public static termExploreModel explore(string word, ILogBuilder response, termExploreMode mode, bool verbose = false)
{
termExploreModel model = new termExploreModel(word);
termExploreModelSet outset = semanticLexiconManager.manager.constructor.loadTermModels(word, true);
if (response != null)
{
response.consoleAltColorToggle();
response.AppendHorizontalLine();
response.AppendLine("Exploring term[" + model.inputForm + "] with [" + mode.ToString() + "]");
response.consoleAltColorToggle();
}
if (Enumerable.Any(outset))
{
model = Enumerable.First(outset);
if (response != null)
{
response.AppendLine("term[" + model.inputForm + "]->lemma[" + model.lemma.inputForm + "]");
}
}
else
{
model.lemmaForm = "";
if (response != null)
{
response.AppendLine("term[" + word + "]->missingLemma[]");
}
}
var output = response;
if (!verbose)
{
response = null;
}
switch (mode)
{
case termExploreMode.apertium_direct:
model = getSynonymsWithApertium(model, response);
break;
case termExploreMode.apertium_wordnet_eng:
model = getSynonymsWithWordnetViaApertium(model, response);
break;
case termExploreMode.apertium_wordnet_srb:
model = getSynonymsWithSerbianWordNetAndApertium(model, response);
break;
case termExploreMode.corpus:
model = getSynonymsByCorpus(model, response);
break;
case termExploreMode.hunspell_srb:
model = getSynonymsWithHunspell(model, response);
break;
case termExploreMode.none:
break;
case termExploreMode.wordnet_srb:
model = getSynonymsWithSerbianWordNet(model, response);
break;
case termExploreMode.unitex:
model = exploreWithUnitex(word, response);
break;
}
model.PostProcess();
if (output != null)
{
model.ToString(output, verbose, false);
}
return(model);
}
/// <summary>
/// Method: word -- translation --- synset ---- other synsets --- collecting all words --- translation --- word
/// </summary>
/// <param name="model">The model.</param>
/// <param name="response">The response.</param>
/// <param name="disableCodePrefixFilter">if set to <c>true</c> [disable code prefix filter].</param>
/// <param name="disableCodeBranchFilter">if set to <c>true</c> [disable code branch filter].</param>
/// <returns></returns>
public static termExploreModel getSynonymsWithWordnetViaApertium(termExploreModel model, ILogBuilder response, bool disableCodePrefixFilter = false, bool disableCodeBranchFilter = false)
{
tokenGraph result = model.graph;
result = languageManagerApertium.manager.queryForGraph(model.lemma.inputForm, apertiumDictQueryScope.exact);
model.translations.AddRange(result.getAllLeafs().getNames());
if (response != null)
{
response.consoleAltColorToggle();
string rst = result.ToStringTreeview();
response.Append(rst);
response.consoleAltColorToggle();
}
languageManagerWordnet.manager.queryWithGraph(result, response, WordnetSource.english, WordnetQueryType.getSymsetCodesByWord);
if (response != null)
{
response.consoleAltColorToggle();
string st = result.ToStringTreeview();
response.Append(st);
response.consoleAltColorToggle();
}
model.wordnetSecondarySymsets.AddRange(result.getAllLeafs().getDeepest().getNames());
if (!disableCodePrefixFilter)
{
string codeStart = model.lemma.gramSet.getPosType().GetWordNetCodeStart().ToString();
Regex codeCriteria = new Regex("^" + codeStart + "");
var badCodes = result.getAllLeafs(codeCriteria, true);
if (response != null)
{
response.AppendHorizontalLine();
response.AppendLine("Reducing to proper codes [" + codeStart + "]->filtered-out[" + badCodes.Count() + "]");
}
badCodes.removeFromParent();
if (response != null)
{
response.consoleAltColorToggle();
string rst = result.ToStringTreeview();
response.Append(rst);
response.consoleAltColorToggle();
}
}
model.wordnetPrimarySymsets.AddRange(result.getAllLeafs().getDeepest().getNames());
languageManagerWordnet.manager.queryWithGraph(result, response, WordnetSource.english, WordnetQueryType.getWordsBySymsetCode);
model.translationRelated.AddRange(result.getAllLeafs().getDeepest().getNames());
if (response != null)
{
response.AppendHorizontalLine();
response.AppendLine("Getting English words by symsetcodes via WordNet");
response.consoleAltColorToggle();
string rst = result.ToStringTreeview();
response.Append(rst);
response.consoleAltColorToggle();
}
languageManagerApertium.manager.queryByGraphNode(result, apertiumDictQueryScope.exact, apertiumDictNeedleSide.translated);
model.wordnetSynonyms.AddRange(result.getAllLeafs().getDeepest().getNames());
if (response != null)
{
response.AppendHorizontalLine();
response.AppendLine("Translating back to Serbian via Apertium");
response.consoleAltColorToggle();
string rst = result.ToStringTreeview();
response.Append(rst);
response.consoleAltColorToggle();
}
if (!disableCodeBranchFilter) // <------ removes the symset nodes that contain none of first-level translation words
{
var codeLevel = result.getAllChildren().getOnLevel(3);
List <IObjectWithPathAndChildren> toTakeOut = new List <IObjectWithPathAndChildren>();
foreach (var clb in codeLevel)
{
foreach (var clb_c in clb)
{
bool takeOut = true;
foreach (var clb_cc in clb_c)
{
if (clb_cc.name == model.lemma.inputForm)
{
takeOut = false;
break;
}
}
if (takeOut)
{
if (response != null)
{
response.AppendLine("-- take out: " + clb.path);
}
toTakeOut.Add(clb);
break;
}
}
}
toTakeOut.removeFromParent();
int wps = Enumerable.Count(model.wordnetSecondarySymsets);
int tr = Enumerable.Count(model.translationRelated);
int ws = Enumerable.Count(model.wordnetSynonyms);
if (response != null)
{
response.AppendLine("----- Branch-node filter ----");
response.AppendLine("Symsets: " + wps);
response.AppendLine("Translations: " + tr);
response.AppendLine("Terms: " + ws);
response.consoleAltColorToggle();
string rst = result.ToStringTreeview();
response.Append(rst);
response.consoleAltColorToggle();
}
model.wordnetPrimarySymsets = result.getAllChildren().getOnLevel(3).getNames(true);
model.translations = result.getAllChildren().getOnLevel(4).getNames(true);
model.synonyms = result.getAllChildren().getOnLevel(5).getNames(true);
wps = wps - Enumerable.Count(model.wordnetPrimarySymsets);
tr = tr - Enumerable.Count(model.translations);
ws = ws - Enumerable.Count(model.synonyms);
if (response != null)
{
//response.AppendLine("----- Branch-node filter ----");
response.AppendLine("Reduction of Symsets: " + wps);
response.AppendLine("Reduction of Translations: " + tr);
response.AppendLine("Reduction of Terms: " + ws);
response.consoleAltColorToggle();
string rst = result.ToStringTreeview();
response.Append(rst);
response.consoleAltColorToggle();
}
}
/*
* String rgex_pat = "^([\\w]*\\\\[\\w]*\\\\[\\w]*\\\\[\\w]*\\\\{0}$)";
*
* Regex rgex = new Regex(String.Format(rgex_pat, model.lemma.inputForm));
*
* var onlyWithLemma = result.getAllLeafs().getFilterOut(rgex);
*/
//languageManagerApertium.manager.queryByGraphNode(result, apertiumDictQueryScope.exact, apertiumDictNeedleSide.english);
model.graph = result;
return(model);
}
/// <summary>
/// Describes the sample group to <c>output</c>
/// </summary>
/// <param name="output">The output object</param>
public void describe(ILogBuilder output = null)
{
if (output == null)
{
return;
}
//output.log();
int tl = output.tabLevel;
output.rootTabLevel();
// output.AppendHeading("SampleGroup description", 1);
output.open("desc", name, "");
// output.AppendHeading(name, 2);
// output.AppendHorizontalLine();
int ci = 1;
foreach (sampleGroupItem item in this)
{
item.parent = this;
output.open("group", item.groupTitle, item.groupDescription);
//output.AppendHeading(, 3);
// output.AppendPair("Description", );
output.AppendPair("ID", ci);
output.AppendPair("Tag", item.groupTag);
if (item.groupSizeLimit == -1)
{
output.AppendPair("Size (ratio)", item.weight + " / " + totalWeight.ToString());
}
else
{
output.AppendPair("Size (limit)", item.groupSizeLimit);
}
output.AppendPair("Count", item.count);
output.AppendPair("Border", item.groupBorder);
if (item.isClosed)
{
output.AppendLine("The group is closed for new members");
}
else
{
output.AppendLine("The group may receive new members");
}
ci++;
output.close();
}
output.AppendHorizontalLine();
output.open("info", "Info", "");
output.AppendPair("Counted collection", countedCollectionName);
output.AppendPair("Items with group tag/s", countHasGroup);
output.AppendPair("Items without group tag", countNoGroup);
output.close();
output.log("-- end");
output.close();
output.tabLevel = tl;
}