예제 #1
0
/*
Compute frequency of all words.
*/
func (ftr *WordsAllFrequency) Compute(dataset tabula.DatasetInterface) {
	allWords := GetAllWordList()

	col := dataset.GetColumnByName("additions")

	for _, rec := range col.Records {
		r := tabula.NewRecordReal(float64(0))

		s := rec.String()
		if len(s) == 0 {
			ftr.PushBack(r)
			continue
		}

		s = clean.WikiText(s)
		if len(s) == 0 {
			ftr.PushBack(r)
			continue
		}

		inWords := tekstus.StringSplitWords(s, true, false)

		freq := tekstus.WordsFrequenciesOf(inWords, allWords, false)

		r.SetFloat(Round(freq))

		ftr.PushBack(r)
	}
}
예제 #2
0
func TestWordsFrequenciesOf(t *testing.T) {
	words := []string{"a", "b", "a", "b", "a", "c"}
	tokens := []string{"a", "b"}
	wordslen := float64(len(words))
	exp := (3.0 / wordslen) + (2.0 / wordslen)

	got := tekstus.WordsFrequenciesOf(words, tokens, false)

	assert(t, exp, got, true)
}
예제 #3
0
/*
Compute the frequency of inserted words.
*/
func (ftr *TermFrequency) Compute(dataset tabula.DatasetInterface) {
	newrevidx := dataset.GetColumnByName("newrevisionid")
	adds := dataset.GetColumnByName("additions")
	recordslen := len(adds.Records)

	for x, rec := range adds.Records {
		r := tabula.NewRecordReal(float64(0))
		// Get inserted words.
		intext := rec.String()

		if len(intext) == 0 {
			ftr.PushBack(r)
			continue
		}

		intext = clean.WikiText(intext)
		inWords := tekstus.StringSplitWords(intext, true, true)

		// Get content of new revision.
		revid := newrevidx.Records[x].String()

		if DEBUG >= 2 {
			fmt.Printf("[feature] term_frequency: %d/%d processing %q\n",
				x, recordslen, revid)
		}

		newtext, e := revision.GetContentClean(revid)
		if e != nil {
			ftr.PushBack(r)
			continue
		}

		newWords := tekstus.StringSplitWords(newtext, true, false)

		freq := tekstus.WordsFrequenciesOf(newWords, inWords, false)

		r.SetFloat(Round(freq))

		ftr.PushBack(r)
	}
}