static void Main(string[] args)
{
try
{
if (args.Length < 3)
{
OutputHelp();
}
else
{
string tbl_file_name = null, tg3_file_name = null, orange_file_name = null;
if (ParseParams(args, ref m_verbose, ref tbl_file_name, ref tg3_file_name, ref orange_file_name))
{
Verbose("Nalagam tabelo oznak ...\r\n");
MetaTaggerData.LoadAttributes(tbl_file_name);
Verbose("Nalagam učni korpus ...\r\n");
MetaTaggerData.LoadData(tg3_file_name);
Verbose("Pišem datoteko učnih primerov za Orange ...\r\n");
MetaTaggerData.WriteDatasetOrange(orange_file_name, /*null_val=*/ "0");
Verbose("Končano.\r\n");
}
}
}
catch (Exception exception)
{
Console.WriteLine("*** Nepričakovana napaka. Podrobnosti: {0}\r\n{1}", exception, exception.StackTrace);
}
}
static void Main(string[] args)
{
try
{
if (args.Length < 5)
{
OutputHelp();
}
else
{
string tags_1_file_name = null, tags_2_file_name = null, tbl_file_name = null, tree_file_name = null,
output_file_name = null;
if (ParseParams(args, ref m_verbose, ref m_consistency_check, ref tags_1_file_name, ref tags_2_file_name,
ref tbl_file_name, ref tree_file_name, ref output_file_name))
{
Verbose("Nalagam izhod prvega označevalnika ...\r\n");
Corpus corpus_1 = new Corpus();
corpus_1.LoadFromXml(tags_1_file_name, /*tag_len=*/ -1);
Verbose("Nalagam izhod drugega označevalnika ...\r\n");
Corpus corpus_2 = new Corpus();
corpus_2.LoadFromXml(tags_2_file_name, /*tag_len=*/ -1);
if (!m_consistency_check)
{
if (corpus_1.TaggedWords.Count != corpus_2.TaggedWords.Count)
{
Console.WriteLine("*** Napaka! Dolžini vhodnih korpusov se ne ujemata.");
return;
}
}
else
{
Verbose("Preverjam ujemanje besed v vhodnih korpusih ...\r\n");
for (int i = 0; i < corpus_1.TaggedWords.Count; i++)
{
if (corpus_1.TaggedWords[i].Word.ToLower() != corpus_2.TaggedWords[i].Word.ToLower())
{
Console.WriteLine("*** Napaka! Besede v vhodnih korpusih se ne ujemajo.");
return;
}
}
}
Verbose("Nalagam tabelo oznak ...\r\n");
MetaTaggerData.LoadAttributes(tbl_file_name);
Verbose("Nalagam odločitveno drevo ...\r\n");
Tree tree = new Tree(tree_file_name);
Verbose("Označujem besedilo ...\r\n");
MetaTaggerData.LoadTestData(corpus_1, corpus_2);
for (int i = 0; i < MetaTaggerData.Items.Count; i++)
{
if (MetaTaggerData.Items[i].Tag1 != MetaTaggerData.Items[i].Tag2)
{
ArrayList <KeyDat <string, string> > test_example = MetaTaggerData.CreateExample(i);
if (tree.Classify(test_example) != 1)
{
corpus_1.TaggedWords[i].Lemma = MetaTaggerData.Items[i].Lemma2;
corpus_1.TaggedWords[i].Tag = MetaTaggerData.Items[i].Tag2;
}
}
}
Verbose("Pišem izhodno datoteko ...\r\n");
StreamWriter writer = new StreamWriter(output_file_name);
writer.Write(corpus_1.ToString("XML-MI"));
writer.Close();
Verbose("Končano.\r\n");
}
}
}
catch (Exception exception)
{
Console.WriteLine("*** Nepričakovana napaka. Podrobnosti: {0}\r\n{1}", exception, exception.StackTrace);
}
}