private NLGElement getNLFromExpressions(List<Expr> expressions) {
List<NLGElement> nlgs = new ArrayList<NLGElement>();
NLGElement elt;
for (Expr e : expressions) {
elt = getNLFromSingleExpression(e);
if (elt != null) {
nlgs.add(elt);
}
}
// now process
if (nlgs.isEmpty()) {
return null;
}
if (nlgs.size() == 1) {
return nlgs.get(0);
} else {
CoordinatedPhraseElement cpe;
cpe = nlgFactory.createCoordinatedPhrase(nlgs.get(0), nlgs.get(1));
for (int i = 2; i < nlgs.size(); i++) {
cpe.addComplement(nlgs.get(i));
}
cpe.setConjunction("and");
return cpe;
}
}
public NLGElement getNLForTripleList(List<Triple> triples, String conjunction) {
if (triples.isEmpty()) {
return null;
}
if (triples.size() == 1) {
return getNLForTriple(triples.get(0));
} else {
CoordinatedPhraseElement cpe;
Triple t0 = triples.get(0);
Triple t1 = triples.get(1);
cpe = nlgFactory.createCoordinatedPhrase(getNLForTriple(t0), getNLForTriple(t1));
for (int i = 2; i < triples.size(); i++) {
cpe.addCoordinate(getNLForTriple(triples.get(i)));
}
cpe.setConjunction(conjunction);
return cpe;
}
}
/**
* Processes a list of elements. These can be elements of the where clause or of an optional
* clause
*
* @param e List of query elements
* @return Conjunctive natural representation of the list of elements.
*/
public NLGElement getNLFromElements(List<Element> e) {
if (e.isEmpty()) {
return null;
}
if (e.size() == 1) {
return getNLFromSingleClause(e.get(0));
} else {
CoordinatedPhraseElement cpe;
cpe =
nlgFactory.createCoordinatedPhrase(
getNLFromSingleClause(e.get(0)), getNLFromSingleClause(e.get(1)));
for (int i = 2; i < e.size(); i++) {
cpe.addCoordinate(getNLFromSingleClause(e.get(i)));
}
cpe.setConjunction("and");
return cpe;
}
}
private NLGElement processTypes(
Map<String, Set<String>> typeMap, Set<String> vars, boolean count, boolean distinct) {
List<NPPhraseSpec> objects = new ArrayList<NPPhraseSpec>();
// process the type information to create the object(s)
for (String s : typeMap.keySet()) {
if (vars.contains(s)) {
// contains the objects to the sentence
NPPhraseSpec object;
object = nlgFactory.createNounPhrase("?" + s);
Set<String> types = typeMap.get(s);
if (types.size() == 1) {
NPPhraseSpec np = getNPPhrase(types.iterator().next(), true);
if (distinct) {
np.addModifier("distinct");
}
object.addPreModifier(np);
} else {
Iterator<String> typeIterator = types.iterator();
String type0 = typeIterator.next();
String type1 = typeIterator.next();
NPPhraseSpec np0 = getNPPhrase(type0, true);
// if (distinct) {
// np0.addModifier("distinct");
// }
NPPhraseSpec np1 = getNPPhrase(type1, true);
// if (distinct) {
// np1.addModifier("distinct");
// }
CoordinatedPhraseElement cpe = nlgFactory.createCoordinatedPhrase(np0, np1);
while (typeIterator.hasNext()) {
NPPhraseSpec np = getNPPhrase(typeIterator.next(), true);
// if (distinct) {
// np.addModifier("distinct");
// }
cpe.addCoordinate(np);
}
cpe.setConjunction("as well as");
if (distinct) {
cpe.addPreModifier("distinct");
}
object.addPreModifier(cpe);
}
object.setFeature(Feature.CONJUNCTION, "or");
objects.add(object);
}
}
if (objects.size() == 1) {
// if(count) objects.get(0).addPreModifier("the number of");
return objects.get(0);
} else {
CoordinatedPhraseElement cpe =
nlgFactory.createCoordinatedPhrase(objects.get(0), objects.get(1));
if (objects.size() > 2) {
for (int i = 2; i < objects.size(); i++) {
cpe.addCoordinate(objects.get(i));
}
}
// if(count) cpe.addPreModifier("the number of");
return cpe;
}
}
/**
* Generates a natural language representation for SELECT queries
*
* @param query Input query
* @return Natural Language Representation
*/
public DocumentElement convertSelectAndAsk(Query query) {
// List of sentences for the output
List<DocumentElement> sentences = new ArrayList<DocumentElement>();
// System.out.println("Input query = " + query);
// preprocess the query to get the relevant types
TypeExtractor tEx = new TypeExtractor(endpoint);
Map<String, Set<String>> typeMap = tEx.extractTypes(query);
// System.out.println("Processed query = " + query);
// contains the beginning of the query, e.g., "this query returns"
SPhraseSpec head = nlgFactory.createClause();
String conjunction = "such that";
NLGElement body;
NLGElement postConditions;
List<Element> whereElements = getWhereElements(query);
List<Element> optionalElements = getOptionalElements(query);
// first sort out variables
Set<String> projectionVars = typeMap.keySet();
Set<String> whereVars = getVars(whereElements, projectionVars);
// case we only have stuff such as rdf:type queries
if (whereVars.isEmpty()) {
whereVars = projectionVars;
}
Set<String> optionalVars = getVars(optionalElements, projectionVars);
// important. Remove variables that have already been declared in first
// sentence from second sentence
for (String var : whereVars) {
if (optionalVars.contains(var)) {
optionalVars.remove(var);
}
}
// process SELECT queries
if (query.isSelectType()) {
// process head
// we could create a lexicon from which we could read these
head.setSubject("This query");
if (!tEx.isCount()) {
head.setVerb("retrieve");
} else {
head.setVerb("retrieve the number of");
}
} // process ASK queries
else {
// process factual ask queries (no variables at all)
if (typeMap.isEmpty()) {
head.setSubject("This query");
head.setVerb("ask whether");
head.setObject(getNLFromElements(whereElements));
head.setFeature(Feature.SUPRESSED_COMPLEMENTISER, true);
phrases.add(head);
sentences.add(nlgFactory.createSentence(head));
return nlgFactory.createParagraph(sentences);
}
// process head
// we could create a lexicon from which we could read these
head.setSubject("This query");
head.setVerb("ask for the existence of");
}
NLGElement e = processTypes(typeMap, whereVars, tEx.isCount(), query.isDistinct());
head.setObject(e);
// now generate body
if (!whereElements.isEmpty()) {
body = getNLFromElements(whereElements);
// now add conjunction
CoordinatedPhraseElement phrase1 = nlgFactory.createCoordinatedPhrase(head, body);
phrase1.setConjunction("such that");
// add as first sentence
sentences.add(nlgFactory.createSentence(phrase1));
phrases.add(phrase1);
// this concludes the first sentence.
} else {
sentences.add(nlgFactory.createSentence(head));
phrases.add(head);
}
// The second sentence deals with the optional clause (if it exists)
if (optionalElements != null && !optionalElements.isEmpty()) {
// the optional clause exists
// if no supplementary projection variables are used in the clause
if (optionalVars.isEmpty()) {
SPhraseSpec optionalHead = nlgFactory.createClause();
optionalHead.setSubject("it");
optionalHead.setVerb("retrieve");
optionalHead.setObject("data");
optionalHead.setFeature(Feature.CUE_PHRASE, "Additionally, ");
NLGElement optionalBody = getNLFromElements(optionalElements);
CoordinatedPhraseElement optionalPhrase =
nlgFactory.createCoordinatedPhrase(optionalHead, optionalBody);
optionalPhrase.setConjunction("such that");
optionalPhrase.addComplement("if such exist");
sentences.add(nlgFactory.createSentence(optionalPhrase));
} // if supplementary projection variables are used in the clause
else {
SPhraseSpec optionalHead = nlgFactory.createClause();
optionalHead.setSubject("it");
optionalHead.setVerb("retrieve");
optionalHead.setObject(
processTypes(typeMap, optionalVars, query.isDistinct(), query.isDistinct()));
optionalHead.setFeature(Feature.CUE_PHRASE, "Additionally, ");
if (!optionalElements.isEmpty()) {
NLGElement optionalBody;
optionalBody = getNLFromElements(optionalElements);
// now add conjunction
CoordinatedPhraseElement optionalPhrase =
nlgFactory.createCoordinatedPhrase(optionalHead, optionalBody);
optionalPhrase.setConjunction("such that");
// add as second sentence
optionalPhrase.addComplement("if such exist");
sentences.add(nlgFactory.createSentence(optionalPhrase));
phrases.add(optionalPhrase);
// this concludes the second sentence.
} else {
optionalHead.addComplement("if such exist");
sentences.add(nlgFactory.createSentence(optionalHead));
phrases.add(optionalHead);
}
}
}
// The last sentence deals with the result modifiers
if (query.hasHaving()) {
SPhraseSpec modifierHead = nlgFactory.createClause();
modifierHead.setSubject("it");
modifierHead.setVerb("return exclusively");
modifierHead.setObject("results");
modifierHead.getObject().setPlural(true);
modifierHead.setFeature(Feature.CUE_PHRASE, "Moreover, ");
List<Expr> expressions = query.getHavingExprs();
CoordinatedPhraseElement phrase =
nlgFactory.createCoordinatedPhrase(modifierHead, getNLFromExpressions(expressions));
phrase.setConjunction("such that");
sentences.add(nlgFactory.createSentence(phrase));
}
if (query.hasOrderBy()) {
SPhraseSpec order = nlgFactory.createClause();
order.setSubject("The results");
order.getSubject().setPlural(true);
order.setVerb("be ordered by");
List<SortCondition> sc = query.getOrderBy();
if (sc.size() == 1) {
Expr expr = sc.get(0).getExpression();
if (expr instanceof ExprVar) {
ExprVar ev = (ExprVar) expr;
order.setObject(ev.toString());
}
if (sc.get(0).direction < 0) {
order.addComplement("in descending order");
} else {
order.addComplement("in ascending order");
}
}
phrases.add(order);
sentences.add(nlgFactory.createSentence(order));
}
if (query.hasLimit()) {
SPhraseSpec limitOffset = nlgFactory.createClause();
long limit = query.getLimit();
if (query.hasOffset()) {
long offset = query.getOffset();
limitOffset.setSubject("The query");
limitOffset.setVerb("return");
limitOffset.setObject("results between number " + limit + " and " + (offset + limit));
} else {
limitOffset.setSubject("The query");
limitOffset.setVerb("return");
if (limit > 1) {
if (query.hasOrderBy()) {
limitOffset.setObject("at most the first " + limit + " results");
} else {
limitOffset.setObject("at most " + limit + " results");
}
} else {
if (query.hasOrderBy()) {
limitOffset.setObject("at most the first result");
} else {
limitOffset.setObject("at most one result");
}
}
}
phrases.add(limitOffset);
sentences.add(nlgFactory.createSentence(limitOffset));
}
DocumentElement result = nlgFactory.createParagraph(sentences);
return result;
}