSource.getName(),
outName,
GraphVizLabelType.LocalId,
GraphVizLabelType.LocalUri,
false,
false);
// }
numberOfKnownModels++;
}
// resultFile.println("=======================================================");
}
for (StringBuffer s : resultsArray) resultFile.println(s.toString());
resultFile.close();
}
@Override
public UpdateContainer doIt(Workspace workspace) throws CommandException {
Worksheet worksheet = workspace.getWorksheet(worksheetId);
SuperSelection selection = getSuperSelection(worksheet);
HNodePath selectedPath = null;
List<HNodePath> columnPaths = worksheet.getHeaders().getAllPaths();
for (HNodePath path : columnPaths) {
if (path.getLeaf().getId().equals(hNodeId)) {
selectedPath = path;
}
}
Collection<Node> nodes = new ArrayList<Node>();
workspace
.getFactory()
.getWorksheet(worksheetId)
.getDataTable()
.collectNodes(selectedPath, nodes, selection);
try {
JSONArray requestJsonArray = new JSONArray();
for (Node node : nodes) {
String id = node.getId();
String originalVal = node.getValue().asString();
JSONObject jsonRecord = new JSONObject();
jsonRecord.put("id", id);
originalVal = originalVal == null ? "" : originalVal;
jsonRecord.put("value", originalVal);
requestJsonArray.put(jsonRecord);
}
String jsonString = null;
jsonString = requestJsonArray.toString();
// String url =
// "http://localhost:8080/cleaningService/IdentifyData";
// String url = "http://localhost:8070/myWS/IdentifyData";
String url =
ServletContextParameterMap.getParameterValue(ContextParameter.CLEANING_SERVICE_URL);
HttpClient httpclient = new DefaultHttpClient();
HttpPost httppost = null;
HttpResponse response = null;
HttpEntity entity;
StringBuffer out = new StringBuffer();
URI u = null;
u = new URI(url);
List<NameValuePair> formparams = new ArrayList<NameValuePair>();
formparams.add(new BasicNameValuePair("json", jsonString));
httppost = new HttpPost(u);
httppost.setEntity(new UrlEncodedFormEntity(formparams, "UTF-8"));
response = httpclient.execute(httppost);
entity = response.getEntity();
if (entity != null) {
BufferedReader buf = new BufferedReader(new InputStreamReader(entity.getContent()));
String line = buf.readLine();
while (line != null) {
out.append(line);
line = buf.readLine();
}
}
// logger.trace(out.toString());
// logger.info("Connnection success : " + url + " Successful.");
final JSONObject data1 = new JSONObject(out.toString());
// logger.trace("Data--->" + data1);
return new UpdateContainer(
new AbstractUpdate() {
@Override
public void generateJson(String prefix, PrintWriter pw, VWorkspace vWorkspace) {
JSONObject response = new JSONObject();
// logger.trace("Reached here");
try {
response.put("updateType", "CleaningServiceOutput");
response.put("chartData", data1);
response.put("hNodeId", hNodeId);
// logger.trace(response.toString(4));
} catch (JSONException e) {
pw.print("Error");
}
pw.print(response.toString());
}
});
} catch (Exception e) {
e.printStackTrace();
return new UpdateContainer(new ErrorUpdate("Error!"));
}
}
