func ImportIndex(s ImportSpec) *pipeline.Corpus {
c := pipeline.NewCorpus()
err := c.UpdateIndex(s.Path, s.Tokenizer)
if err != nil {
panic(err)
}
return s.Run(c)
}
// NewWordLabelCorpusWordFeature creates a LabeledCorpus with word-feature pairs
// as features and document tiles as filepaths for labels.
func NewLabelCorpusWordFeature(c *pipeline.Corpus, z [][]int) *LabeledCorpus {
featured := pipeline.NewCorpus()
for d := 0; d < c.M; d++ {
tokens := make([]string, c.N[d])
for n := 0; n < c.N[d]; n++ {
tokens[n] = fmt.Sprintf("%s-%d", c.Vocab.Tokens[c.W[d][n]], z[d][n])
}
featured.AddDocument(c.Titles[d], tokens)
}
return NewLabeledCorpusWord(featured)
}
// NewLabelCorpusFeature creates a LabeledCorcpus with feature assignments in
// place of word feature, with using document titles as filepaths for labels.
func NewLabeledCorpusFeature(c *pipeline.Corpus, z [][]int) *LabeledCorpus {
featured := pipeline.NewCorpus()
for d, zd := range z {
tokens := make([]string, len(zd))
for n, z := range zd {
tokens[n] = fmt.Sprintf("%d", z)
}
featured.AddDocument(c.Titles[d], tokens)
}
return NewLabeledCorpusWord(featured)
}