/// <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();
}
private void LoadBayesianFilter()
{
mFilter = new SpamFilter();
Corpus bad = new Corpus();
Corpus good = new Corpus();
bad.LoadFromFile(mBadFile);
good.LoadFromFile(mGoodFile);
mFilter.Load(good, bad);
/*
// Just for grins, we'll dump out some statistics about the data we just loaded.
lstResults.Items.Clear();
lstResults.Items.Add(String.Format(@"Bayesian Filter Training Stats: Good:{0} Bad:{1} Prob:{2}"
, mFilter.Good.Tokens.Count
, mFilter.Bad.Tokens.Count
, mFilter.Prob.Count));
// ... and some probabilities for keys
foreach (string key in mFilter.Prob.Keys)
{
if (mFilter.Prob[key] > 0.02)
{
lstResults.Items.Add(String.Format("{0},{1}", mFilter.Prob[key].ToString(".0000"), key));
}
}
*/
}
/// <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();
}
/// <summary>
/// Initialize the SpamFilter based on the supplied text
/// </summary>
/// <param name="goodReader"></param>
/// <param name="badReader"></param>
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>
/// <param name="good"></param>
/// <param name="bad"></param>
public void Load(Corpus good, Corpus bad)
{
_good = good;
_bad = bad;
CalculateProbabilities();
}
