explainability-core-example-opennlp.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Explainability Core example\n",
    "\n",
    "\n",
    "## Dependencies\n",
    "\n",
    "Start by specifying the necessary dependencies:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%loadFromPOM\n",
    "<repository>\n",
    "    <id>jboss-public-repository-group</id>\n",
    "    <name>JBoss Public Repository Group</name>\n",
    "    <url>https://repository.jboss.org/nexus/content/groups/public/</url>\n",
    "</repository>\n",
    "\n",
    "<dependency>\n",
    "    <groupId>org.kie.kogito</groupId>\n",
    "    <artifactId>kogito-bom</artifactId>\n",
    "    <version>1.0.0.Final</version>\n",
    "</dependency>\n",
    "\n",
    "<dependency>\n",
    "    <groupId>org.kie.kogito</groupId>\n",
    "    <artifactId>kogito-apps</artifactId>\n",
    "    <version>1.0.0.Final</version>\n",
    "</dependency>\n",
    "\n",
    "<dependency>\n",
    "    <groupId>org.kie.kogito</groupId>\n",
    "    <artifactId>explainability-core</artifactId>\n",
    "    <version>1.0.0.Final</version>\n",
    "</dependency>\n",
    "\n",
    "<dependency>\n",
    "  <groupId>org.apache.opennlp</groupId>\n",
    "  <artifactId>opennlp-tools</artifactId>\n",
    "  <version>1.9.2</version>\n",
    "</dependency>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import io.github.spencerpark.ijava.IJava;\n",
    "import io.github.spencerpark.jupyter.kernel.magic.common.Shell;\n",
    "IJava.getKernelInstance().getMagics().registerMagics(Shell.class);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "%sh wget https://downloads.apache.org/opennlp/models/langdetect/1.8.3/langdetect-183.bin"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import java.io.InputStream;\n",
    "import java.io.FileInputStream;\n",
    "import opennlp.tools.langdetect.Language;\n",
    "import opennlp.tools.langdetect.LanguageDetector;\n",
    "import opennlp.tools.langdetect.LanguageDetectorME;\n",
    "import opennlp.tools.langdetect.LanguageDetectorModel;\n",
    "\n",
    "InputStream is = new FileInputStream(\"langdetect-183.bin\");\n",
    "LanguageDetectorModel languageDetectorModel = new LanguageDetectorModel(is);\n",
    "LanguageDetector languageDetector = new LanguageDetectorME(languageDetectorModel);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "import java.util.LinkedList;\n",
    "import java.util.List;\n",
    "import java.util.Map;\n",
    "import java.util.Random;\n",
    "import java.util.concurrent.CompletableFuture;\n",
    "\n",
    "import org.kie.kogito.explainability.model.Type;\n",
    "import org.kie.kogito.explainability.model.Value;\n",
    "import org.kie.kogito.explainability.model.Feature;\n",
    "import org.kie.kogito.explainability.model.FeatureFactory;\n",
    "import org.kie.kogito.explainability.model.Output;\n",
    "import org.kie.kogito.explainability.model.PerturbationContext;\n",
    "import org.kie.kogito.explainability.model.Prediction;\n",
    "import org.kie.kogito.explainability.model.PredictionInput;\n",
    "import org.kie.kogito.explainability.model.PredictionOutput;\n",
    "import org.kie.kogito.explainability.model.PredictionProvider;\n",
    "\n",
    "PredictionProvider model = inputs -> CompletableFuture.supplyAsync(() -> {\n",
    "    List<PredictionOutput> results = new LinkedList<>();\n",
    "    for (PredictionInput predictionInput : inputs) {\n",
    "        StringBuilder builder = new StringBuilder();\n",
    "        for (Feature f : predictionInput.getFeatures()) {\n",
    "            if (builder.length() > 0) {\n",
    "                builder.append(' ');\n",
    "            }\n",
    "            builder.append(f.getValue().asString());\n",
    "        }\n",
    "        Language language = languageDetector.predictLanguage(builder.toString());\n",
    "        PredictionOutput predictionOutput = new PredictionOutput(List.of(new Output(\"lang\", Type.TEXT, new Value<>(language.getLang()), language.getConfidence())));\n",
    "        results.add(predictionOutput);\n",
    "    }\n",
    "    return results;\n",
    "});"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Output{value=Value{ita}, type=text, score=0.028884185063368605, name='lang'}]\n"
     ]
    }
   ],
   "source": [
    "String inputText = \"italiani spaghetti pizza mandolino\";\n",
    "List<Feature> features = new LinkedList<>();\n",
    "features.add(FeatureFactory.newFulltextFeature(\"text\", inputText));\n",
    "PredictionInput input = new PredictionInput(features);\n",
    "\n",
    "List<PredictionOutput> predictionOutputs = model.predictAsync(List.of(input)).get();\n",
    "PredictionOutput output = predictionOutputs.get(0);\n",
    "Prediction prediction = new Prediction(input, output);\n",
    "\n",
    "System.out.println(output.getOutputs())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature{name='text', type=text, value=Value{italiani}}: 0.0\n",
      "Feature{name='text', type=text, value=Value{spaghetti}}: 0.09680093109947152\n",
      "Feature{name='text', type=text, value=Value{pizza}}: 0.053478635073901735\n",
      "Feature{name='text', type=text, value=Value{mandolino}}: 0.0\n"
     ]
    }
   ],
   "source": [
    "import java.util.concurrent.TimeUnit;\n",
    "import org.kie.kogito.explainability.Config;\n",
    "import org.kie.kogito.explainability.model.Saliency;\n",
    "import org.kie.kogito.explainability.model.FeatureImportance;\n",
    "import org.kie.kogito.explainability.local.lime.LimeExplainer;\n",
    "\n",
    "LimeExplainer limeExplainer = new LimeExplainer(1, 100);\n",
    "\n",
    "Map<String, Saliency> saliencyMap = limeExplainer.explainAsync(prediction, model)\n",
    "                    .get(Config.INSTANCE.getAsyncTimeout(), Config.INSTANCE.getAsyncTimeUnit());\n",
    "\n",
    "for (FeatureImportance fi : saliencyMap.get(\"lang\").getPerFeatureImportance()) {\n",
    "    System.out.println(fi.getFeature() + \": \" + fi.getScore());\n",
    "}"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Java",
   "language": "java",
   "name": "java"
  },
  "language_info": {
   "codemirror_mode": "java",
   "file_extension": ".jshell",
   "mimetype": "text/x-java-source",
   "name": "Java",
   "pygments_lexer": "java",
   "version": "11.0.3+7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}