utterances_items_intent.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c3b00d1d",
   "metadata": {},
   "source": [
    "# Utterance Generation for ItemsIntent"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "43d0741f",
   "metadata": {},
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "13604cb0",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import random\n",
    "import re\n",
    "\n",
    "import nltk\n",
    "from nltk import CFG\n",
    "from nltk.parse.generate import generate\n",
    "\n",
    "random.seed(123)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cbbb5c25",
   "metadata": {},
   "source": [
    "### Items Dictionary from Items Intent\n",
    "Including\n",
    "* item name\n",
    "* slot name\n",
    "* location details\n",
    "* neighboring items\n",
    "* content items (optional)\n",
    "* adjectives\n",
    "* hyperonyms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "4eb906d1",
   "metadata": {},
   "outputs": [],
   "source": [
    "items_dict = {\n",
    "    \"astronautinnen\": {\n",
    "        \"slot\": \"astronautinnen\",\n",
    "        \"location\": [\"links\", \"oben\", \"in der ecke\"],\n",
    "        \"neighbors\": [\"ufo\", \"raumschiff\"],\n",
    "        \"content\": [\"anzügen\"],\n",
    "        \"adjectives\": [\"3\", \"weiß\"],\n",
    "\t\t\"hyperonyms\": [\"menschen\", \"personen\", \"frauen\"],\n",
    "    },\n",
    "    \"ballon\": {\n",
    "        \"slot\": \"ballon\",\n",
    "        \"location\": [\n",
    "            \"rechts\",\n",
    "            \"unten\",\n",
    "            \"in der ecke\",\n",
    "            \"am rand\",\n",
    "            \"auf der rechten seite\",\n",
    "            \"über\",\n",
    "        ],\n",
    "        \"neighbors\": [\"boot\", \"schiff\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"rot\", \"rund\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\", \"spielzeug\"],\n",
    "    },\n",
    "    \"biene\": {\n",
    "        \"slot\": \"biene\",\n",
    "        \"location\": [\"rechts\", \"am rand\", \"auf der rechten seite\"],\n",
    "        \"neighbors\": [\"zahnrad\", \"treppe\"],\n",
    "        \"content\": [\"streifen\"],\n",
    "        \"adjectives\": [\"gelb\", \"gestreift\"],\n",
    "\t\t\"hyperonyms\": [\"tier\", \"insekt\", \"lebewesen\"],\n",
    "    },\n",
    "    \"boot\": {\n",
    "        \"slot\": \"boot\",\n",
    "        \"location\": [\n",
    "            \"rechts\",\n",
    "            \"unten\",\n",
    "            \"in der ecke\",\n",
    "            \"am rand\",\n",
    "            \"auf der rechten seite\",\n",
    "            \"unter\",\n",
    "            \"auf\",\n",
    "        ],\n",
    "        \"neighbors\": [\"ballon\", \"wasser\", \"fluss\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"grün\", \"lächelnd\", \"schwimmend\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\", \"fahrzeug\"],\n",
    "    },\n",
    "    \"buchLinksOben\": {\n",
    "        \"slot\": \"buch\",\n",
    "        \"location\": [\"links\", \"oben\", \"in der ecke\"],\n",
    "        \"neighbors\": [\"buch\", \"uhr\", \"säulen\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"grün\", \"blau\", \"geschlossen\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"buchMitteUnten\": {\n",
    "        \"slot\": \"buch\",\n",
    "        \"location\": [\"in der mitte\", \"unten\", \"am rand\"],\n",
    "        \"neighbors\": [\"pfeilen\", \"tisch\", \"kurven\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"grün\", \"geschlossen\", \"klein\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"buchLinksObenOffen\": {\n",
    "        \"slot\": \"buch\",\n",
    "        \"location\": [\"links\", \"oben\", \"in der ecke\"],\n",
    "        \"neighbors\": [\"buch\", \"uhr\", \"säulen\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"aufgeschlagen\", \"offen\", \"geöffnet\", \"klein\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"buchRechtsMitte\": {\n",
    "        \"slot\": \"buch\",\n",
    "        \"location\": [\"rechts\", \"auf der rechten seite\", \"unter\"],\n",
    "        \"neighbors\": [\"globus\", \"spinne\", \"rauch\", \"mikroskop\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"aufgeschlagen\", \"offen\", \"geöffnet\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"diagramm\": {\n",
    "        \"slot\": \"diagramm\",\n",
    "        \"location\": [\"rechts\", \"oben\", \"am rand\", \"über\", \"neben\", \"an der wand\"],\n",
    "        \"neighbors\": [\"mikroskop\", \"uhr\"],\n",
    "        \"content\": [\"x\", \"y\", \"sigma\", \"summe\", \"summenzeichen\", \"normalverteilung\", \"gaußkurve\", \"mathe\", \"mathematik\", \"achse\"],\n",
    "        \"adjectives\": [\"bunt\", \"schwarz\", \"mathematisch\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"europa\": {\n",
    "        \"slot\": \"europa\",\n",
    "        \"location\": [\n",
    "            \"unten\",\n",
    "            \"links\",\n",
    "            \"auf der linken seite\",\n",
    "            \"über\",\n",
    "            \"unter\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\"tisch\", \"schultasche\"],\n",
    "        \"content\": [\"sterne\", \"sternen\"],\n",
    "        \"adjectives\": [\"blau\", \"gelb\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"fff\": {\n",
    "        \"slot\": \"kinder\",\n",
    "        \"location\": [\"oben\", \"links\", \"hinter\", \"unter\"],\n",
    "        \"neighbors\": [\"terminal\", \"konsole\", \"astronautinnen\", \"wolke\"],\n",
    "        \"content\": [\"fahne\"],\n",
    "        \"adjectives\": [\"grün\", \"beid\"],\n",
    "\t\t\"hyperonyms\": [\"menschen\", \"personen\", \"kinder\"],\n",
    "    },\n",
    "    \"gleichung\": {\n",
    "        \"slot\": \"gleichung\",\n",
    "        \"location\": [\"unten\", \"rechts\"],\n",
    "        \"neighbors\": [\"stift\", \"kolben\", \"chemie\", \"labor\"],\n",
    "        \"content\": [\"x\", \"y\", \"Gleichheitszeichen\"],\n",
    "        \"adjectives\": [\"weiß\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"globus\": {\n",
    "        \"slot\": \"globus\",\n",
    "        \"location\": [\"mittig\", \"zentral\", \"in der mitte\", \"im zentrum\"],\n",
    "        \"neighbors\": [\"wolke\", \"treppe\"],\n",
    "        \"content\": [\"sockel\"],\n",
    "        \"adjectives\": [\"rund\", \"blau\", \"grün\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"gluehbirneObenMitteGross\": {\n",
    "        \"slot\": \"glühbirne\",\n",
    "        \"location\": [\"oben\", \"im oberen teil\", \"in der mitte\", \"neben\"],\n",
    "        \"neighbors\": [\"glühbirne\", \"teleskop\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"gelb\", \"groß\", \"rund\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"gluehbirneObenMitteKlein\": {\n",
    "        \"slot\": \"glühbirne\",\n",
    "        \"location\": [\"oben\", \"im oberen teil\", \"in der mitte\", \"neben\"],\n",
    "        \"neighbors\": [\"glühbirne\", \"globus\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"gelb\", \"klein\", \"rund\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"kurven\": {\n",
    "        \"slot\": \"kurven\",\n",
    "        \"location\": [\"unten\", \"links\", \"mittig\", \"links von der mitte\", \"in der mitte\"],\n",
    "        \"neighbors\": [\"buch\", \"lineal\", \"tafel\", \"wolken\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"weiß\", \"bunt\", \"gelb\", \"rosa\", \"pink\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"lab\": {\n",
    "        \"slot\": \"labor\",\n",
    "        \"location\": [\"unten\", \"rechts\", \"am rand\", \"neben\"],\n",
    "        \"neighbors\": [\"wasser\", \"boot\", \"ballon\"],\n",
    "        \"content\": [\"terminal\", \"diy\"],\n",
    "        \"adjectives\": [\"eckig\", \"bunt\", \"rot\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"mikroskop\": {\n",
    "        \"slot\": \"mikroskop\",\n",
    "        \"location\": [\"oben\", \"rechts\", \"neben\", \"unter\"],\n",
    "        \"neighbors\": [\"diagramm\", \"buch\", \"tisch\", \"pflanze\", \"kugeln\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"grau\", \"grün\", \"petrolfarben\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"personLinksMitte\": {\n",
    "        \"slot\": \"person\",\n",
    "        \"location\": [\"links\", \"in der mitte\", \"neben\"],\n",
    "        \"neighbors\": [\"taube\", \"vogel\", \"wolke\", \"stift\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"sitzend\", \"kauernd\", \"blond\", \"lila\"],\n",
    "\t\t\"hyperonyms\": [\"mensch\", \"person\"],\n",
    "    },\n",
    "    \"personMitteMitte\": {\n",
    "        \"slot\": \"person\",\n",
    "        \"location\": [\"in der mitte\", \"im zentrum\", \"neben\"],\n",
    "        \"neighbors\": [\"terminal\", \"konsole\", \"ordner\", \"kästen\"],\n",
    "        \"content\": [\"kittel\", \"brille\", \"klemmbrett\"],\n",
    "        \"adjectives\": [\"weiß\", \"grau\"],\n",
    "\t\t\"hyperonyms\": [\"mensch\", \"person\"],\n",
    "    },\n",
    "    \"personRechtsMitte\": {\n",
    "        \"slot\": \"person\",\n",
    "        \"location\": [\"in der mitte\", \"rechts\", \"neben\"],\n",
    "        \"neighbors\": [\"labor\", \"magnet\", \"kolben\", \"sprechblase\"],\n",
    "        \"content\": [\"kittel\", \"brille\"],\n",
    "        \"adjectives\": [\"weiß\", \"blond\", \"gelb\"],\n",
    "\t\t\"hyperonyms\": [\"mensch\", \"person\"],\n",
    "    },\n",
    "    \"planet\": {\n",
    "        \"slot\": \"planet\",\n",
    "        \"location\": [\n",
    "            \"oben\",\n",
    "            \"links\",\n",
    "            \"auf der linken seite\",\n",
    "            \"am Rand\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\"buch\", \"astronautinnen\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"gelb\", \"golden\", \"rund\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"raumschiff\": {\n",
    "        \"slot\": \"raumschiff\",\n",
    "        \"location\": [\n",
    "            \"oben\",\n",
    "            \"links\",\n",
    "            \"auf der linken seite\",\n",
    "            \"am Rand\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\"astronautinnen\", \"teleskop\", \"glühbirne\", \"terminal\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"rot\", \"eckig\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\", \"fahrzeug\", \"flugobjekt\"],\n",
    "    },\n",
    "    \"spielKonsole\": {\n",
    "        \"slot\": \"spielkonsole\",\n",
    "        \"location\": [\"in der mitte\", \"im zentrum\", \"neben\"],\n",
    "        \"neighbors\": [\"glühbirne\", \"terminal\", \"astronautinnen\", \"würfel\"],\n",
    "        \"content\": [\"tasten\", \"knöpfen\"],\n",
    "        \"adjectives\": [\"eckig\", \"schwarz\", \"grau\", \"flach\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"stiftLinksMitte\": {\n",
    "        \"slot\": \"stift\",\n",
    "        \"location\": [\n",
    "            \"mittig\",\n",
    "            \"links\",\n",
    "            \"auf der linken seite\",\n",
    "            \"am Rand\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\"person\", \"würfel\", \"tafel\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"grün\", \"spitz\", \"länglich\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\", \"schreibutensil\"],\n",
    "    },\n",
    "    \"stiftRechtsUnten\": {\n",
    "        \"slot\": \"stift\",\n",
    "        \"location\": [\n",
    "            \"unten\",\n",
    "            \"rechts\",\n",
    "            \"auf der rechten seite\",\n",
    "            \"am Rand\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\"labor\", \"ballon\", \"boot\", \"gleichung\", \"konsole\", \"terminal\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"rot\", \"spitz\", \"länglich\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\", \"schreibutensil\"],\n",
    "    },\n",
    "    \"tafel\": {\n",
    "        \"slot\": \"tafel\",\n",
    "        \"location\": [\n",
    "            \"unten\",\n",
    "            \"links\",\n",
    "            \"auf der linken seite\",\n",
    "            \"am Rand\",\n",
    "            \"neben\",\n",
    "            \"bei\",\n",
    "        ],\n",
    "        \"neighbors\": [\"würfel\", \"stift\", \"buch\", \"kurven\", \"europa\"],\n",
    "        \"content\": [\"zwei\", \"vier\", \"komma\", \"punkt\"],\n",
    "        \"adjectives\": [\"grün\", \"beschriftet\", \"beschrieben\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"teleskop\": {\n",
    "        \"slot\": \"teleskop\",\n",
    "        \"location\": [\"am Rand\", \"mittig\", \"neben\", \"neben\"],\n",
    "        \"neighbors\": [\"glühbirne\", \"raumschiff\"],\n",
    "        \"content\": [],\n",
    "        \"adjectives\": [\"rot\", \"rund\", \"mikroskopähnlich\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"terminalMitteMitte\": {\n",
    "        \"slot\": \"terminal\",\n",
    "        \"location\": [\n",
    "            \"in der mitte\",\n",
    "            \"im zentrum\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\n",
    "            \"globus\",\n",
    "            \"spielkonsole\",\n",
    "            \"glühbirne\",\n",
    "            \"astronautinnen\",\n",
    "            \"würfel\",  \n",
    "        ],\n",
    "        \"content\": [\"schrift\", \"hello\", \"world\"],\n",
    "        \"adjectives\": [\"schwarz\", \"beschriftet\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"uhrLinksOben\": {\n",
    "        \"slot\": \"uhr\",\n",
    "        \"location\": [\"oben\", \"links\", \"auf der linken seite\", \"neben\"],\n",
    "        \"neighbors\": [\"buch\", \"astronautinnen\", \"person\", \"planet\"],\n",
    "        \"content\": [\"zeit\", \"zeigern\"],\n",
    "        \"adjectives\": [\"gelb\", \"rund\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"uhrRechtsOben\": {\n",
    "        \"slot\": \"uhr\",\n",
    "        \"location\": [\n",
    "            \"oben\",\n",
    "            \"rechts\",\n",
    "            \"auf der rechten seite\",\n",
    "            \"am Rand\",\n",
    "            \"an der Wand\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\"diagramm\", \"mikroskop\"],\n",
    "        \"content\": [\"zeit\", \"zeigern\"],\n",
    "        \"adjectives\": [\"blau\", \"rund\", \"grün\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "    \"wuerfel\": {\n",
    "        \"slot\": \"würfel\",\n",
    "        \"location\": [\n",
    "            \"mittig\",\n",
    "            \"links\",\n",
    "            \"auf der linken seite\",\n",
    "            \"neben\",\n",
    "        ],\n",
    "        \"neighbors\": [\"person\", \"spielkonsole\", \"kurven\", \"terminal\"],\n",
    "        \"content\": [\"buchstaben\", \"a\", \"b\"],\n",
    "        \"adjectives\": [\"blau\", \"türkis\", \"gelb\", \"lila\", \"bunt\"],\n",
    "\t\t\"hyperonyms\": [\"ding\", \"gegenstand\", \"objekt\"],\n",
    "    },\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "def0d7dd",
   "metadata": {},
   "source": [
    "### Interaction Model with detailed slot names, values, and synonyms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "707e4cf8",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = json.load(open('interaction_model.json', 'r', encoding='utf-8'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "096ff063",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'other': ['Absolvent', 'Student', 'Gesicht', 'Smiley', 'Sterne', 'Stern', 'Bildschirm', 'Podest', 'Binärcode', 'Stühle', 'Stuhl', 'Seeigel', 'Mine', 'Pfeile', 'Pfeil', 'Legobaustein', 'Legostein', 'Tisch', 'ABC', 'Molekül', 'Atom', 'Schrank', 'Delfin', 'Medaille', 'Tauben', 'Taube', 'Vögel', 'Vogel', 'Pfeile', 'Pfeil', 'Regenbogen', 'Reagenzglas', 'Schultasche', 'Säulen', 'Spinne', 'Wolken', 'Wolke', 'Blumentopf', 'Pflanze', 'Treppe', 'Lineal', 'Büro', 'Schreibtisch', 'Giraffenkopf', 'Giraffe', 'Zahnrad'], 'Ding': ['Sachen', 'Sache', 'Teile', 'Gegenstände', 'Dinger', 'Dingens', 'Dings', 'Objekt', 'Gegenstand', 'Teil'], 'Uhr': [], 'Würfel': ['Buchstabenwürfel'], 'Terminal': ['Programmfenster', 'Bildschirm', 'Computerbildschirm', 'Konsole'], 'Teleskop': ['Fernrohr'], 'Tafel': ['Schultafel'], 'Stift': ['Schreibutensil', 'Buntstift'], 'Spielkonsole': ['Gameboy', 'Switch', 'Spielekonsole'], 'Raumschiff': ['Spaceshuttle', 'Flugzeug'], 'Planet': ['Saturn'], 'Person': ['Ärztin', 'Arzt', 'Ingenieurin', 'Ingenieur', 'Mann', 'Frau', 'Wissenschaftlerin', 'Wissenschaftler', 'Forscherin', 'Forscher', 'Personen', 'Mensch'], 'Mikroskop': ['Teleskop'], 'Labor': ['Kasten', 'Lab'], 'Kurven': ['Grafiken', 'Diagramm', 'Wellen'], 'Glühbirne': ['Birne'], 'Gleichung': ['Screen'], 'Kinder': ['Demo', 'Klimaaktivistinnen', 'Klimaaktivisten', 'Fridays for Future', 'FFF', 'Kinder mit Flagge', 'Friends For Future', 'Friends', 'Future'], 'Europa': ['Flagge', 'Fahne', 'EU', 'Europaflagge'], 'Diagramm': ['Tafel', 'Koordinatensystem'], 'Buch': ['Heft'], 'Boot': ['Schiff'], 'Astronautinnen': ['Drillinge', 'Raumfahrerinnen', 'Frauen'], 'Ballon': ['Luftballon', 'Ball'], 'Biene': ['Tier', 'Insekt'], 'Globus': ['Weltkugel', 'Erdkugel']}\n"
     ]
    }
   ],
   "source": [
    "# create a nested dictionary from interaction model slots\n",
    "slots = dict()\n",
    "# structure: {slot: dict of slot values: list of their synonyms}\n",
    "for s in model[\"interactionModel\"][\"languageModel\"][\"types\"]:\n",
    "    slots[s['name']] = dict()\n",
    "    for v in s['values']:\n",
    "        slots[s['name']][v['name']['value']] = list()\n",
    "        if 'synonyms' in v['name']:\n",
    "            slots[s['name']][v['name']['value']] = v['name']['synonyms']\n",
    "print(slots['object'])  # example for slot 'object'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "382d4248",
   "metadata": {},
   "source": [
    "### Encode grammatical properties of nouns, determiners, and verbs/sentence beginnings (BOS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "663efe7d",
   "metadata": {},
   "outputs": [],
   "source": [
    "OBJECTS = \"\"\"\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Absolvent'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Student'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Gesicht'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Smiley'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Sterne'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Sternen'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Stern'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Bildschirm'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Podest'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Binärcode'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Stühle'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Stuhl'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Seeigel'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Mine'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Pfeile'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Pfeilen'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Pfeil'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Legobaustein'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Legostein'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Tisch'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'ABC'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Molekül'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Atom'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Schrank'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Delfin'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Medaille'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Tauben'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Taube'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Vögel'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Vogel'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Pfeile'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Pfeil'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Regenbogen'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Reagenzglas'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Schultasche'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Säulen'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Spinne'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Wolken'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Wolke'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Blumentopf'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Pflanze'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Treppe'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Lineal'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Büro'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Schreibtisch'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Giraffenkopf'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Giraffe'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Zahnrad'\n",
    "N[CAS=?c,NUM=sg,GEN=?g] -> 'other'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Sachen'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Sache'\n",
    "N[CAS=?c,NUM=pl,GEN=n] -> 'Teile'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Gegenstände'\n",
    "N[CAS=?c,NUM=pl,GEN=n] -> 'Dinger'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Dingens'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Dings'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Objekt'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Gegenstand'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Teil'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Ding'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Uhr'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Buchstabenwürfel'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Würfel'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Programmfenster'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Bildschirm'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Computerbildschirm'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Konsole'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Terminal'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Fernrohr'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Teleskop'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Schultafel'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Tafel'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Schreibutensil'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Buntstift'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Stift'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Gameboy'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Switch'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Spielekonsole'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Spielkonsole'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Spaceshuttle'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Flugzeug'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Raumschiff'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Saturn'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Planet'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Ärztin'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Arzt'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Ingenieurin'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Ingenieur'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Mann'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Frau'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Wissenschaftlerin'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Wissenschaftler'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Forscherin'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Forscher'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Personen'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Mensch'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Person'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Teleskop'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Mikroskop'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Kasten'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Lab'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Labor'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Grafiken'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Diagramm'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Wellen'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Kurven'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Birne'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Glühbirne'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Screen'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Gleichung'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Demo'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Klimaaktivistinnen'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Klimaaktivisten'\n",
    "N[CAS=?c,NUM=pl,GEN=?g] -> 'Fridays for Future'\n",
    "N[CAS=?c,NUM=pl,GEN=?g] -> 'FFF'\n",
    "N[CAS=?c,NUM=pl,GEN=n] -> 'Kinder mit Flagge'\n",
    "N[CAS=?c,NUM=pl,GEN=?g] -> 'Friends For Future'\n",
    "N[CAS=?c,NUM=pl,GEN=n] -> 'Kinder'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Flagge'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Fahne'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Europaflagge'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Tafel'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Koordinatensystem'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Diagramm'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Heft'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Buch'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Schiff'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Boot'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Drillinge'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Raumfahrerinnen'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Frauen'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Astronautinnen'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Luftballon'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Ball'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Ballon'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Tier'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Insekt'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Biene'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Weltkugel'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Erdkugel'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Globus'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Menschen'\n",
    "N[CAS=?c,NUM=pl,GEN=f] -> 'Personen'\n",
    "N[CAS=?c,NUM=pl,GEN=n] -> 'Kinder'\n",
    "N[CAS=?c,NUM=pl,GEN=m] -> 'Streifen'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Flugobjekt'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Ufo'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Anzügen'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Spielzeug'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Lebewesen'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Wasser'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Fluss'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Roboter'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Rauch'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'X'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Y'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Sigma'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Summenzeichen'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Gleichheitszeichen'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Summe'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Normalverteilung'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Gaußkurve'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Kolben'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Kugeln'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Kästen'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Ordner'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Kittel'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Brille'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Sprechblase'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Tasten'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Knöpfen'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Zwei'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Vier'\n",
    "N[CAS=?c,NUM=sg,GEN=m] -> 'Punkt'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'Komma'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Schrift'\n",
    "N[CAS=?c,NUM=sg,GEN=f] -> 'Zeit'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Buchstaben'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'A'\n",
    "N[CAS=?c,NUM=sg,GEN=n] -> 'B'\n",
    "N[CAS=dat,NUM=pl,GEN=m] -> 'Zeigern'\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "177275f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "ARTICLES = \"\"\"\n",
    "Det[CAS=nom,NUM=sg,GEN=m] -> 'der'\n",
    "Det[CAS=gen,NUM=sg,GEN=m] -> 'des'\n",
    "Det[CAS=dat,NUM=sg,GEN=m] -> 'dem'\n",
    "Det[CAS=acc,NUM=sg,GEN=m] -> 'den'\n",
    "Det[CAS=nom,NUM=sg,GEN=f] -> 'die'\n",
    "Det[CAS=gen,NUM=sg,GEN=f] -> 'der'\n",
    "Det[CAS=dat,NUM=sg,GEN=f] -> 'der'\n",
    "Det[CAS=acc,NUM=sg,GEN=f] -> 'die'\n",
    "Det[CAS=nom,NUM=sg,GEN=n] -> 'das'\n",
    "Det[CAS=gen,NUM=sg,GEN=n] -> 'des'\n",
    "Det[CAS=dat,NUM=sg,GEN=n] -> 'dem'\n",
    "Det[CAS=acc,NUM=sg,GEN=n] -> 'das'\n",
    "Det[CAS=nom,NUM=pl,GEN=?g] -> 'die'\n",
    "Det[CAS=gen,NUM=pl,GEN=?g] -> 'der'\n",
    "Det[CAS=dat,NUM=pl,GEN=?g] -> 'den'\n",
    "Det[CAS=acc,NUM=pl,GEN=?g] -> 'die'\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "d3075d78",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Bos1: phrases for first turn only\n",
    "BOS = \"\"\"\n",
    "Bos1[CAS=nom,NUM=?n] -> 'und jetzt'\n",
    "Bos[CAS=acc,NUM=?n] -> 'bitte zeig mir'\n",
    "Bos[CAS=nom,NUM=sg] -> 'das objekt ist'\n",
    "Bos[CAS=nom,NUM=sg] -> 'es ist'\n",
    "Bos[CAS=acc,NUM=?n] -> 'zeig mal'\n",
    "Bos[CAS=acc,NUM=?n,NUM=?n] -> 'wähl mal'\n",
    "Bos[CAS=nom,NUM=sg] -> 'das ding ist'\n",
    "Bos1[CAS=acc,NUM=?n] -> 'ich nehme'\n",
    "Bos[CAS=acc,NUM=?n] -> 'bitte wähle'\n",
    "Bos[CAS=acc,NUM=?n] -> 'bitte finde'\n",
    "Bos1[CAS=nom,NUM=?n] -> 'dann noch'\n",
    "Bos1[CAS=nom,NUM=?n] -> 'dann'\n",
    "Bos[CAS=acc,NUM=?n] -> 'finde'\n",
    "Bos1[CAS=acc,NUM=?n] -> 'wähle mal'\n",
    "Bos1[CAS=acc,NUM=?n,NUM=?n] -> 'ich sehe'\n",
    "Bos1[CAS=acc,NUM=?n,NUM=?n,NUM=?n] -> 'nimm mal'\n",
    "Bos[CAS=nom,NUM=sg] -> 'der gegenstand ist'\n",
    "Bos[CAS=acc,NUM=?n,NUM=?n,NUM=?n,NUM=?n] -> 'ich denke an'\n",
    "Bos[CAS=acc,NUM=?n] -> 'ich wähle'\n",
    "Bos[CAS=acc,NUM=?n] -> 'wähle'\n",
    "Bos[CAS=acc,NUM=?n] -> 'siehst du'\n",
    "Bos[CAS=acc,NUM=?n,NUM=?n] -> 'zeig'\n",
    "Bos1[CAS=nom,NUM=?n] -> 'als nächstes'\n",
    "Bos1[CAS=dat,NUM=?n,NUM=?n] -> 'weiter mit'\n",
    "Bos[CAS=acc,NUM=?n] -> 'bitte zeige'\n",
    "Bos1[CAS=nom,NUM=sg] -> 'das nächste objekt ist'\n",
    "Bos[CAS=acc,NUM=?n] -> 'ich beschreibe'\n",
    "Bos1[CAS=nom,NUM=?n] -> 'jetzt'\n",
    "Bos1[CAS=nom,NUM=sg] -> 'der nächste gegenstand ist'\n",
    "Bos[CAS=acc,NUM=?n] -> 'du siehst'\n",
    "Bos[CAS=acc,NUM=?n] -> 'wähl'\n",
    "Bos[CAS=acc,NUM=?n] -> 'zeig mir'\n",
    "Bos[CAS=acc,NUM=?n] -> 'ich meine'\n",
    "Bos1[CAS=dat,NUM=?n] -> 'weiter gehts mit'\n",
    "Bos[CAS=acc,NUM=?n] -> 'nimm'\n",
    "Bos[CAS=acc,NUM=?n] -> 'bitte nimm'\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "66112d43",
   "metadata": {},
   "outputs": [],
   "source": [
    "ADJECTIVE_ENDINGS = \"\"\"\n",
    "AdjEnd[CAS=nom,NUM=sg,GEN=m] -> 'e'\n",
    "AdjEnd[CAS=gen,NUM=sg,GEN=m] -> 'en'\n",
    "AdjEnd[CAS=dat,NUM=sg,GEN=m] -> 'en'\n",
    "AdjEnd[CAS=acc,NUM=sg,GEN=m] -> 'en'\n",
    "AdjEnd[CAS=nom,NUM=sg,GEN=f] -> 'e'\n",
    "AdjEnd[CAS=gen,NUM=sg,GEN=f] -> 'en'\n",
    "AdjEnd[CAS=dat,NUM=sg,GEN=f] -> 'en'\n",
    "AdjEnd[CAS=acc,NUM=sg,GEN=f] -> 'e'\n",
    "AdjEnd[CAS=nom,NUM=sg,GEN=n] -> 'e'\n",
    "AdjEnd[CAS=gen,NUM=sg,GEN=n] -> 'en'\n",
    "AdjEnd[CAS=dat,NUM=sg,GEN=n] -> 'en'\n",
    "AdjEnd[CAS=acc,NUM=sg,GEN=n] -> 'e'\n",
    "AdjEnd[CAS=?c,NUM=pl,GEN=?g] -> 'en'\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "857ee9a4",
   "metadata": {},
   "source": [
    "## Transform rules: Specify for all grammatical features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "fa6b752c",
   "metadata": {},
   "outputs": [],
   "source": [
    "NUM = {'sg', 'pl'}\n",
    "CAS = {'nom', 'acc', 'gen', 'dat'}\n",
    "GEN = {'m', 'f', 'n'}\n",
    "\n",
    "def specify_rule(r):\n",
    "    \"\"\"specify feature grammar rule containing ? for all possible features and values\"\"\"\n",
    "    new_rules = set([r])\n",
    "    if '?g' in r:\n",
    "        for g in GEN:\n",
    "            new = set([rule.replace('?g', g) for rule in new_rules])\n",
    "            new_rules = new_rules.union(new)  \n",
    "    if '?c' in r:\n",
    "        for c in CAS:\n",
    "            new = set([rule.replace('?c', c) for rule in new_rules])\n",
    "            new_rules = new_rules.union(new)\n",
    "    if '?n' in r:\n",
    "        for n in NUM:\n",
    "            new = set([rule.replace('?n', n) for rule in new_rules])\n",
    "            new_rules = new_rules.union(new)\n",
    "    new_rules = new_rules.difference(set([rule for rule in new_rules if '?' in rule]))\n",
    "    return list(new_rules)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "38ac9c21",
   "metadata": {},
   "outputs": [],
   "source": [
    "def adjective_rules(a):\n",
    "    \"\"\"generate rules with inflection for an adjective\"\"\"\n",
    "    adj_rules = []\n",
    "    if a[-1].isdigit():  # no inflection for digits\n",
    "        rule = f'Adj[CAS=?c,NUM=?n,GEN=?g] -> \\'{a}\\''\n",
    "        for sr in specify_rule(rule):  # specify for all CNG combis\n",
    "            adj_rules.append(sr)\n",
    "    else:\n",
    "        if a in {'lila', 'rosa'}:\n",
    "            a += 'n'\n",
    "        for rule in ADJECTIVE_ENDINGS.split('\\n'):\n",
    "            r = rule.replace('AdjEnd', 'Adj')\n",
    "            r = r.replace('AdjEnd', 'Adj')\n",
    "            r = r.replace('-> \\'', f'-> \\'{a}')\n",
    "            for sr in specify_rule(r):  # specify for all CNG combis\n",
    "                adj_rules.append(sr)\n",
    "    return adj_rules"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3241b0e0",
   "metadata": {},
   "source": [
    "## Non-terminal rules by turn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "0b2ab623",
   "metadata": {},
   "outputs": [],
   "source": [
    "NONTERMINALS1 = \"\"\"\n",
    "S -> Bos1[CAS=?c,NUM=?n] NP[CAS=?c,NUM=?n] Loc\n",
    "S -> Bos[CAS=?c,NUM=?n] NP[CAS=?c,NUM=?n] Loc\n",
    "NP[CAS=?c,NUM=?n] -> Det[CAS=?c,NUM=?n,GEN=?g] Adj[CAS=?c,NUM=?n,GEN=?g] N[CAS=?c,NUM=?n,GEN=?g]\"\"\"\n",
    "\n",
    "NONTERMINALS2 = \"\"\"\n",
    "S -> Bos[CAS=?c,NUM=?n] NP[CAS=?c,NUM=?n] ContP Loc\n",
    "S -> Bos[CAS=?c,NUM=?n] NP[CAS=?c,NUM=?n] NeighP Loc\n",
    "NP[CAS=?c,NUM=?n] -> Det[CAS=?c,NUM=?n,GEN=?g] Adj[CAS=?c,NUM=?n,GEN=?g] N[CAS=?c,NUM=?n,GEN=?g]\n",
    "ContP -> \\'mit\\' Det[CAS=dat,NUM=?n,GEN=?g] Cont[CAS=dat,NUM=?n,GEN=?g]\n",
    "NeighP -> Nextto Det[CAS=dat,NUM=?n,GEN=?g] Neigh[CAS=dat,NUM=?n,GEN=?g]\n",
    "Nextto -> \\'neben\\' | \\'bei\\' | \\'in der Nähe von\\'\n",
    "\"\"\"\n",
    "\n",
    "NONTERMINALS3 = \"\"\"\n",
    "S -> Bos[CAS=?c,NUM=?n] NP[CAS=?c,NUM=?n] ContP NeighP Loc\n",
    "S -> Bos[CAS=?c,NUM=?n] NP[CAS=?c,NUM=?n] NeighP Loc \n",
    "NP[CAS=?c,NUM=?n] -> Det[CAS=?c,NUM=?n,GEN=?g] Adj[CAS=?c,NUM=?n,GEN=?g] N[CAS=?c,NUM=?n,GEN=?g]\n",
    "ContP -> \\'mit\\' Det[CAS=dat,NUM=?n,GEN=?g] Cont[CAS=dat,NUM=?n,GEN=?g]\n",
    "NeighP -> Nextto Det[CAS=dat,NUM=?n,GEN=?g] Neigh[CAS=dat,NUM=?n,GEN=?g]\n",
    "Nextto -> \\'neben\\' | \\'bei\\' | \\'in der Nähe von\\'\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ff2aee7c",
   "metadata": {},
   "source": [
    "## Transform rules to CFG, split by turn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "b7835ccf",
   "metadata": {},
   "outputs": [],
   "source": [
    "def CFG_transform(rule):\n",
    "    \"\"\"transform feature grammar rules to CFG rules\"\"\"\n",
    "    rule = rule.replace('[', '_')\n",
    "    rule = rule.replace(',', '_')\n",
    "    rule = rule.replace('=', '')\n",
    "    rule = rule.replace(']', '')\n",
    "    rule = rule.replace('?', '')\n",
    "    return rule"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "ef85686f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def split_rules(l, n=3):\n",
    "    \"\"\"split rules to avoid same information by turn\"\"\"\n",
    "    if len(l) > 2:\n",
    "        split_len = len(l) // n\n",
    "        return l[:split_len], l[split_len:(split_len*2)], l[(split_len*2):]\n",
    "    else:\n",
    "        if len(l) == 1:\n",
    "            return l, l, l\n",
    "        split_len = 1  # turn 2=0\n",
    "        return l[:split_len], l[split_len:(split_len*2)], l[:split_len]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb02d1b5",
   "metadata": {},
   "source": [
    "## Merge slots and grammatical information to feature grammar"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "c8ad93c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "def featgram(i):\n",
    "    \"\"\"\n",
    "    create a feature grammar for an item from\n",
    "    - item-specific terminal rules\n",
    "    - turn-specific non-terminal rules\n",
    "    \"\"\"\n",
    "    d = items_dict[i]  # load item's dict\n",
    "    rules = [\"% start S\"]  # initialize rules\n",
    "    # specified article rules\n",
    "    for r in ARTICLES.split('\\n'):\n",
    "        rules += specify_rule(r)\n",
    "    # specified BOS rules\n",
    "    for r in BOS.split('\\n'):\n",
    "        rules += specify_rule(r)\n",
    "\n",
    "    # ITEM SPECIFIC terminal rules\n",
    "    # item synonyms from slot\n",
    "    synonyms = slots['object'][d[\"slot\"].capitalize()]\n",
    "    # check for object name and synonyms\n",
    "    objs = slots['object'][d['slot'].capitalize()] + [d['slot'].capitalize()]\n",
    "    obj_rules = [r for r in OBJECTS.split('\\n') if '\\'' in r and r.split('\\'')[1] in objs]\n",
    "    # location information\n",
    "    locs = d['location']\n",
    "    locs_transitive = {'neben', 'über', 'unter', 'hinter', 'auf', 'bei', 'vor'}\n",
    "    # ignore transitive location for now\n",
    "    loc_rules = [f'Loc -> \\'{l}\\'' for l in locs if l not in locs_transitive]\n",
    "    loc_rules1,loc_rules2,loc_rules3 = split_rules(loc_rules)\n",
    "    # adjectives\n",
    "    adj = d['adjectives']\n",
    "    adj1, adj2, adj3 = split_rules(adj)\n",
    "    \n",
    "    # neighbors\n",
    "    neighbors = d['neighbors']\n",
    "    # content\n",
    "    cont = d['content']\n",
    "    \n",
    "    # TURN 1\n",
    "    rules1 = [] + rules\n",
    "    # BOS + Det + Adj + hyperonym + location\n",
    "    # e.g. wähle die blonde Person in der mitte\n",
    "    # non terminal rules, replace general rule by specified rules\n",
    "    for r in NONTERMINALS1.split('\\n'):\n",
    "        rules1 += specify_rule(r)\n",
    "    hyps = d['hyperonyms']\n",
    "    hyp_rules = [r for r in OBJECTS.split('\\n') if '\\'' in r and r.split('\\'')[1].lower() in hyps]\n",
    "    for r in hyp_rules:\n",
    "        rules1 += specify_rule(r)\n",
    "    for a in adj1:\n",
    "        rules1 += adjective_rules(a)\n",
    "    rules1 += loc_rules1\n",
    "    rules1 = [CFG_transform(r) for r in rules1 if r]\n",
    "    fg1 = CFG.fromstring(rules1)\n",
    "    \n",
    "    # TURN 2\n",
    "    rules2 = [] + rules\n",
    "    # Bos + Det + Adj + hyperonym + content + location\n",
    "    # e.g. Bitte finde die weißen Frauen mit den Anz ̈ugen oben OR\n",
    "    # Bos + Det + Adj + hyperonym + neighbor + location\n",
    "    # e.g. Ich meine die weißen Frauen bei dem Raumschiff oben\n",
    "    for r in NONTERMINALS2.split('\\n'):\n",
    "        rules2 += specify_rule(r)\n",
    "    for r in hyp_rules:\n",
    "        rules2 += specify_rule(r)\n",
    "    # content rules\n",
    "    cont_rules = [r.replace('N[', 'Cont[') for r in OBJECTS.split('\\n') if '\\'' in r\n",
    "                  and r.split('\\'')[1].lower() in cont]\n",
    "    for r in cont_rules:\n",
    "        rules2 += specify_rule(r)\n",
    "    # neighbor rules\n",
    "    neigh_rules = [r.replace('N[', 'Neigh[') for r in OBJECTS.split('\\n') if '\\'' in r\n",
    "                  and r.split('\\'')[1].lower() in neighbors]\n",
    "    for r in neigh_rules:\n",
    "        rules2 += specify_rule(r)\n",
    "    for a in adj2:\n",
    "        rules2 += adjective_rules(a)\n",
    "    rules2 += loc_rules2\n",
    "    rules2 = [CFG_transform(r) for r in rules2 if r]\n",
    "    fg2 = CFG.fromstring(rules2)\n",
    "            \n",
    "    # TURN 3\n",
    "    rules3 = [] + rules\n",
    "    # Bos + Det + Adj + object + content + neighbor + location\n",
    "    # e.g. Bitte zeige die 3 Astronautinnen mit den Anzügen in der Nähe\n",
    "    # von dem Raumschiff in der Ecke OR\n",
    "    # Bos + Det + Adj + object + neighbor + location\n",
    "    # e.g. Nimm die 3 Astronautinnen bei dem Ufo in der Ecke\n",
    "    for r in NONTERMINALS3.split('\\n'):\n",
    "        rules3 += specify_rule(r)\n",
    "    for r in obj_rules:\n",
    "        rules3 += specify_rule(r)\n",
    "    for r in neigh_rules:\n",
    "        rules3 += specify_rule(r)\n",
    "    for r in cont_rules:\n",
    "        rules3 += specify_rule(r)\n",
    "    rules3 += loc_rules3\n",
    "    for a in adj3:\n",
    "        rules3 += adjective_rules(a)\n",
    "    rules3 = [CFG_transform(r) for r in rules3 if r]\n",
    "    fg3 = CFG.fromstring(rules3)\n",
    "    return fg1, fg2, fg3"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "167182d9",
   "metadata": {},
   "source": [
    "## Run for all items"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "dc7f80bc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def run(turns=3, output=True):\n",
    "    for item in items_dict.keys():\n",
    "        utterance_dict = dict()  \n",
    "        featgrams = featgram(item)  # returns 3 feature grammars\n",
    "        for i in range(turns):\n",
    "            sents = list()\n",
    "            for sentence in generate(featgrams[i], depth=5, n=100000):\n",
    "                # split multiword terminals\n",
    "                sents.append([t.replace('_', ',') for tokens in sentence for t in tokens.split()])\n",
    "            print(item, i, len(sents))\n",
    "            try:\n",
    "                utterance_dict[str(i)] = random.sample(sents, k=90)  # keep only 90 utterances per turn\n",
    "            except ValueError:  # less sentences\n",
    "                utterance_dict[str(i)] = sents\n",
    "        if output:\n",
    "            with open(f'utterances/{item}.json', 'w', encoding='utf-8') as f:\n",
    "                json.dump(utterance_dict, f, indent=4, ensure_ascii=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "5e6fa205",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "astronautinnen 0 96\n",
      "astronautinnen 1 448\n",
      "astronautinnen 2 768\n",
      "ballon 0 120\n",
      "ballon 1 480\n",
      "ballon 2 1080\n",
      "biene 0 90\n",
      "biene 1 420\n",
      "biene 2 720\n",
      "boot 0 90\n",
      "boot 1 540\n",
      "boot 2 1080\n",
      "buchLinksOben 0 90\n",
      "buchLinksOben 1 540\n",
      "buchLinksOben 2 360\n",
      "buchMitteUnten 0 90\n",
      "buchMitteUnten 1 540\n",
      "buchMitteUnten 2 360\n",
      "buchLinksObenOffen 0 90\n",
      "buchLinksObenOffen 1 540\n",
      "buchLinksObenOffen 2 720\n",
      "buchRechtsMitte 0 90\n",
      "buchRechtsMitte 1 720\n",
      "buchRechtsMitte 2 480\n",
      "diagramm 0 90\n",
      "diagramm 1 780\n",
      "diagramm 2 9600\n",
      "europa 0 90\n",
      "europa 1 480\n",
      "europa 2 1080\n",
      "fff 0 120\n",
      "fff 1 1040\n",
      "fff 2 5856\n",
      "gleichung 0 90\n",
      "gleichung 1 660\n",
      "gleichung 2 1080\n",
      "globus 0 90\n",
      "globus 1 360\n",
      "globus 2 720\n",
      "gluehbirneObenMitteGross 0 90\n",
      "gluehbirneObenMitteGross 1 540\n",
      "gluehbirneObenMitteGross 2 360\n",
      "gluehbirneObenMitteKlein 0 90\n",
      "gluehbirneObenMitteKlein 1 360\n",
      "gluehbirneObenMitteKlein 2 240\n",
      "kurven 0 90\n",
      "kurven 1 900\n",
      "kurven 2 11880\n",
      "lab 0 90\n",
      "lab 1 660\n",
      "lab 2 1620\n",
      "mikroskop 0 90\n",
      "mikroskop 1 1080\n",
      "mikroskop 2 1080\n",
      "personLinksMitte 0 60\n",
      "personLinksMitte 1 480\n",
      "personLinksMitte 2 6528\n",
      "personMitteMitte 0 60\n",
      "personMitteMitte 1 560\n",
      "personMitteMitte 2 9792\n",
      "personRechtsMitte 0 60\n",
      "personRechtsMitte 1 440\n",
      "personRechtsMitte 2 7344\n",
      "planet 0 90\n",
      "planet 1 360\n",
      "planet 2 480\n",
      "raumschiff 0 120\n",
      "raumschiff 1 1200\n",
      "raumschiff 2 1800\n",
      "spielKonsole 0 90\n",
      "spielKonsole 1 840\n",
      "spielKonsole 2 5760\n",
      "stiftLinksMitte 0 120\n",
      "stiftLinksMitte 1 960\n",
      "stiftLinksMitte 2 1440\n",
      "stiftRechtsUnten 0 120\n",
      "stiftRechtsUnten 1 1440\n",
      "stiftRechtsUnten 2 2160\n",
      "tafel 0 90\n",
      "tafel 1 960\n",
      "tafel 2 7200\n",
      "teleskop 0 90\n",
      "teleskop 1 360\n",
      "teleskop 2 360\n",
      "terminalMitteMitte 0 90\n",
      "terminalMitteMitte 1 960\n",
      "terminalMitteMitte 2 3600\n",
      "uhrLinksOben 0 90\n",
      "uhrLinksOben 1 840\n",
      "uhrLinksOben 2 720\n",
      "uhrRechtsOben 0 90\n",
      "uhrRechtsOben 1 660\n",
      "uhrRechtsOben 2 1620\n",
      "wuerfel 0 90\n",
      "wuerfel 1 900\n",
      "wuerfel 2 5760\n"
     ]
    }
   ],
   "source": [
    "run()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}