private void initModelFromRepo()
{
string msg;
DateTime endt = DateTime.UtcNow;
DateTime startt = DateTime.UtcNow.AddHours(-1 * MaxHoursConsidered);
//DateTime startt = DateTime.UtcNow.AddHours(-1 * 36); // for testing
msg = String.Format("initModelFromRepo(): Building Initial Model for Articles in Vertical {0} \n\t from startTime [{1}] to endTime [{2}].",
VerticalConstants.GetFriendlyName(VerticalId), startt.ToString(), endt.ToString());
logger.Info(msg);
logger.Debug("Retrieving Articles from Pickscomb repository.");
List <Article> articles = null;
try
{
IPickscombRepository repo = PickscombRepository.Instance;
logger.Info("PickscombRepository instantiated. Fetching Articles . . .");
articles = repo.GetArticles(VerticalId, startt, endt).ToList <Article>();
}
catch (Exception ex)
{
logger.Error("VerticalCitationStore::initModelFromRepo() failed to fetch Articles from Pickscomb repository.");
logger.Error(ex.Message);
throw new Exception("VerticalCitationStore::initModelFromRepo() failed to fetch Articles from Pickscomb repository.");
}
msg = String.Format("Retrieved {0} Articles from the repository. Building the initial model . . .", articles.Count());
logger.Info(msg);
uint count = 0;
foreach (Article art in articles)
{
// Test function.
// testArticleURLs(art);
msg = string.Format("Adding #[{0}]", ++count);
logger.Info(msg);
AddCitationsInArticle(art);
//logger.Info("ADDED");
}
logger.Info("Initial CitationStore Model build complete.\n\n");
// testCitationStore();
}
public void rankOnCitationPopularity(uint vIndex, uint numRankedArticlesToExtract)
{
NPreFetched = 0;
dataSetV[vIndex] = false;
Guid vid = VerticalConstants.AllVerticals[vIndex];
IPickscombRepository repo = PickscombRepository.Instance;
/* // Do not fetch all articles from repo; too expensive.
* // Fetch from database Articles going back MaxHoursConsidered ; default MaxHoursConsidered = 5 days * 24
* uint numHours = CitationStore.Instance.CStore[vid].MaxHoursConsidered;
* DateTime start = DateTime.Now.AddHours(-1 * numHours);
* List<Article> articles = repo.GetArticles( vid, start, start.AddHours(numHours)).ToList<Article>();
* */
Guid artid;
List <Guid> articlesWithCitations = new List <Guid>();
uint ranking = 0, inDBCount = 0;
CitationStore.Instance.CStore[vid].setTimeWeightedCitationPopularity();
Dictionary <string, double> rankedURLs = CitationStore.Instance.CStore[vid].getTimeWeightedCitationPopularityScores();
// make sure the URLs are sorted descending according to ranking score
rankedURLs.OrderByDescending(y => y.Value).ToDictionary(y => y.Key, y => y.Value);
foreach (KeyValuePair <string, double> url in rankedURLs)
{
++ranking;
if (CitationStore.RepoArticleUrlGuidIndex[vid].ContainsKey(url.Key))
{
// Article for this URL is in DB; map to Guid
++inDBCount;
artid = CitationStore.RepoArticleUrlGuidIndex[vid][url.Key];
articlesWithCitations.Add(artid);
_allRankedArticleMetrics[vIndex].Add(url.Value, artid);
// Log only the top 100
if (inDBCount < 100)
{
logger.DebugFormat(" Found Article Scored [{0}] in DB with Guid [{1}] Ranking [{2}]",
url.Value, artid, inDBCount);
}
/*
* // Fetch and update in repo scores of only those Articles to be indexed; others scored at 0.0.
* if (inDBCount <= numRankedArticlesToExtract)
* {
* Article art = repo.GetArticle(vid, artid);
* art.Score = url.Value;
* art.PopularityScore = url.Value;
* art.PopularityRanking = ranking;
* repo.Save(art);
* logger.DebugFormat("[#{0}] Updated DB Article Score to [{1:F20}].", inDBCount, art.Score);
* }
*/
}
else //( !articleInDB )
{
_rankedURLNotInDB[vIndex].Add(url.Key, new RankingMetrics(ranking, url.Value));
}
}
logger.Debug("\n");
logger.DebugFormat("~~~ Found [{0}] Cited Articles in DB.", inDBCount);
logger.DebugFormat("~~~ Found [{0}] Total Cited Article URLs.", ranking);
if (ranking > 0)
{
logger.DebugFormat("~~~ [{0}] ({1:F4}%) Citated URL's do not have Articles in the database.\n\n",
ranking - inDBCount, (ranking - inDBCount) * 100 / ranking);
}
/* Don't fetch from repo here. Fetch operation too expensive.
* Now add the Articles in the DB within the time frame but with no citations, to the
* bottom if the list with a lower score than the lowest citation score.
* double noCitationScore = lowestScore / 10;
++ranking;
* foreach (Article art in articles)
* {
* if ( !articlesWithCitations.Contains( art.GetArticleId() ) )
* {
* // Add this article with a score lower than the lowest score.
* // They all have the same ranking and the same lowest score at the bottom.
* art.PopularityRanking = ranking;
* art.PopularityScore = noCitationScore;
* repo.Save(art);
*
* _allRankedArticleMetrics[vIndex].Add(art.PopularityScore, art.GetArticleId());
* if (++inDBCount <= NPreFetched)
* {
* _topNRankedArticles[vIndex].Add(art.PopularityScore, art);
* }
* }
* }
* */
_averagePopularityScore[vIndex] = CitationStore.Instance.CStore[vid].AveragePopularityScore;
dataSetV[vIndex] = true;
}
