/// <summary>
/// return a clone of the token send to the function
/// </summary>
/// <param name="token"></param>
/// <returns></returns>
public override LexiconToken getClone(LexiconToken token)
{
LexiconLiteral literalToReplace = new LexiconLiteral();
literalToReplace.URI = token.URI;
literalToReplace.label = token.label;
literalToReplace.typeOfOwner = (token as LexiconLiteral).typeOfOwner.ToList();
literalToReplace.QuestionMatch = token.QuestionMatch;
literalToReplace.score = token.score;
return(literalToReplace);
}
/// <summary>
/// get predicates is a method in lexicon class that get all LexiconLiterals objects that match some words in the Question
/// </summary>
/// <param name="question">question to get matched predicates of it </param>
/// <param name="topN">the number of top matching results to be returned, default = 10</param>
/// <param name="Limit">the limit of the number of returned results in the query, default = 20</param>
/// <returns>list of top matching LexiconLiterals with it's type of owner and predicate </returns>
public List<LexiconLiteral> getLiterals(string question, int topN = 30, int Limit = 30)
{
DateTime dt = DateTime.Now; // capturing time for testing
List<LexiconLiteral> __literalList = new List<LexiconLiteral>();
//getting all permutation of words formed from the question string
List<string> permutationList = getPermutations(question);
//removing permutations that most propbably wont return results and will take time in querying
permutationList = trimPermutations(permutationList);
// to check if the literals are filled before - so returning the matching Literals only or not
if (literalFilled)
{
foreach (LexiconLiteral literal in this.literalList)
{
if (permutationList.Contains(literal.QuestionMatch))
{
__literalList.Add(literal);
}
}
return __literalList;
}
else
{
// iterating over each permutation of Question left and Query them from virtuoso and return predicate list and add them
foreach (string questionleft in permutationList)
{
string Query = "select distinct ?subject ?literal ?redirects ?typeOfOwner ?redirectsTypeOfOwner where{" +
"?subject <http://www.w3.org/2000/01/rdf-schema#label> ?literal." +
"optional { ?subject <http://dbpedia.org/ontology/wikiPageRedirects> ?redirects . " +
"optional {?redirects <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?redirectsTypeOfOwner ." +
"}}." +
"optional {?subject <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?typeOfOwner}." +
"Filter ( !bound(?typeOfOwner) || " +
" ( !(?typeOfOwner = <http://www.w3.org/2004/02/skos/core#Concept>)" +
" && !(?typeOfOwner = <http://www.w3.org/2002/07/owl#Thing>) " +
" && !(?typeOfOwner = <http://www.opengis.net/gml/_Feature>) " +
" && !(?typeOfOwner = <http://www.w3.org/2002/07/owl#ObjectProperty>) " +
" && !(?typeOfOwner = <http://www.w3.org/1999/02/22-rdf-syntax-ns#Property> ) " +
" && !(?typeOfOwner = <http://www.w3.org/2002/07/owl#DatatypeProperty> )))." +
"?literal bif:contains '\"" + questionleft + "\"'.} limit " + Limit;
SparqlRemoteEndpoint remoteEndPoint = new SparqlRemoteEndpoint(new Uri("http://localhost:8890/sparql"));
SparqlResultSet resultSet = new SparqlResultSet();
try
{
resultSet = remoteEndPoint.QueryWithResultSet(Query);
}
// skipping results that raised timeout exceptions
catch
{
util.log("skipped Query3: " + questionleft + " ---- due to time out ");
}
//iterating over matched Literals in the resultset
foreach (SparqlResult result in resultSet)
{
string resultTypeOfOwner = "";
string resultURI;
string resultLabel = result.Value("literal").ToString();
string resultquestionMatch = questionleft;
if (result.Value("redirects") != null)
{
resultURI = result.Value("redirects").ToString();
if (result.Value("redirectsTypeOfOwner") != null)
{
resultTypeOfOwner = result.Value("redirectsTypeOfOwner").ToString();
}
}
else
{
resultURI = result.Value("subject").ToString();
if (result.Value("typeOfOwner") != null)
{
resultTypeOfOwner = result.Value("typeOfOwner").ToString();
}
}
// check that the predicate doesn't exists in the predicateslist before
bool exists = false; // URI + Label only Exists
bool totallyExists = false; // URI + Label + TypeofOwner exists in the literal list
foreach (LexiconLiteral x in __literalList)
{
if (x.URI == resultURI && x.label == resultLabel && x.QuestionMatch == resultquestionMatch)
{
exists = true;
if (x.typeOfOwner.Contains(resultTypeOfOwner) && resultTypeOfOwner.Length > 0)
{
totallyExists = true;
break;
}
}
}
// adding the new literals to the literallist.
if (!exists)
{
LexiconLiteral tmpLexiconLiteral = new LexiconLiteral(resultURI, resultLabel, resultquestionMatch, resultTypeOfOwner);
__literalList.Add(tmpLexiconLiteral);
}
if (!totallyExists && exists)
{
foreach (LexiconLiteral let in __literalList)
{
if (let.URI == resultURI && let.label == resultLabel)
{
let.typeOfOwner.Add(resultTypeOfOwner);
}
}
}
}
}
//scoring literals . trimming duplicates ,
__literalList = scoreLiterals(__literalList, topN);
util.log("total time taken :" + DateTime.Now.Subtract(dt).TotalMilliseconds.ToString() + " msecs ");
literalFilled = true;
this.literalList = __literalList;
return __literalList;
}
}
/// removing the non used predicates domains and the literals type of owners
/// </summary>
/// <param name="tokens">list </param>of tokens
/// <returns>cleaned list of tokens </returns>
private List <QueryBucket> cleanBucket(List <QueryBucket> queryBuckets)
{
#region removing Buckets which still have question left >1
foreach (QueryBucket querybucket in queryBuckets.ToList())
{
if (querybucket.questionLeft.Length > 0)
{
queryBuckets.Remove(querybucket);
}
}
#endregion
#region remove Predicates domains and type of owners
foreach (QueryBucket bucket in queryBuckets.ToList())
{
//adding predicates and literals to a list
List <LexiconPredicate> predicateList = new List <LexiconPredicate>();
List <LexiconLiteral> literalList = new List <LexiconLiteral>();
foreach (LexiconToken token in bucket.tokens)
{
if (token is LexiconPredicate)
{
predicateList.Add(token as LexiconPredicate);
}
if (token is LexiconLiteral)
{
literalList.Add(token as LexiconLiteral);
}
}
if (predicateList.Count > 0)
{
//removing domains and ranges that are not used
foreach (LexiconToken token in bucket.tokens.ToList())
{
if (token is LexiconPredicate)
{
//casting the lexicontoken to lexicon predicate
LexiconPredicate oldPredicate = token as LexiconPredicate;
//cloning the token to be modified
LexiconPredicate predicateToReplace = (LexiconPredicate)token.getClone(token);
foreach (string oldPredDomain in oldPredicate.domains.ToList())
{
bool exist = false;
foreach (LexiconLiteral tmpliteral in literalList)
{
if (tmpliteral.typeOfOwner.Contains(oldPredDomain))
{
exist = true;
}
}
//if this domains doesn't contained in any of literals type of owners remove it as it wont match|join with anything
if (!exist)
{
//old bucket = new bucket and then modify in the new in order then to be able to remove the old
predicateToReplace = oldPredicate.getClone(oldPredicate) as LexiconPredicate;
//removing domain not used
predicateToReplace.domains.Remove(oldPredDomain);
//remove the old bucket and replace it with new modified one // needed because of reference issues
bucket.tokens.Remove(oldPredicate);
bucket.tokens.Add(predicateToReplace);
oldPredicate = predicateToReplace;
//remove the predicate if it doesnt have any domains left
if (oldPredicate.domains.Count == 0)
{
bucket.tokens.Remove(oldPredicate);
predicateList.Remove(oldPredicate as LexiconPredicate);
//remove the bucket if it's free from predicates
if (bucket.tokens.Count == 0)
{
queryBuckets.Remove(bucket);
}
}
}
}
}
if (token is LexiconLiteral)
{
LexiconLiteral oldLiteral = token as LexiconLiteral;
LexiconLiteral newLiteral = token.getClone(token) as LexiconLiteral;
foreach (string typeofowner in oldLiteral.typeOfOwner.ToList())
{
bool exist = false;
foreach (LexiconPredicate tmmpredicate in predicateList)
{
if (tmmpredicate.domains.Contains(typeofowner))
{
exist = true;
}
}
if (!exist)
{
//taking a copy from the old literal in order to remove it from the bucket when replacing it with the newliteral
newLiteral = oldLiteral.getClone(oldLiteral) as LexiconLiteral;
// removing typeofowner not used
newLiteral.typeOfOwner.Remove(typeofowner);
// updating the bucket tokens by replacing the old literal with the new one
bucket.tokens.Remove(oldLiteral);
bucket.tokens.Add(newLiteral);
oldLiteral = newLiteral;
if (oldLiteral.typeOfOwner.Count == 0)
{
bucket.tokens.Remove(oldLiteral);
literalList.Remove(oldLiteral as LexiconLiteral);
//remove the bucket if it's free from Tokens
if (bucket.tokens.Count == 0)
{
queryBuckets.Remove(bucket);
}
}
}
}
}
}
}
else
{
bucket.literalOnly = true;
}
}
#endregion
#region remove the multiple domains and multiple ranges
foreach (QueryBucket bucket in queryBuckets)
{
foreach (LexiconToken predicateToken in bucket.tokens)
{
if (predicateToken is LexiconPredicate)
{
foreach (LexiconToken literalToken in bucket.tokens)
{
if (literalToken is LexiconLiteral && Enumerable.SequenceEqual((predicateToken as LexiconPredicate).domains, (literalToken as LexiconLiteral).typeOfOwner))
{
(predicateToken as LexiconPredicate).domains.RemoveRange(1, (predicateToken as LexiconPredicate).domains.Count - 1);
(literalToken as LexiconLiteral).typeOfOwner.RemoveRange(1, (literalToken as LexiconLiteral).typeOfOwner.Count - 1);
}
}
}
}
}
#endregion
return(queryBuckets);
}
/// <summary>
/// get predicates is a method in lexicon class that get all LexiconLiterals objects that match some words in the Question
/// </summary>
/// <param name="question">question to get matched predicates of it </param>
/// <param name="topN">the number of top matching results to be returned, default = 10</param>
/// <param name="Limit">the limit of the number of returned results in the query, default = 20</param>
/// <returns>list of top matching LexiconLiterals with it's type of owner and predicate </returns>
public List <LexiconLiteral> getLiterals(string question, int topN = 30, int Limit = 30)
{
DateTime dt = DateTime.Now; // capturing time for testing
List <LexiconLiteral> __literalList = new List <LexiconLiteral>();
//getting all permutation of words formed from the question string
List <string> permutationList = getPermutations(question);
//removing permutations that most propbably wont return results and will take time in querying
permutationList = trimPermutations(permutationList);
// to check if the literals are filled before - so returning the matching Literals only or not
if (literalFilled)
{
foreach (LexiconLiteral literal in this.literalList)
{
if (permutationList.Contains(literal.QuestionMatch))
{
__literalList.Add(literal);
}
}
return(__literalList);
}
else
{
// iterating over each permutation of Question left and Query them from virtuoso and return predicate list and add them
foreach (string questionleft in permutationList)
{
string Query = "select distinct ?subject ?literal ?redirects ?typeOfOwner ?redirectsTypeOfOwner where{" +
"?subject <http://www.w3.org/2000/01/rdf-schema#label> ?literal." +
"optional { ?subject <http://dbpedia.org/ontology/wikiPageRedirects> ?redirects . " +
"optional {?redirects <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?redirectsTypeOfOwner ." +
"}}." +
"optional {?subject <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?typeOfOwner}." +
"Filter ( !bound(?typeOfOwner) || " +
" ( !(?typeOfOwner = <http://www.w3.org/2004/02/skos/core#Concept>)" +
" && !(?typeOfOwner = <http://www.w3.org/2002/07/owl#Thing>) " +
" && !(?typeOfOwner = <http://www.opengis.net/gml/_Feature>) " +
" && !(?typeOfOwner = <http://www.w3.org/2002/07/owl#ObjectProperty>) " +
" && !(?typeOfOwner = <http://www.w3.org/1999/02/22-rdf-syntax-ns#Property> ) " +
" && !(?typeOfOwner = <http://www.w3.org/2002/07/owl#DatatypeProperty> )" +
" && !(?typeOfOwner = <http://www.w3.org/2002/07/owl#Class> )))." +
"?literal bif:contains '\"" + questionleft + "\"'.} limit " + Limit;
//SparqlRemoteEndpoint remoteEndPoint = new SparqlRemoteEndpoint(new Uri("http://localhost:8890/sparql"));
SparqlResultSet resultSet = new SparqlResultSet();
try
{
//resultSet = remoteEndPoint.QueryWithResultSet(Query);
resultSet = Request.RequestWithHTTP(Query);
}
// skipping results that raised timeout exceptions
catch
{
util.log("skipped Query3: " + questionleft + " ---- due to time out ");
}
//iterating over matched Literals in the resultset
foreach (SparqlResult result in resultSet)
{
string resultTypeOfOwner = "";
string resultURI;
string resultLabel = result.Value("literal").ToString();
string resultquestionMatch = questionleft;
if (result.Value("redirects") != null)
{
resultURI = result.Value("redirects").ToString();
if (result.Value("redirectsTypeOfOwner") != null)
{
resultTypeOfOwner = result.Value("redirectsTypeOfOwner").ToString();
}
}
else
{
resultURI = result.Value("subject").ToString();
if (result.Value("typeOfOwner") != null)
{
resultTypeOfOwner = result.Value("typeOfOwner").ToString();
}
}
// check that the predicate doesn't exists in the predicateslist before
bool exists = false; // URI + Label only Exists
bool totallyExists = false; // URI + Label + TypeofOwner exists in the literal list
foreach (LexiconLiteral x in __literalList)
{
if (x.URI == resultURI && x.label == resultLabel && x.QuestionMatch == resultquestionMatch)
{
exists = true;
if (x.typeOfOwner.Contains(resultTypeOfOwner) && resultTypeOfOwner.Length > 0)
{
totallyExists = true;
break;
}
}
}
// adding the new literals to the literallist.
if (!exists)
{
LexiconLiteral tmpLexiconLiteral = new LexiconLiteral(resultURI, resultLabel, resultquestionMatch, resultTypeOfOwner);
__literalList.Add(tmpLexiconLiteral);
}
if (!totallyExists && exists)
{
foreach (LexiconLiteral let in __literalList)
{
if (let.URI == resultURI && let.label == resultLabel)
{
let.typeOfOwner.Add(resultTypeOfOwner);
}
}
}
}
}
//scoring literals . trimming duplicates ,
__literalList = scoreLiterals(__literalList, topN);
util.log("total time taken :" + DateTime.Now.Subtract(dt).TotalMilliseconds.ToString() + " msecs ");
literalFilled = true;
this.literalList = __literalList;
return(__literalList);
}
}
/// <summary>
/// return a clone of the token send to the function
/// </summary>
/// <param name="token"></param>
/// <returns></returns>
public override LexiconToken getClone(LexiconToken token)
{
LexiconLiteral literalToReplace = new LexiconLiteral();
literalToReplace.URI = token.URI;
literalToReplace.label = token.label;
literalToReplace.typeOfOwner = (token as LexiconLiteral).typeOfOwner.ToList();
literalToReplace.QuestionMatch = token.QuestionMatch;
literalToReplace.score = token.score;
return literalToReplace;
}