/// <summary> Strips standard prepositions from the given <paramref name="buddyText"/>. </summary>
/// <param name="buddyText">Buddy text to process</param>
/// <returns><paramref name="buddyText"/> without articles</returns>
public virtual string StripPrepositions(string buddyText)
{
// Prepositions to remove
HashSet <string> prepositions = new HashSet <string>(
new[] {
"in", "im", "aus", "ein", "ob", "bis", "auf", "zu", "unter",
"da", "wo"
},
StringComparer.InvariantCultureIgnoreCase
);
string strippedBuddyText = buddyText;
for (WordIterator itr = buddyText.GetWordIterator(); itr.MoveNext();)
{
WordIterator.Word word = itr.Current;
if (word == null)
{
break;
}
// Check if the word is an preposition
if (!prepositions.Contains(word.Text))
{
continue;
}
word.Remove();
// Update return value
strippedBuddyText = itr.GetStringData();
}
return(strippedBuddyText);
}
/// <summary> Strips standard auxiliary verbs from the given <paramref name="buddyText"/>. </summary>
/// <param name="buddyText">Buddy text to process</param>
/// <returns><paramref name="buddyText"/> without auxiliary verbs</returns>
public virtual string StripAuxiliaryVerbs(string buddyText)
{
// Auxiliary verbs to remove
HashSet <string> auxiliaryVerbs = new HashSet <string>(
new[] { "ist" },
StringComparer.InvariantCultureIgnoreCase
);
string strippedBuddyText = buddyText;
for (WordIterator itr = buddyText.GetWordIterator(); itr.MoveNext();)
{
WordIterator.Word word = itr.Current;
if (word == null)
{
break;
}
// Check if the word is an preposition
if (!auxiliaryVerbs.Contains(word.Text))
{
continue;
}
word.Remove();
// Update return value
strippedBuddyText = itr.GetStringData();
}
return(strippedBuddyText);
}
/// <summary> Strips standard articles from the given <paramref name="buddyText"/>. </summary>
/// <param name="buddyText">Buddy text to process</param>
/// <returns><paramref name="buddyText"/> without articles</returns>
public virtual string StripArticles(string buddyText)
{
// Articles to remove
HashSet <string> articles = new HashSet <string>(
new[] { "der", "die", "das", "den", "dem", "ein", "eine" },
StringComparer.InvariantCultureIgnoreCase
);
string strippedBuddyText = buddyText;
for (WordIterator itr = buddyText.GetWordIterator(); itr.MoveNext();)
{
WordIterator.Word word = itr.Current;
if (word == null)
{
break;
}
// Check if the word is an article
if (!articles.Contains(word.Text))
{
continue;
}
word.Remove();
// Update return value
strippedBuddyText = itr.GetStringData();
}
return(strippedBuddyText);
}
/// <summary> Strips typically unneeded substantives from the given <paramref name="buddyText"/>. </summary>
/// <param name="buddyText">Buddy text to process</param>
/// <returns><paramref name="buddyText"/> without substantives</returns>
public virtual string StripSubstantives(string buddyText) // TODO Add unit test
// Substantives to remove
{
HashSet <string> substantives = new HashSet <string>(
new[] {
"Wert", "Button", "Schaltfläche", "Navigation", "Spalte", "Auswahlbox",
"Dokument", "Auftrag", "Bestellung",
// GTUE-related -> Remove all upper-case words?
"Kennzeichen",
"X", "XX", "XXX"
},
StringComparer.CurrentCultureIgnoreCase
);
string strippedBuddyText = buddyText;
for (WordIterator itr = buddyText.GetWordIterator(); itr.MoveNext();)
{
WordIterator.Word word = itr.Current;
if (word == null)
{
break;
}
// Check if the word is an preposition
if (!substantives.Contains(word.Text))
{
continue;
}
word.Remove();
// Update return value
strippedBuddyText = itr.GetStringData();
}
return(strippedBuddyText);
}