// Construye bloques de último momento (sin detalle, quick), o pequeños, basados en tweets (i.e. no en otros bloques menores).
public DbTimeBlock BuildTimeBlockFromTweets(bool quick, TimeSpan length)
{
IEnumerable <DbTweet> globalToProc = GetTweetsInTimeInterval(DateTime.Now.Add(-length), DateTime.Now, null);
AnalysisResults globalAR = null;
if (quick)
{
globalAR = TA.AnalyzeTweetSet(globalToProc, true);
}
else
{
globalAR = TA.AnalyzeTweetSet(globalToProc, false);
}
Dictionary <DbTopic, AnalysisResults> allTopicsAR = new Dictionary <DbTopic, AnalysisResults>();
foreach (DbTopic t in db.Query <DbTopic>())
{
IEnumerable <DbTweet> topicToProc = GetTweetsInTimeInterval(DateTime.Now.Add(-length), DateTime.Now, t);
AnalysisResults topicAR = null;
if (quick)
{
topicAR = TA.AnalyzeTweetSet(topicToProc, true);
}
else
{
topicAR = TA.AnalyzeTweetSet(topicToProc, false);
}
allTopicsAR.Add(t, topicAR);
}
DbTimeBlock toRet = new DbTimeBlock(DateTime.Now.Add(-length), length, globalAR, allTopicsAR);
db.Store(toRet);
Console.Out.WriteLine("\n\n[ CONSTRUIDO TIME BLOCK, quick: " + quick + "+ ]\n");
return(toRet);
}
// Construye bloques de último momento (sin detalle, quick), o pequeños, basados en tweets (i.e. no en otros bloques menores).
public DbTimeBlock BuildTimeBlockFromTweets(bool quick, TimeSpan length)
{
IEnumerable<DbTweet> globalToProc = GetTweetsInTimeInterval(DateTime.Now.Add(-length), DateTime.Now, null);
AnalysisResults globalAR = null;
if(quick)
globalAR = TA.AnalyzeTweetSet(globalToProc, true);
else
globalAR = TA.AnalyzeTweetSet(globalToProc, false);
Dictionary<DbTopic, AnalysisResults> allTopicsAR = new Dictionary<DbTopic, AnalysisResults>();
foreach (DbTopic t in db.Query<DbTopic>())
{
IEnumerable<DbTweet> topicToProc = GetTweetsInTimeInterval(DateTime.Now.Add(-length), DateTime.Now, t);
AnalysisResults topicAR = null;
if(quick)
topicAR = TA.AnalyzeTweetSet(topicToProc, true);
else
topicAR = TA.AnalyzeTweetSet(topicToProc, false);
allTopicsAR.Add(t, topicAR);
}
DbTimeBlock toRet = new DbTimeBlock(DateTime.Now.Add(-length), length, globalAR, allTopicsAR);
db.Store(toRet);
Console.Out.WriteLine("\n\n[ CONSTRUIDO TIME BLOCK, quick: "+quick+"+ ]\n");
return toRet;
}
// Construye bloques medianos o grandes, basados en bloques chicos o medianos respectivamente.
public void BuildTimeBlockFromBlocks(TimeSpan length)
{
TimeSpan childBlocksLength = TimeSpan.FromSeconds(0);
if (length == TS_medium)
childBlocksLength = TS_short;
else if (length == TS_long)
childBlocksLength = TS_medium;
IEnumerable<DbTimeBlock> ch = from DbTimeBlock t in db
where t.Length == childBlocksLength && t.Used == false
orderby t.Start descending
select t;
int cant = ch.Count();
AnalysisResults GlobalAR = new AnalysisResults();
Dictionary<DbTopic, AnalysisResults> TopicAR = new Dictionary<DbTopic, AnalysisResults>();
foreach (DbTopic t in db.Query<DbTopic>())
TopicAR.Add(t, new AnalysisResults());
if (cant > this.TimeBlockConsolidationThreshold)
{
foreach (DbTimeBlock cht in ch)
{
GlobalAR.Popularity += cht.GlobalAR.Popularity;
GlobalAR.PosVal += cht.GlobalAR.PosVal;
GlobalAR.NegVal += cht.GlobalAR.NegVal;
GlobalAR.Ambiguity += cht.GlobalAR.Ambiguity;
// Para poder comparar las palabras importantes de un subbloque con otro, ya que cada uno fue calculado con tfidf distintos.
double normalizationFactor = 0;
foreach(KeyValuePair<string,double> kv in cht.GlobalAR.relevantList)
normalizationFactor += kv.Value;
foreach(KeyValuePair<string,double> kv in cht.GlobalAR.relevantList)
if(GlobalAR.RelevantTerms.ContainsKey(kv.Key))
GlobalAR.RelevantTerms[kv.Key] += kv.Value/normalizationFactor;
else
GlobalAR.RelevantTerms[kv.Key] = kv.Value/normalizationFactor;
// ídem a lo anterior, para cada topic
foreach (KeyValuePair<DbTopic, AnalysisResults> tar in cht.TopicAR)
{
TopicAR[tar.Key].Popularity += tar.Value.Popularity;
TopicAR[tar.Key].PosVal += tar.Value.PosVal;
TopicAR[tar.Key].NegVal += tar.Value.NegVal;
TopicAR[tar.Key].Ambiguity += tar.Value.Ambiguity;
normalizationFactor = 0;
foreach(KeyValuePair<string,double> kv in cht.TopicAR[tar.Key].relevantList)
normalizationFactor += kv.Value;
foreach(KeyValuePair<string,double> kv in cht.TopicAR[tar.Key].relevantList)
if(TopicAR[tar.Key].RelevantTerms.ContainsKey(kv.Key))
TopicAR[tar.Key].RelevantTerms[kv.Key] += kv.Value/normalizationFactor;
else
TopicAR[tar.Key].RelevantTerms[kv.Key] = kv.Value / normalizationFactor;
}
cht.Used = true;
db.Store(cht);
}
GlobalAR.Ambiguity /= cant;
GlobalAR.DictionaryToList();
foreach (DbTopic t in db.Query<DbTopic>())
{
TopicAR[t].Ambiguity /= cant;
TopicAR[t].DictionaryToList();
}
DbTimeBlock toAdd = new DbTimeBlock(DateTime.Now.Add(-length), length, GlobalAR, TopicAR);
db.Store(toAdd);
Console.Out.WriteLine("\n\n[ CONSTRUIDO TIME BLOCK (desde otros bloques), tamaño: " + length +"+ ]\n");
}
}
// Construye bloques medianos o grandes, basados en bloques chicos o medianos respectivamente.
public void BuildTimeBlockFromBlocks(TimeSpan length)
{
TimeSpan childBlocksLength = TimeSpan.FromSeconds(0);
if (length == TS_medium)
{
childBlocksLength = TS_short;
}
else if (length == TS_long)
{
childBlocksLength = TS_medium;
}
IEnumerable <DbTimeBlock> ch = from DbTimeBlock t in db
where t.Length == childBlocksLength && t.Used == false
orderby t.Start descending
select t;
int cant = ch.Count();
AnalysisResults GlobalAR = new AnalysisResults();
Dictionary <DbTopic, AnalysisResults> TopicAR = new Dictionary <DbTopic, AnalysisResults>();
foreach (DbTopic t in db.Query <DbTopic>())
{
TopicAR.Add(t, new AnalysisResults());
}
if (cant > this.TimeBlockConsolidationThreshold)
{
foreach (DbTimeBlock cht in ch)
{
GlobalAR.Popularity += cht.GlobalAR.Popularity;
GlobalAR.PosVal += cht.GlobalAR.PosVal;
GlobalAR.NegVal += cht.GlobalAR.NegVal;
GlobalAR.Ambiguity += cht.GlobalAR.Ambiguity;
// Para poder comparar las palabras importantes de un subbloque con otro, ya que cada uno fue calculado con tfidf distintos.
double normalizationFactor = 0;
foreach (KeyValuePair <string, double> kv in cht.GlobalAR.relevantList)
{
normalizationFactor += kv.Value;
}
foreach (KeyValuePair <string, double> kv in cht.GlobalAR.relevantList)
{
if (GlobalAR.RelevantTerms.ContainsKey(kv.Key))
{
GlobalAR.RelevantTerms[kv.Key] += kv.Value / normalizationFactor;
}
else
{
GlobalAR.RelevantTerms[kv.Key] = kv.Value / normalizationFactor;
}
}
// ídem a lo anterior, para cada topic
foreach (KeyValuePair <DbTopic, AnalysisResults> tar in cht.TopicAR)
{
TopicAR[tar.Key].Popularity += tar.Value.Popularity;
TopicAR[tar.Key].PosVal += tar.Value.PosVal;
TopicAR[tar.Key].NegVal += tar.Value.NegVal;
TopicAR[tar.Key].Ambiguity += tar.Value.Ambiguity;
normalizationFactor = 0;
foreach (KeyValuePair <string, double> kv in cht.TopicAR[tar.Key].relevantList)
{
normalizationFactor += kv.Value;
}
foreach (KeyValuePair <string, double> kv in cht.TopicAR[tar.Key].relevantList)
{
if (TopicAR[tar.Key].RelevantTerms.ContainsKey(kv.Key))
{
TopicAR[tar.Key].RelevantTerms[kv.Key] += kv.Value / normalizationFactor;
}
else
{
TopicAR[tar.Key].RelevantTerms[kv.Key] = kv.Value / normalizationFactor;
}
}
}
cht.Used = true;
db.Store(cht);
}
GlobalAR.Ambiguity /= cant;
GlobalAR.DictionaryToList();
foreach (DbTopic t in db.Query <DbTopic>())
{
TopicAR[t].Ambiguity /= cant;
TopicAR[t].DictionaryToList();
}
DbTimeBlock toAdd = new DbTimeBlock(DateTime.Now.Add(-length), length, GlobalAR, TopicAR);
db.Store(toAdd);
Console.Out.WriteLine("\n\n[ CONSTRUIDO TIME BLOCK (desde otros bloques), tamaño: " + length + "+ ]\n");
}
}