public static List<ICSM11DataSet> ImportDatasets(FilePath XMLfile)
{
XmlDocument XMLdoc = new XmlDocument();
XMLdoc.Load(XMLfile.Absolute);
string[] s = XMLfile.Absolute.Split(new char[] { '\\' }, StringSplitOptions.RemoveEmptyEntries);
StringBuilder path = new StringBuilder();
for (long i = 0; i < s.Length - 1; i++)
{
path.Append(s[i]);
path.Append("\\");
}
path.Append(XMLdoc.GetElementsByTagName("Datasets")[0].Attributes["Directory"].Value.Replace('/', '\\'));
string RTMsimspath = XMLdoc.GetElementsByTagName("Datasets")[0].Attributes["PrecomputedRTMSimilaritiesDirectory"].Value.Replace('/', '\\');
List<ICSM11DataSet> datasets = new List<ICSM11DataSet>();
XmlNodeList nodelist = XMLdoc.GetElementsByTagName("Dataset");
foreach (XmlNode node in nodelist)
{
ICSM11DataSet dataset = new ICSM11DataSet();
dataset.SourceArtifacts = path + node["Corpus"].Attributes["Directory"].Value + node["Corpus"]["SourceArtifacts"].InnerText
+ "#" + node["Corpus"]["SourceArtifacts"].Attributes["Extension"].Value;
dataset.TargetArtifacts = path + node["Corpus"].Attributes["Directory"].Value + node["Corpus"]["TargetArtifacts"].InnerText
+ "#" + node["Corpus"]["TargetArtifacts"].Attributes["Extension"].Value;
dataset.Oracle = path + node["Corpus"].Attributes["Directory"].Value + node["Corpus"]["Oracle"].InnerText;
dataset.Name = node.Attributes["Name"].Value;
dataset.PrecomputedRTMSimilarities = path + RTMsimspath + node["Settings"]["PrecomputedRTMSimilarities"].InnerText;
dataset.RTM = new RTMSettings
{
Alpha = Convert.ToDouble(node["Settings"]["RTM"]["Alpha"].InnerText),
Beta = Convert.ToDouble(node["Settings"]["RTM"]["Beta"].InnerText),
Eta = Convert.ToDouble(node["Settings"]["RTM"]["Eta"].InnerText),
NumTopics = Convert.ToInt32(node["Settings"]["RTM"]["NumTopics"].InnerText),
};
dataset.PCA = new PCASettings
{
VSM_JS = Convert.ToDouble(node["Settings"]["PCA"]["VSM_JS"].InnerText),
VSM_RTM = Convert.ToDouble(node["Settings"]["PCA"]["VSM_RTM"].InnerText),
JS_RTM = Convert.ToDouble(node["Settings"]["PCA"]["JS_RTM"].InnerText),
};
datasets.Add(dataset);
}
return datasets;
}
private SummaryData ExtractSummaryData(ICSM11DataSet ds, string modelName, DataSetPairsType type)
{
DataSetPairs model = null;
foreach (DataSetPairs dsp in ds.Metrics)
{
if (dsp.Name.Split(new char[] { '.' })[0].Equals(modelName))
{
model = dsp;
}
}
if (model == null)
throw new ArgumentException("Model \"" + modelName + "\" not found.");
switch (type)
{
case DataSetPairsType.Precision:
return CreateSummaryData(model.PrecisionData, modelName);
case DataSetPairsType.Recall:
return CreateSummaryData(model.RecallData, modelName);
case DataSetPairsType.AveragePrecision:
return CreateSummaryData(model.AveragePrecisionData, modelName);
case DataSetPairsType.MeanAveragePrecision:
return CreateSummaryData(model.MeanAveragePrecisionData, modelName);
default:
throw new ArgumentException("Unknown DataSetPairsType");
}
}
