MiguelPeralvo
/
spark-nlp-workshop
mirror of https://github.com/MiguelPeralvo/spark-nlp-workshop


  
1

	
2

	
3

	
4

	
5

	
6

	
7

	
8

	
9

	
10

	
11

	
12

	
13

	
14

	
15

	
16

	
17

	
18

	
19

	
20

	
21

	
22

	
23

	
24

	
25

	
26

	
27

	
28

	
29

	
30

	
31

	
32

	
33

	
34

	
35

	
36

	
37

	
38

	
39

	
40

	
41

	
42

	
43

	
44

	
45

	
46

	
47

	
48

	
49

	
50

	
51

	
52

	
53

	
54

	
55

	
56

	
57

	
58

	
59

	
60

	
61

	
62

	
63

	
64

	
65

	
66

	
67

	
68

	
69

	
70

	
71

	
72

	
73

	
74

	
75

	
76

	
77

	
78

	
79

	
80

	
81

	
82

	
83

	
84

	
85

	
86

	
87

	
88

	
89

	
90

	
91

	
92

	
93

	
94

	
95

	
96

	
97

	
98

	
99

	
100

	
101

	
102

	
103

	
104

	
105

	
106

	
107

	
108

	
109

	
110

	
111

	
112

	
113

	
114

	
115

	
116

	
117

	
118

	
119

	
120

	
121

	
122

	
123

	
124

	
125

	
126

	
127

	
128

	
129

	
130

	
131

	
132

	
133

	
134

	
135

	
136

	
137

	
138

	
139

	
140

	
141

	
142

	
143

	
144

	
145

	
146

	
147

	
148

	
149

	
150

	
151

	
152

	
153

	
154

	
155

	
156

	
157

	
158

	
159

	
160

	
161

	
162

	
163

	
164

	
165

	
166

	
167

	
168

	
169

	
170

	
171

	
172

	
173

	
174

	
175

	
176

	
177

	
178

	
179

	
180

	
181

	
182

	
183

	
184

	
185

	
186

	
187

	
188

	
189

	
190

	
191

	
192

	
193

	
194

	
195

	
196

	
197

	
198

	
199

	
200

	
201

	
202

	
203

	
204

	
205

	
206

	
207

	
208

	
209

	
210

	
211

	
212

	
213

	
214

	
215

	
216

	
217

	
218

	
219

	
220

	
221

	
222

	
223

	
224

	
225

	
226

	
227

	
228

	
229

	
230

	
231

	
232

	
233

	
234

	
235

	
236

	
237

	
238

	
239

	
240

	
241

	
            import com.johnsnowlabs.nlp.annotator._
import com.johnsnowlabs.nlp.annotators.ner.NerConverter
import com.johnsnowlabs.nlp.base._
import com.johnsnowlabs.nlp.embeddings.WordEmbeddingsFormat
import com.johnsnowlabs.nlp.util.io.ResourceHelper
import com.johnsnowlabs.util.Benchmark

object NerCrfTraining extends App {

  import ResourceHelper.spark.implicits._

  val documentAssembler = new DocumentAssembler().
    setInputCol("text").
    setOutputCol("document")

  val tokenizer = new Tokenizer().
    setInputCols(Array("document")).
    setOutputCol("token")

  val pos = PerceptronModel.pretrained().
    setInputCols("document", "token").
    setOutputCol("pos")

  val ner = new NerCrfApproach().
    setInputCols("document", "token", "pos").
    setOutputCol("ner").
    setLabelColumn("label").
    setOutputCol("ner").
    setMinEpochs(1).
    setMaxEpochs(5).
    setEmbeddingsSource("data/embeddings/glove.6B.100d.txt", 100, WordEmbeddingsFormat.TEXT).
    setExternalFeatures("data/ner/dict.txt", ",").
    setExternalDataset("data/ner/eng.train", "SPARK_DATASET").
    setC0(1250000).
    setRandomSeed(0).
    setVerbose(2)

  val finisher = new Finisher().
    setInputCols("ner")

  val recursivePipeline = new RecursivePipeline().
    setStages(Array(
      documentAssembler,
      tokenizer,
      pos,
      ner,
      finisher
    ))

  val nermodel = recursivePipeline.fit(Seq.empty[String].toDF("text"))
  val nerlpmodel = new LightPipeline(nermodel)

  val res = Benchmark.time("Light annotate NerCRF") {
    nerlpmodel.annotate("Peter is a very good person from Germany, he is working at IBM.")
  }

  println(res.mapValues(_.mkString(", ")).mkString(", "))

}

object NerDLTraining extends App {

  ResourceHelper.spark

  import ResourceHelper.spark.implicits._

  val documentAssembler = new DocumentAssembler().
    setInputCol("text").
    setOutputCol("document")

  val tokenizer = new Tokenizer().
    setInputCols(Array("document")).
    setOutputCol("token")

  val ner = new NerDLApproach().
    setInputCols("document", "token").
    setOutputCol("ner").
    setLabelColumn("label").
    setOutputCol("ner").
    setMinEpochs(1).
    setMaxEpochs(30).
    setEmbeddingsSource("data/embeddings/glove.6B.100d.txt", 100, WordEmbeddingsFormat.TEXT).
    setExternalDataset("data/ner/eng.train", "SPARK_DATASET").
    setRandomSeed(0).
    setVerbose(2).
    setDropout(0.8f).
    setBatchSize(18)

  val finisher = new Finisher().
    setInputCols("ner")

  val recursivePipeline = new RecursivePipeline().
    setStages(Array(
      documentAssembler,
      tokenizer,
      ner,
      finisher
    ))

  val nermodel = recursivePipeline.fit(Seq.empty[String].toDF("text"))
  val nerlpmodel = new LightPipeline(nermodel)

  val res = Benchmark.time("Light annotate NerDL") {
    nerlpmodel.annotate("Peter is a very good person from Germany, he is working at IBM.")
  }

  println(res.mapValues(_.mkString(", ")).mkString(", "))

  nermodel.stages(2).asInstanceOf[NerDLModel].write.overwrite().save("./models/nerdl-deid-30")

}

object NerDLPretrained extends App {

  ResourceHelper.spark

  import ResourceHelper.spark.implicits._

  val documentAssembler = new DocumentAssembler().
    setInputCol("text").
    setOutputCol("document")

  val sentenceDetector = new SentenceDetector()
    .setInputCols("document")
    .setOutputCol("sentence")
    .setUseAbbreviations(false)

  val tokenizer = new Tokenizer().
    setInputCols(Array("sentence")).
    setOutputCol("token")

  val ner = NerDLModel.pretrained().
    setInputCols("sentence", "token").
    setOutputCol("ner")

  val converter = new NerConverter()
    .setInputCols("sentence", "token", "ner")
    .setOutputCol("nerconverter")

  val finisher = new Finisher().
    setInputCols("token", "sentence", "nerconverter", "ner")

  val recursivePipeline = new RecursivePipeline().
    setStages(Array(
      documentAssembler,
      sentenceDetector,
      tokenizer,
      ner,
      converter,
      finisher
    ))

  val nermodel = recursivePipeline.fit(Seq.empty[String].toDF("text"))
  val nerlpmodel = new LightPipeline(nermodel)

  val res1 = Benchmark.time("Light annotate NerDL") {
    nerlpmodel.fullAnnotate("Peter is a very good person from Germany, he is working at IBM.")
  }
  val res2 = Benchmark.time("Light annotate NerDL") {
    nerlpmodel.fullAnnotate("I saw the patient with Dr. Andrew Newhouse.")
  }
  val res3 = Benchmark.time("Light annotate NerDL") {
    nerlpmodel.fullAnnotate("Ms. Louise Iles is a 70 yearold")
  }
  val res4 = Benchmark.time("Light annotate NerDL") {
    nerlpmodel.fullAnnotate("Ms.")
  }

  println(res1.mapValues(_.mkString(", ")).mkString(", "))
  println(res2.mapValues(_.mkString(", ")).mkString(", "))
  println(res3.mapValues(_.mkString(", ")).mkString(", "))
  println(res4.mapValues(_.mkString(", ")).mkString(", "))

}

object NerCrfPretrained extends App {

  import ResourceHelper.spark.implicits._

  val documentAssembler = new DocumentAssembler().
    setInputCol("text").
    setOutputCol("document")

  val sentenceDetector = new SentenceDetector()
    .setInputCols("document")
    .setOutputCol("sentence")
    .setUseAbbreviations(false)

  val tokenizer = new Tokenizer().
    setInputCols(Array("sentence")).
    setOutputCol("token")

  val pos = PerceptronModel.pretrained().
    setInputCols("document", "token").
    setOutputCol("pos")

  val ner = NerCrfModel.pretrained().
    setInputCols("sentence", "token", "pos").
    setOutputCol("ner")

  val converter = new NerConverter()
    .setInputCols("sentence", "token", "ner")
    .setOutputCol("nerconverter")

  val finisher = new Finisher().
    setInputCols("token", "sentence", "nerconverter", "ner")

  val recursivePipeline = new RecursivePipeline().
    setStages(Array(
      documentAssembler,
      sentenceDetector,
      tokenizer,
      pos,
      ner,
      converter,
      finisher
    ))

  val nermodel = recursivePipeline.fit(Seq.empty[String].toDF("text"))
  val nerlpmodel = new LightPipeline(nermodel)

  val res1 = Benchmark.time("Light annotate NerCrf") {
    nerlpmodel.fullAnnotate("Peter is a very good person from Germany, he is working at IBM.")
  }
  val res2 = Benchmark.time("Light annotate NerCrf") {
    nerlpmodel.fullAnnotate("I saw the patient with Dr. Andrew Newhouse.")
  }
  val res3 = Benchmark.time("Light annotate NerCrf") {
    nerlpmodel.fullAnnotate("Ms. Louise Iles is a 70yearold")
  }
  val res4 = Benchmark.time("Light annotate NerCrf") {
    nerlpmodel.fullAnnotate("Ms.")
  }

  println(res1.mapValues(_.mkString(", ")).mkString(", "))
  println(res2.mapValues(_.mkString(", ")).mkString(", "))
  println(res3.mapValues(_.mkString(", ")).mkString(", "))
  println(res4.mapValues(_.mkString(", ")).mkString(", "))

}