/**
* Calculate each default InterestProfile
*
* @param author
* @param interestProfileDefault
* @param publicationClustersMap
*/
public void calculateEachInterestProfileDefault(
Author author,
Set<Interest> newInterests,
InterestProfile interestProfileDefault,
Map<String, PublicationClusterHelper> publicationClustersMap) {
// get author interest profile
Calendar calendar = Calendar.getInstance();
// default profile name [DEFAULT_PROFILENAME]
String authorInterestProfileName = interestProfileDefault.getName();
// create new author interest profile for c-value
AuthorInterestProfile authorInterestProfile = new AuthorInterestProfile();
authorInterestProfile.setCreated(calendar.getTime());
authorInterestProfile.setDescription(
"Interest mining using " + interestProfileDefault.getName() + " algorithm");
authorInterestProfile.setName(authorInterestProfileName);
// CorePhrase and WordFreq specific, according to Svetoslav Evtimov thesis
// yearFactor Map format Map< Language-Year , value >
// totalYearsFactor Map< Language, value >
Map<String, Double> yearFactorMap = new HashMap<String, Double>();
Map<String, Double> totalYearsFactorMap = new HashMap<String, Double>();
// calculate some weighting factors
// if ( interestProfileDefault.getName().toLowerCase().equals( "corephrase" ) ||
// interestProfileDefault.getName().toLowerCase().equals( "wordfreq" ) )
// {
// yearFactorMap = CorePhraseAndWordFreqHelper.calculateYearFactor( publicationClustersMap,
// 0.25 );
// totalYearsFactorMap = CorePhraseAndWordFreqHelper.calculateTotalYearsFactor(
// publicationClustersMap );
// }
// get the number of active extraction services
int numberOfExtractionService = applicationService.getExtractionServices().size();
// loop to each cluster and calculate default profiles
for (Map.Entry<String, PublicationClusterHelper> publicationClusterEntry :
publicationClustersMap.entrySet()) {
PublicationClusterHelper publicationCluster = publicationClusterEntry.getValue();
if (publicationCluster.getTermMap() == null || publicationCluster.getTermMap().isEmpty())
continue;
// prepare variables
AuthorInterest authorInterest = new AuthorInterest();
// assign author interest method
if (interestProfileDefault.getName().toLowerCase().equals("cvalue")) {
cValueInterestProfile.doCValueCalculation(
authorInterest, newInterests, publicationCluster, numberOfExtractionService);
} else if (interestProfileDefault.getName().toLowerCase().equals("corephrase")) {
Double yearFactor =
yearFactorMap.get(publicationCluster.getLanguage() + publicationCluster.getYear());
Double totalYearFactor = totalYearsFactorMap.get(publicationCluster.getLanguage());
corePhraseInterestProfile.doCorePhraseCalculation(
authorInterest,
newInterests,
publicationCluster,
yearFactor,
totalYearFactor,
numberOfExtractionService);
} else if (interestProfileDefault.getName().toLowerCase().equals("wordfreq")) {
Double yearFactor =
yearFactorMap.get(publicationCluster.getLanguage() + publicationCluster.getYear());
Double totalYearFactor = totalYearsFactorMap.get(publicationCluster.getLanguage());
wordFreqInterestProfile.doWordFreqCalculation(
authorInterest,
newInterests,
publicationCluster,
yearFactor,
totalYearFactor,
numberOfExtractionService);
}
// Put other default interest profiles
else if (interestProfileDefault.getName().toLowerCase().equals("lda")) {
}
// at the end persist new interests
// for ( Interest newInterest : newInterests )
// persistenceStrategy.getInterestDAO().persist( newInterest );
// check author interest calculation result
if (authorInterest.getTermWeights() != null && !authorInterest.getTermWeights().isEmpty()) {
authorInterest.setAuthorInterestProfile(authorInterestProfile);
authorInterestProfile.addAuthorInterest(authorInterest);
authorInterestProfile.setInterestProfile(interestProfileDefault);
// persistenceStrategy.getAuthorInterestProfileDAO().persist( authorInterestProfile );
}
}
// at the end persist
if (authorInterestProfile.getAuthorInterests() != null
&& !authorInterestProfile.getAuthorInterests().isEmpty()) {
authorInterestProfile.setAuthor(author);
author.addAuthorInterestProfiles(authorInterestProfile);
persistenceStrategy.getAuthorDAO().persist(author);
}
}
/**
* Calculated derived interest profile (Intersection and/or Union between interest profile) in an
* author
*
* @param author
* @param interestProfilesDerived
*/
private void calculateInterestProfilesDerived(
Author author, List<InterestProfile> interestProfilesDerived) {
// get authorInterest set on profile
for (InterestProfile interestProfileDerived : interestProfilesDerived) {
String[] derivedInterestProfileName = interestProfileDerived.getName().split("\\s+");
// at list profile name has three segment
if (derivedInterestProfileName.length < 3) continue;
// prepare variables
AuthorInterestProfile authorInterestProfile1 = null;
AuthorInterestProfile authorInterestProfile2 = null;
AuthorInterestProfile authorInterestProfileResult = null;
String operationType = null;
for (String partOfProfileName : derivedInterestProfileName) {
// ? sometimes problem on encoding
if (partOfProfileName.equals("∩")
|| partOfProfileName.equals("?")
|| partOfProfileName.equals("+")
|| partOfProfileName.equals("∪")) {
if (authorInterestProfileResult != null) {
authorInterestProfile1 = authorInterestProfileResult;
authorInterestProfileResult = null;
}
if (partOfProfileName.equals("∩")
|| partOfProfileName.equals("?")
|| partOfProfileName.equals("+")) operationType = "INTERSECTION";
else operationType = "UNION";
} else {
if (authorInterestProfile1 == null) {
authorInterestProfile1 = author.getSpecificAuthorInterestProfile(partOfProfileName);
if (authorInterestProfile1 == null) {
logger.error("AuthorInterestProfile " + partOfProfileName + " not found");
// continue to next derived author profile, if exist
break;
}
} else {
authorInterestProfile2 = author.getSpecificAuthorInterestProfile(partOfProfileName);
if (authorInterestProfile2 == null) {
logger.error("AuthorInterestProfile " + partOfProfileName + " not found");
// continue to next derived author profile, if exist
break;
}
}
// calculate and persist
if (authorInterestProfile1 != null
&& authorInterestProfile2 != null
&& operationType != null) {
if (operationType.equals("INTERSECTION"))
authorInterestProfileResult =
calculateIntersectionOfAuthorInterestProfiles(
authorInterestProfile1, authorInterestProfile2, interestProfileDerived);
else
authorInterestProfileResult =
calculateUnionOfAuthorInterestProfiles(
authorInterestProfile1, authorInterestProfile2, interestProfileDerived);
}
}
}
// persist result
if (authorInterestProfileResult != null
&& (authorInterestProfileResult.getAuthorInterests() != null
&& !authorInterestProfileResult.getAuthorInterests().isEmpty())) {
authorInterestProfileResult.setAuthor(author);
author.addAuthorInterestProfiles(authorInterestProfileResult);
persistenceStrategy.getAuthorDAO().persist(author);
persistenceStrategy.getAuthorInterestProfileDAO().persist(authorInterestProfileResult);
}
}
}
