BotSharp/BotSharp.NLP/Classify/SentenceFeatureExtractor.cs

25 lines
619 B
C#
Raw Normal View History

2018-09-10 03:56:32 +00:00
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using BotSharp.Algorithm.Features;
using BotSharp.NLP.Tokenize;
namespace BotSharp.NLP.Classify
{
public class SentenceFeatureExtractor : ITextFeatureExtractor
{
public List<Feature> GetFeatures(List<Token> words)
{
var features = new List<Feature>();
2018-09-10 22:25:41 +00:00
words.Where(x => x.IsAlpha)
.Distinct()
2018-09-10 03:56:32 +00:00
.ToList()
2018-09-10 22:25:41 +00:00
.ForEach(w => features.Add(new Feature($"contains {w.Text.ToLower()}", "True")));
2018-09-10 03:56:32 +00:00
return features;
}
}
}