/// <summary>
/// Initialize the SpamFilter based on the supplied text
/// </summary>
public void Load(TextReader goodReader, TextReader badReader)
{
_good = new Corpus(goodReader);
_bad = new Corpus(badReader);
CalculateProbabilities();
}
/// <summary>
/// Initialize the SpamFilter based on the contents of the supplied Corpuseses
/// </summary>
public void Load(Corpus good, Corpus bad)
{
_good = good;
_bad = bad;
CalculateProbabilities();
}
private void btnLoad_Click(object sender, EventArgs e)
{
Corpus bad = new Corpus();
Corpus good = new Corpus();
bad.LoadFromFile("../../TestData/spam.txt");
good.LoadFromFile("../../TestData/good.txt");
_filter = new SpamFilter();
_filter.Load(good, bad);
txtOut.Text = String.Format("Good Count: {0} \nBad Count: {1} ",
_filter.Good.Tokens.Count,
_filter.Bad.Tokens.Count,
_filter.Prob.Count,
Environment.NewLine);
}
/// <summary>
/// Initialize the SpamFilter based on a DataTable containing columns "IsSpam" and "Body".
/// This is only useful to me the author, but hey, it's my code so I can do what I want!
/// </summary>
/// <param name="table"></param>
public void Load(DataTable table)
{
_good = new Corpus();
_bad = new Corpus();
foreach (DataRow row in table.Rows)
{
bool isSpam = (bool)row["IsSpam"];
string body = row["Body"].ToString();
if (isSpam)
{
_bad.LoadFromReader(new StringReader(body));
}
else
{
_good.LoadFromReader(new StringReader(body));
}
}
CalculateProbabilities();
}
public void Load(DataTable table)
{
_good = new Corpus();
_bad = new Corpus();
foreach (DataRow row in table.Rows)
{
bool isSpam = (bool)row["IsSpam"];
string body = row["Body"].ToString();
if (isSpam) {
_bad.LoadFromReader(new StringReader(body));
}
else {
_good.LoadFromReader(new StringReader(body));
}
}
CalculateProbabilities();
}
