/*
Compute the longest word in inserted text.
*/
func (ftr *LongestWord) Compute(dataset tabula.DatasetInterface) {
adds := dataset.GetColumnByName("additions")
addslen := adds.Len()
for x, rec := range adds.Records {
text := rec.String()
textlen := len(text)
if textlen == 0 {
ftr.PushBack(tabula.NewRecordInt(int64(0)))
continue
}
text = clean.WikiText(text)
inWords := tekstus.StringSplitWords(text, true, true)
slong, _ := tekstus.WordsFindLongest(inWords)
if DEBUG >= 2 {
fmt.Printf("[feature] %d/%d longest word: %q\n", x, addslen,
slong)
}
slonglen := int64(len(slong))
ftr.PushBack(tabula.NewRecordInt(slonglen))
}
}
/*
Compute number of good token in inserted text.
*/
func (ftr *GoodToken) Compute(dataset tabula.DatasetInterface) {
col := dataset.GetColumnByName("additions")
for _, rec := range col.Records {
cnt := tekstus.StringCountTokens(rec.String(), tokens, false)
ftr.PushBack(tabula.NewRecordInt(int64(cnt)))
}
}
/*
Compute maximum sequence of character at inserted text.
*/
func (ftr *LongestCharSeq) Compute(dataset tabula.DatasetInterface) {
col := dataset.GetColumnByName("additions")
for _, rec := range col.Records {
text := rec.String()
_, v := tekstus.GetMaxCharSequence(text)
ftr.PushBack(tabula.NewRecordInt(int64(v)))
}
}
// Compute will count number of bytes that is used in comment, NOT including
// the header content "/* ... */".
func (ftr *CommentLength) Compute(dataset tabula.DatasetInterface) {
col := dataset.GetColumnByName("editcomment")
leftcap := []byte("/*")
rightcap := []byte("*/")
for _, rec := range col.Records {
cmt := rec.Bytes()
cmt, _ = tekstus.BytesRemoveUntil(cmt, leftcap, rightcap)
ftr.PushBack(tabula.NewRecordInt(int64(len(cmt))))
}
}
/*
Compute change the classification from text to numeric. The "regular" edit
will become 0 and the "vandalism" will become 1.
*/
func (ftr *Class) Compute(dataset tabula.DatasetInterface) {
col := dataset.GetColumnByName("class")
for _, rec := range col.Records {
r := tabula.NewRecordInt(0)
if rec.String() == "vandalism" {
r.SetInteger(1)
}
ftr.PushBack(r)
}
}
//
// ToRow will convert the stat to tabula.row in the order of Stat field.
//
func (stat *Stat) ToRow() (row *tabula.Row) {
row = &tabula.Row{}
row.PushBack(tabula.NewRecordInt(stat.ID))
row.PushBack(tabula.NewRecordInt(stat.StartTime))
row.PushBack(tabula.NewRecordInt(stat.EndTime))
row.PushBack(tabula.NewRecordInt(stat.ElapsedTime))
row.PushBack(tabula.NewRecordReal(stat.OobError))
row.PushBack(tabula.NewRecordReal(stat.OobErrorMean))
row.PushBack(tabula.NewRecordInt(stat.TP))
row.PushBack(tabula.NewRecordInt(stat.FP))
row.PushBack(tabula.NewRecordInt(stat.TN))
row.PushBack(tabula.NewRecordInt(stat.FN))
row.PushBack(tabula.NewRecordReal(stat.TPRate))
row.PushBack(tabula.NewRecordReal(stat.FPRate))
row.PushBack(tabula.NewRecordReal(stat.TNRate))
row.PushBack(tabula.NewRecordReal(stat.Precision))
row.PushBack(tabula.NewRecordReal(stat.FMeasure))
row.PushBack(tabula.NewRecordReal(stat.Accuracy))
row.PushBack(tabula.NewRecordReal(stat.AUC))
return
}