/// <summary>
/// Gets the web lemma.
/// </summary>
/// <param name="parent">The parent.</param>
/// <param name="node">The node.</param>
/// <param name="options">The options.</param>
/// <returns></returns>
public static webLemmaTerm GetWebLemma(webLemmaTerm parent, freeGraphNodeBase node, lemmaExpansionOptions options)
{
webLemmaTerm term = new webLemmaTerm();
term.name = node.name;
term.nominalForm = node.name;
term.weight = node.weight;
if (options.HasFlag(lemmaExpansionOptions.initialWeightToOne))
{
term.weight = 1;
}
if (options.HasFlag(lemmaExpansionOptions.initialWeightFromParent))
{
term.weight = parent.weight;
}
if (options.HasFlag(lemmaExpansionOptions.weightAsSemanticDistanceFromParent))
{
Double distanceFactor = node.distance.GetRatio(1);
term.weight = term.weight * distanceFactor;
}
return(term);
}
internal void AddOrUpdateNode(freeGraphNodeBase node, freeGraphLink link, freeGraphNodeAndLinks links, Boolean typeToMin, lemmaExpansionOptions options)
{
if (ContainsNode(node.name, true))
{
var nd = GetNode(node.name, true);
if (options.HasFlag(lemmaExpansionOptions.divideWeightByNumberOfSynonims))
{
nd.weight = nd.weight + (node.weight.GetRatio(links.linkedNodeClones.Count));
}
else if (options.HasFlag(lemmaExpansionOptions.weightAsSemanticDistanceThatIsSumOfLinkWeights))
{
nd.weight = nd.weight + node.weight.GetRatio(node.distance);
if (nd.distance > 1)
{
nd.distance = Math.Min(nd.distance, node.distance + link.linkBase.weight);
}
else
{
nd.distance = node.distance + link.linkBase.weight;
}
}
else
{
nd.weight = nd.weight + node.weight;
}
if (typeToMin)
{
nd.type = Math.Min(nd.type, node.type);
}
else
{
nd.type = Math.Max(nd.type, node.type);
}
}
else
{
if (options.HasFlag(lemmaExpansionOptions.divideWeightByNumberOfSynonims))
{
AddNode(node.name, node.weight.GetRatio(links.linkedNodeClones.Count), node.type);
}
else if (options.HasFlag(lemmaExpansionOptions.weightAsSemanticDistanceThatIsSumOfLinkWeights))
{
AddNode(node.name, node.weight.GetRatio(node.distance), node.type).distance = node.distance + link.linkBase.weight;
}
else
{
AddNode(node.name, node.weight, node.type);
}
}
}
/// <summary>
/// Expands the terms.
/// </summary>
/// <param name="lemmas">The lemmas.</param>
/// <param name="expansionSteps">The expansion steps.</param>
/// <param name="options">The options.</param>
/// <returns></returns>
public List <webLemmaTerm> ExpandTerms(IEnumerable <webLemmaTerm> lemmas, Int32 expansionSteps, lemmaExpansionOptions options = lemmaExpansionOptions.initialWeightFromParent | lemmaExpansionOptions.weightAsSemanticDistanceFromParent)
{
Dictionary <String, webLemmaTerm> lemmaDictionary = lemmas.GetLemmaDictionary();
Dictionary <String, webLemmaTerm> outputDictionary = new Dictionary <string, webLemmaTerm>();
Dictionary <String, List <webLemmaTerm> > lemmaSet = new Dictionary <string, List <webLemmaTerm> >();
//aceDictionarySet<String, webLemmaTerm> lemmaSet = new aceDictionarySet<string, webLemmaTerm>();
List <String> terms = lemmaDictionary.Keys.ToList();
List <webLemmaTerm> output = new List <webLemmaTerm>();
foreach (String term in terms)
{
lemmaSet.Add(term, new List <webLemmaTerm>());
lemmaSet[term].Add(lemmaDictionary[term]);
//lemmaSet[term].Add(lemmaDictionary[term]);
}
foreach (String term in terms)
{
webLemmaTerm parent = lemmaDictionary[term];
freeGraphQueryResult termResult = GetLinkedNodes(new String[] { term }, expansionSteps, true, false);
foreach (freeGraphNodeBase node in termResult)
{
webLemmaTerm lem = GetWebLemma(parent, node, options);
if (options.HasFlag(lemmaExpansionOptions.divideWeightByNumberOfSynonims))
{
lem.weight = lem.weight.GetRatio(termResult.Count);
}
if (!lemmaSet.ContainsKey(node.name))
{
lemmaSet.Add(node.name, new List <webLemmaTerm>());
}
lemmaSet[node.name].Add(lem);
}
}
foreach (String key in lemmaSet.Keys)
{
Double weight = 0;
webLemmaTerm firstLemma = null;
foreach (webLemmaTerm lem in lemmaSet[key])
{
weight += lem.weight;
firstLemma = lem;
}
if (!outputDictionary.ContainsKey(key))
{
outputDictionary.Add(key, firstLemma);
}
outputDictionary[key].weight = weight;
}
return(outputDictionary.Values.ToList());
}
/// <summary>
/// Returns expanded cloud from given lemma list - only for matched lemmas
/// </summary>
/// <param name="lemmas">The lemmas.</param>
/// <param name="expansionSteps">The expansion steps.</param>
/// <param name="options">The options.</param>
/// <param name="typeToMin">todo: describe typeToMin parameter on ExpandTermsToCloud</param>
/// <returns></returns>
public lemmaSemanticCloud ExpandTermsToCloud(IEnumerable <String> lemmas, Int32 expansionSteps, Boolean typeToMin = true, lemmaExpansionOptions options = lemmaExpansionOptions.initialWeightFromParent | lemmaExpansionOptions.weightAsSemanticDistanceFromParent)
{
lemmaSemanticCloud output = new lemmaSemanticCloud();
output.name = name + "_subset_exp" + expansionSteps;
output.DisableCheck = true;
StringBuilder sb = new StringBuilder();
sb.Append("Subset expanded from matched query lemmas [");
List <String> nextTerms = new List <string>();
List <String> allTerms = new List <string>();
foreach (String t in lemmas)
{
if (ContainsNode(t))
{
sb.Append(t + " ");
var l = GetNode(t);
output.AddNode(l.name, l.weight, 0).distance = 1;
nextTerms.Add(t);
allTerms.Add(t);
}
}
sb.Append("] using cloud [" + name + "]");
output.description = sb.ToString();
Int32 exp_i = 1;
while (nextTerms.Any())
{
List <String> newNextTerms = new List <string>();
foreach (String t in nextTerms)
{
freeGraphNodeAndLinks links = new freeGraphNodeAndLinks();
if (options.HasFlag(lemmaExpansionOptions.weightAsSemanticDistanceFromParent))
{
links = GetLinks(t, true, false, 1.GetRatio(exp_i), exp_i, true, options.HasFlag(lemmaExpansionOptions.initialWeightFromParent));
}
else if (options.HasFlag(lemmaExpansionOptions.weightAsSemanticDistanceThatIsSumOfLinkWeights))
{
var nd = output.GetNode(t, true);
links = GetLinks(t, true, false, 1, exp_i, true, options.HasFlag(lemmaExpansionOptions.initialWeightFromParent));
}
else
{
links = GetLinks(t, true, false, 1, exp_i, true, options.HasFlag(lemmaExpansionOptions.initialWeightFromParent));
}
foreach (freeGraphLink link in links)
{
if (!allTerms.Contains(link.nodeA.name))
{
newNextTerms.Add(link.nodeA.name);
allTerms.Add(link.nodeA.name);
}
if (link.nodeA.name != t)
{
output.AddOrUpdateNode(link.nodeA, link, links, typeToMin, options);
}
if (!allTerms.Contains(link.nodeB.name))
{
newNextTerms.Add(link.nodeB.name);
allTerms.Add(link.nodeB.name);
}
if (link.nodeB.name != t)
{
output.AddOrUpdateNode(link.nodeB, link, links, typeToMin, options);
}
}
foreach (freeGraphLink link in links)
{
if (!output.ContainsLink(link.linkBase.nodeNameA, link.linkBase.nodeNameB))
{
output.AddLink(link.linkBase.nodeNameA, link.linkBase.nodeNameB, Math.Max(link.linkBase.weight, 1), link.linkBase.type);
//var nd = output.GetNode(link.nodeB.name);
//nd.weight = nd.weight + (link.nodeB.weight.GetRatio(links.linkedNodeClones.Count));
//nd.type = Math.Max(nd.type, link.nodeB.type);
}
else
{
var lnk = output.GetLink(link.linkBase.nodeNameA, link.linkBase.nodeNameB);
lnk.weight += link.linkBase.weight;
if (typeToMin)
{
lnk.type = Math.Min(link.linkBase.type, lnk.type);
}
else
{
lnk.type = Math.Max(link.linkBase.type, lnk.type);
}
}
}
}
nextTerms = newNextTerms;
exp_i++;
if (exp_i > expansionSteps)
{
break;
}
}
output.DisableCheck = false;
output.RebuildIndex();
return(output);
}