private void getSynSets_Click(object sender, EventArgs e)
{
synSets.Items.Clear();
semanticRelations.Items.Clear();
lexicalRelations.Items.Clear();
getRelatedSynSets.Enabled = false;
// retrive synsets
Set <SynSet> synSetsToShow = null;
if (synsetID.Text != "")
{
try { synSetsToShow = new Set <SynSet>(new SynSet[] { _wordNetEngine.GetSynSet(synsetID.Text) }); }
catch (Exception)
{
MessageBox.Show("Invalid SynSet ID");
return;
}
}
else
{
// get single most common synset
if (mostCommon.Checked)
{
try
{
SynSet synset = _wordNetEngine.GetMostCommonSynSet(word.Text, (WordNetEngine.POS)pos.SelectedItem);
if (synset != null)
{
synSetsToShow = new Set <SynSet>(new SynSet[] { synset });
}
}
catch (Exception ex) { MessageBox.Show("Error: " + ex); return; }
}
// get all synsets
else
{
try
{
synSetsToShow = _wordNetEngine.GetSynSets(word.Text, (WordNetEngine.POS)pos.SelectedItem);
}
catch (Exception ex) { MessageBox.Show("Error: " + ex); return; }
}
}
if (synSetsToShow.Count > 0)
{
foreach (SynSet synSet in synSetsToShow)
{
synSets.Items.Add(synSet);
}
}
else
{
MessageBox.Show("No synsets found");
}
}
/*--------------------------------------------------------------------------------------------*/
public static bool InsertLexicalsAndSemantics(ISession pSess, WordNetEngine pEngine,
int pStart, int pCount)
{
List <string> keys = SynsetCache.Keys.ToList();
int i = pStart;
for ( ; i < pStart + pCount; ++i)
{
if (i >= keys.Count)
{
break;
}
InsertLexAndSemForSynSet(pSess, keys[i], pEngine.GetSynSet(keys[i]));
}
return(i < keys.Count);
}
// Defualt Constructor
public SemCluster(string DataFolder)
{
try
{
Console.WriteLine("\tSemCluster Text Analytics Tool");
Console.WriteLine("\t------------------------------");
Console.WriteLine("\t-Wikipedia local server couldn't be found!");
Console.WriteLine("\t-Seeds SemAve is in manual mode!");
Console.WriteLine();
Console.WriteLine();
Console.WriteLine("-> Resources loading ...");
Console.WriteLine();
#region Loading External Resources
_wn = new WordNetEngine(DataFolder + "WordNet", InMemoryWordNet);
_tokenizer = new EnglishRuleBasedTokenizer(TokenizeHyphen);
_sentenceDetector = new EnglishMaximumEntropySentenceDetector(DataFolder + "EnglishSD.nbin");
_posTagger = new EnglishMaximumEntropyPosTagger(DataFolder + "EnglishPOS.nbin", DataFolder + "\\Build\\tagdict");
_chunker = new EnglishTreebankChunker(DataFolder + "EnglishChunk.nbin");
#endregion
PlugInsManager(DataFolder);
Console.WriteLine("\tResources loaded successfully");
Console.WriteLine("\t" + PlugInsNumber + " KB plug-ins found in the repository");
Console.WriteLine("\tPress any key to continue ...");
Console.ReadKey();
Console.WriteLine();
RootVirtualNode = _wn.GetSynSet("Noun:1740");
ap = new AffinityPropagationClustering();
SynSetRelationTypes = new WordNetApi.Core.WordNetEngine.SynSetRelation[2];
SynSetRelationTypes[0] = WordNetApi.Core.WordNetEngine.SynSetRelation.Hypernym;
SynSetRelationTypes[1] = WordNetApi.Core.WordNetEngine.SynSetRelation.InstanceHypernym;
}
catch (Exception ex)
{
Dispose();
throw new Exception(ex.Message);
}
}
public static Category FromString(WordNetEngine wordNetEngine, String text)
{
string categoryName;
int wordCount;
Dictionary<SynSet, int> synSetDictionary;
// split into lines
var lines = Regex.Split(text, "\r\n|\r|\n");
// first line "name=.."
var firstLineSplit = lines[0].Split(new char[] {'='}, StringSplitOptions.RemoveEmptyEntries);
if (firstLineSplit.Length != 2)
{
return null;
}
if (firstLineSplit[0] != STRING_REPRESENTATION_FIRST_LINE_KEY)
{
return null;
}
categoryName = firstLineSplit[1];
// second line "wordcount=.."
var secondLineSplit = lines[1].Split(new char[] {'='}, StringSplitOptions.RemoveEmptyEntries);
if (secondLineSplit.Length != 2)
{
return null;
}
if (secondLineSplit[0] != STRING_REPRESENTATION_SECOND_LINE_KEY || !Int32.TryParse( secondLineSplit[1], out wordCount))
{
return null;
}
synSetDictionary = new Dictionary<SynSet,int>();
// rest of the lines - dictionary
for(int i = 2 ; i < lines.Length ; ++i)
{
var lineSplit = lines[i].Split(new char[] {'|'}, StringSplitOptions.RemoveEmptyEntries);
if (lineSplit.Length != 2)
{
break;
}
// synset word count
int synSetWordCount;
if (!Int32.TryParse(lineSplit[1], out synSetWordCount))
{
return null;
}
// synset
var synSet = wordNetEngine.GetSynSet(lineSplit[0]);
if (synSet == null)
{
return null;
}
synSetDictionary.Add(synSet, synSetWordCount);
}
Utility.Log("loaded " + categoryName + " (" + wordCount + ") with " + synSetDictionary.Count + " synsets");
return new Category(categoryName, wordCount, synSetDictionary);
}
public List <SynSet> Query(WordNetEngine _wn, string term)
{
try
{
HttpWebRequest request;
ExternalSet obj;
int SynsCount;
string InternalSynSetId;
string HypernymInfo;
int tmpDepth;
List <SynSet> Syns = new List <SynSet>();
SynSet tmpSyn;
request = WebRequest.Create("http://lookup.dbpedia.org/api/search.asmx/KeywordSearch?QueryString=" + term) as HttpWebRequest;
request.Method = "Get";
request.Accept = "application/json";
using (HttpWebResponse response = request.GetResponse() as HttpWebResponse)
using (StreamReader reader = new StreamReader(response.GetResponseStream()))
obj = JsonConvert.DeserializeObject <ExternalSet>(reader.ReadToEnd());
SynsCount = obj.SynSets.Count;
if (SynsCount > 0)
{
for (int k = 0; k < SynsCount; k++)
{
if (MeasureOverlap(term, obj.SynSets[k].Synonym))
{
// If the Sense has classes, map directly to WordNet
if (obj.SynSets[k].Hypernyms.Count > 0)
{
tmpDepth = 0;
InternalSynSetId = "";
foreach (ExternalHypernym hypernym in obj.SynSets[k].Hypernyms)
{
// if the class RDF Identifier has URL, extract only the concept
if (hypernym.URI.StartsWith("ht"))
{
hypernym.URI = hypernym.URI.Substring(hypernym.URI.LastIndexOf("/") + 1);
}
hypernym.URI = hypernym.URI.Trim().ToLower();
if (SchemaMap.TryGetValue(hypernym.URI, out HypernymInfo))
{
if (tmpDepth < Convert.ToInt32(HypernymInfo[0]))
{
tmpDepth = Convert.ToInt32(HypernymInfo[0]);
InternalSynSetId = HypernymInfo.Substring(1);
}
}
}
if (InternalSynSetId != "")
{
tmpSyn = _wn.GetSynSet("Noun:" + InternalSynSetId);
if (obj.SynSets[k].Gloss != null)
{
tmpSyn.Gloss = rx.Replace(obj.SynSets[k].Gloss, string.Empty);
}
tmpSyn.URI = obj.SynSets[k].ID;
Syns.Add(tmpSyn);
}
}
else
if (obj.SynSets[k].Synonym.Contains("("))
{
List <SynSet> tmpSyns = _wn.GetSynSets(obj.SynSets[k].Synonym.Split('(', ')')[1], "noun");
if (tmpSyns.Count > 0)
{
tmpSyn = tmpSyns.First();
if (obj.SynSets[k].Gloss != null)
{
tmpSyn.Gloss = rx.Replace(obj.SynSets[k].Gloss, string.Empty);
}
tmpSyn.URI = obj.SynSets[k].ID;
Syns.Add(tmpSyn);
}
}
else
{
// Infer a class from categries
}
}
}
}
return(Syns);
}
catch (Exception ex)
{
return(null);
}
}