/**
* Create a cluster for publications, based on language and year
*
* @param author
* @param publicationClustersMap
*/
public void constructPublicationClusterByLanguageAndYear(
Author author, Map<String, PublicationClusterHelper> publicationClustersMap) {
// fill publication clusters
// prepare calendar for publication year
Calendar calendar = Calendar.getInstance();
// get all publications from specific author and put it into cluster
for (Publication publication : author.getPublications()) {
// only proceed publication that have date, language and abstract
if (publication.getAbstractText() == null || publication.getAbstractText().equals(""))
continue;
if (publication.getPublicationDate() == null) continue;
if (publication.getLanguage() == null) continue;
// get publication year
calendar.setTime(publication.getPublicationDate());
// construct clusterMap key
String clusterMapKey = publication.getLanguage() + calendar.get(Calendar.YEAR);
// construct publication map
if (publicationClustersMap.get(clusterMapKey) == null) {
// not exist create new cluster
PublicationClusterHelper publicationCluster = new PublicationClusterHelper();
publicationCluster.setLangauge(publication.getLanguage());
publicationCluster.setYear(calendar.get(Calendar.YEAR));
publicationCluster.addPublicationAndUpdate(publication);
// add into map
publicationClustersMap.put(clusterMapKey, publicationCluster);
} else {
// exist on map, get the cluster
PublicationClusterHelper publicationCluster = publicationClustersMap.get(clusterMapKey);
publicationCluster.addPublicationAndUpdate(publication);
}
}
}
/**
* 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);
}
}
