/// <summary>
/// Returns the preferred similairy using the provided attribute
/// </summary>
/// <param name="attribute">The target attribute</param>
/// <returns>an instance of the preffered Similarity Measure; otherwise null</returns>
private ISimilarityMeasure GetPrefferedSimilarityMeasure(Data.Attributes.Attribute attribute)
{
// Check if a preffered similarity measure has been set for
// the given attribute
if (string.IsNullOrEmpty(attribute.PreferredSimilarityMeasure))
{
return(null);
}
// Return the MEF maintained instant of the prefered similarity measure
return(GetSimilarityMeasureInstance(attribute.PreferredSimilarityMeasure));
}
/// <summary>
/// Returns the preferred similarity measure using the provided attribute
/// name
/// </summary>
/// <param name="attributeName">The name of the target attribute</param>
/// <returns>an instance of the Preffered Similarity Measure; otherwise null</returns>
private ISimilarityMeasure GetPrefferedSimilarityMeasure(string attributeName)
{
// Retrieve the Attribute instance from the GlobalAttributeCollection
Data.Attributes.Attribute attribute = GlobalAttributeCollection.GetInstance(this.scope).GetAttribute(attributeName);
if (attribute == null)
{
return(null);
}
else
{
return(GetPrefferedSimilarityMeasure(attribute));
}
}
/// <summary>
/// Returns a collection of all Similarity Measures that are valid for
/// the specified attribute
/// </summary>
/// <param name="attributeName">The attribute name to retrieve valid Similarity
/// Measures for</param>
/// <returns>a collection of all Similarity Measures that are valid
/// for the specified attribute</returns>
public ICollection <ISimilarityMeasure> GetValidSimilarityMeasures(string attributeName)
{
// Validate the parameter
if (string.IsNullOrEmpty(attributeName))
{
throw new ArgumentNullException("AttributeName", "No valid attribute name was provided");
}
List <ISimilarityMeasure> validMeasures = new List <ISimilarityMeasure>();
// Retrieve the Attribute instance from the GlobalAttributeCollection
Data.Attributes.Attribute attribute = GlobalAttributeCollection.GetInstance(this.scope).GetAttribute(attributeName);
// Make sure that the SimilairyMeasures collection has values
//if (this.SimilarityMeasures.IsValueCreated)
//{
// Loop over the MEF maintained similarity collection
foreach (ISimilarityMeasure currentMeasure in SimilarityMeasures)
{
SimilarityMeasureBase measure = currentMeasure as SimilarityMeasureBase;
// Check if current measure is valid based on allowed
// semantic types
if (measure.SemanticTypes.HasFlag(attribute.SemanticType))
{
validMeasures.Add(measure);
}
else
{
// If we are here, appropriate semantic types not matching so
// something was setup wrong initially
//TODO: HANDLE THIS
}
}
//}
return(validMeasures);
}
private void GenerateAttributes(Node newNode)
{
// Randomly select a record from the sample data
int recordIndex = rand.Next(0, data.Count - 1);
string name = string.Empty;
foreach (Tuple<string, string> fieldData in this.data[recordIndex])
{
Data.Attributes.Attribute attribute = new Data.Attributes.Attribute(fieldData.Item1);
AttributeValue attributeValue = null;
switch (fieldData.Item1)
{
case ("GivenName"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.Name;
name = fieldData.Item2;
break;
case ("MiddleInitial"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.Name;
name = name + " " + fieldData.Item2;
break;
case ("Surname"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.Name;
name = name + ". " + fieldData.Item2;
attribute = new Data.Attributes.Attribute("Full Name");
attributeValue = new AttributeValue(name, name);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("Gender"):
attribute.SetPreferredSimilarityMeasure(typeof(ExactMatchSimilarityMeasure));
attribute.SemanticType = SemanticType.GeneralString;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("EmailAddress"):
attribute.SetPreferredSimilarityMeasure(typeof(EmailDomainSimilarityMeasure));
attribute.SemanticType = SemanticType.EmailAddress;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("TelephoneNumber"):
attribute.SetPreferredSimilarityMeasure(typeof(LevenshteinDistanceStringSimilarityMeasure));
attribute.SemanticType = SemanticType.Number;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("Birthday"):
attribute.SetPreferredSimilarityMeasure(typeof(DateTimeSimilarityMeasure));
attribute.SemanticType = SemanticType.Date;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("Occupation"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.GeneralString;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
default:
attribute.SetPreferredSimilarityMeasure(typeof(LevenshteinDistanceStringSimilarityMeasure));
attribute.SemanticType = SemanticType.GeneralString;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
}
}
// Make the node's display name the name field
newNode.DisplayValue = name;
newNode.Description = "Generated by internal generator";
// Remove the item we used
this.data.RemoveAt(recordIndex);
}
private void GenerateAttributes(Node newNode)
{
// Randomly select a record from the sample data
int recordIndex = rand.Next(0, data.Count - 1);
string name = string.Empty;
foreach (Tuple <string, string> fieldData in this.data[recordIndex])
{
Data.Attributes.Attribute attribute = new Data.Attributes.Attribute(fieldData.Item1);
AttributeValue attributeValue = null;
switch (fieldData.Item1)
{
case ("GivenName"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.Name;
name = fieldData.Item2;
break;
case ("MiddleInitial"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.Name;
name = name + " " + fieldData.Item2;
break;
case ("Surname"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.Name;
name = name + ". " + fieldData.Item2;
attribute = new Data.Attributes.Attribute("Full Name");
attributeValue = new AttributeValue(name, name);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("Gender"):
attribute.SetPreferredSimilarityMeasure(typeof(ExactMatchSimilarityMeasure));
attribute.SemanticType = SemanticType.GeneralString;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("EmailAddress"):
attribute.SetPreferredSimilarityMeasure(typeof(EmailDomainSimilarityMeasure));
attribute.SemanticType = SemanticType.EmailAddress;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("TelephoneNumber"):
attribute.SetPreferredSimilarityMeasure(typeof(LevenshteinDistanceStringSimilarityMeasure));
attribute.SemanticType = SemanticType.Number;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("Birthday"):
attribute.SetPreferredSimilarityMeasure(typeof(DateTimeSimilarityMeasure));
attribute.SemanticType = SemanticType.Date;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
case ("Occupation"):
attribute.SetPreferredSimilarityMeasure(typeof(DoubleMetaphoneSimilarityMeasure));
attribute.SemanticType = SemanticType.GeneralString;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
default:
attribute.SetPreferredSimilarityMeasure(typeof(LevenshteinDistanceStringSimilarityMeasure));
attribute.SemanticType = SemanticType.GeneralString;
attributeValue = new AttributeValue(fieldData.Item2, fieldData.Item2);
// Add the attribute and it's value to the provided node
UpdateAttributeCollection(newNode, attribute, attributeValue);
break;
}
}
// Make the node's display name the name field
newNode.DisplayValue = name;
newNode.Description = "Generated by internal generator";
// Remove the item we used
this.data.RemoveAt(recordIndex);
}
/// <summary>
/// Returns the similarity measure for the specified Attribute
/// </summary>
/// <param name="attribute">The Attribute</param>
/// <returns>the default similarity measure for the specified attribute</returns>
public ISimilarityMeasure GetDefaultSimilarityMeasure(Data.Attributes.Attribute attribute)
{
return(GetDefaultSimilarityMeasure(attribute.Name));
}
