diff --git a/code/.ipynb_checkpoints/generalized_graph_building-checkpoint.ipynb b/code/.ipynb_checkpoints/generalized_graph_building-checkpoint.ipynb
index 1092f949d6e0f11040f79eaf34262f80c5cf395e..e65187f57c44818a6cdef1358f801b5d9cfbabf7 100644
--- a/code/.ipynb_checkpoints/generalized_graph_building-checkpoint.ipynb
+++ b/code/.ipynb_checkpoints/generalized_graph_building-checkpoint.ipynb
@@ -9,7 +9,7 @@
},
{
"cell_type": "code",
- "execution_count": 19,
+ "execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
@@ -66,7 +66,7 @@
},
{
"cell_type": "code",
- "execution_count": 6,
+ "execution_count": 29,
"metadata": {},
"outputs": [
{
@@ -76,12 +76,16 @@
"(b / be-located-at-91\n",
" :ARG1 (p / person\n",
" :name (n / name\n",
- " :op1 \"Addie\"))\n",
- " :ARG2 (z / zoo)\n",
- " :accompanier (p2 / person\n",
- " :ARG0-of (h / have-rel-role-91\n",
- " :ARG1 p\n",
- " :ARG2 (f / family))))\n"
+ " :op1 \"Max\")\n",
+ " :ARG0-of (h / have-org-role-91\n",
+ " :ARG1 (o / organization\n",
+ " :name (n2 / name\n",
+ " :op1 \"Some\"\n",
+ " :op2 \"Germans\"))\n",
+ " :ARG2 (m / member)))\n",
+ " :ARG2 (c / country\n",
+ " :name (n3 / name\n",
+ " :op1 \"France\")))\n"
]
}
],
@@ -90,7 +94,7 @@
" with open(file,'r') as f:\n",
" return(json.load(f))\n",
"amr_dict = open_amr_dict('../amr/amr_dict.json')\n",
- "print(amr_dict['Addie was at the zoo with her family'])"
+ "print(amr_dict['Max who is a member of Some Germans is at France'])"
]
},
{
@@ -121,7 +125,7 @@
},
{
"cell_type": "code",
- "execution_count": 6,
+ "execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
@@ -551,7 +555,7 @@
},
{
"cell_type": "code",
- "execution_count": 21,
+ "execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
@@ -594,7 +598,7 @@
},
{
"cell_type": "code",
- "execution_count": 71,
+ "execution_count": 8,
"metadata": {},
"outputs": [],
"source": [
@@ -615,7 +619,7 @@
},
{
"cell_type": "code",
- "execution_count": 72,
+ "execution_count": 9,
"metadata": {},
"outputs": [
{
@@ -632,7 +636,7 @@
},
{
"cell_type": "code",
- "execution_count": 8,
+ "execution_count": 10,
"metadata": {},
"outputs": [],
"source": [
@@ -643,7 +647,7 @@
},
{
"cell_type": "code",
- "execution_count": 68,
+ "execution_count": 11,
"metadata": {},
"outputs": [],
"source": [
@@ -662,7 +666,7 @@
},
{
"cell_type": "code",
- "execution_count": 73,
+ "execution_count": 32,
"metadata": {},
"outputs": [],
"source": [
@@ -674,50 +678,22 @@
" # get general rule\n",
" general_rule = annotation['2parse'][1]\n",
" part1, part2 = general_rule[0], general_rule[2]\n",
- " \n",
- " # get general rule\n",
- " #general_rule = annotation['general_0']\n",
- " #part1, part2 = general_rule[0].lower(), general_rule[2].lower()\n",
- " #print(part1)\n",
- " #print(part2)\n",
"\n",
" # rename fillers\n",
" filler1 = [replacers[word.replace(')','')] for word in annotation['general_0'][0].split(' ') if '_' in word]\n",
" filler2 = [replacers[word.replace(')','')] for word in annotation['general_0'][2].split(' ') if '_' in word]\n",
" filler = [el for el in filler1 if el in filler2]\n",
- " #filler = {el:[] for el in filler1 if el in filler2}\n",
- " #part1, part2 = rename_fillers(part1), rename_fillers(part2)\n",
- " #print(part1)\n",
- " #print(part2)\n",
- " # save index of fillers in original sentence\n",
- " #for fill in filler:\n",
- " #filler[fill] = [part1.split().index(fill), part2.split().index(fill)]\n",
- " #print(filler)\n",
- "\n",
- " # parse to amr\n",
- " #doc1, doc2 = nlp(part1), nlp(part2)\n",
- " #parse_part1, parse_part2 = doc1._.to_amr(), doc2._.to_amr()\n",
- " #triples_part1, triples_part2 = get_graph_triples(parse_part1[0]), get_graph_triples(parse_part2[0])\n",
- "\n",
- " #print(parse_part1)\n",
- " #print(parse_part2)\n",
- " \n",
- " \n",
- " # parse to amr\n",
+ " left_fillers = [el for el in filler1 if el not in filler]+[el for el in filler2 if el not in filler]\n",
+ " #print(left_fillers,' fillers left')\n",
+ " # look up amr parse\n",
" parse_part1, parse_part2 = replace_names(amr_dict[part1]),replace_names(amr_dict[part2])\n",
" triples_part1, triples_part2 = get_graph_triples(parse_part1), get_graph_triples(parse_part2)\n",
- " #print(parse_part1)\n",
- " #print(triples_part1)\n",
- " #print(parse_part2)\n",
- " #print(triples_part2)\n",
" \n",
" for fill in filler:\n",
" fill = re.sub('[()]', '', replace_names(amr_dict[fill]).split('/ ')[1].split('\\n')[0])\n",
- " #fill = filler[fill][0]\n",
- " #fill = re.sub('[()]', '', doc1[fill:fill+1]._.to_amr()[0].split('/ ')[1].split('\\n')[0])\n",
- " #print(fill)\n",
+ " \n",
" for triple in triples_part1:\n",
- " #print(triple)\n",
+ " \n",
" triple1 = (None,None)\n",
" if not '_of' in triple[1]:\n",
" if fill in triple[2]:\n",
@@ -741,134 +717,143 @@
" mapping = (triple1,triple2)\n",
" argument_mappings.append(mapping)\n",
" else:\n",
- " left_over.append(rule)\n",
+ " left_over.append(annotation)\n",
+ " #print(triple1,triple2)\n",
" continue\n",
- " print(list(set(argument_mappings)))\n",
+ " #print(list(set(argument_mappings)))\n",
" #print('\\n\\n')\n",
" return((list(set(argument_mappings)),left_over))"
]
},
{
"cell_type": "code",
- "execution_count": 74,
+ "execution_count": 21,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "[(('w / want-01', ':ARG0', 'p / \"Germany\"'), ('b / be-located-at-91', ':ARG1', 'p / \"Germany\"'))]\n"
+ "['France'] fillers left\n",
+ "[(('w / want-01', ':ARG0', 'p / \"Germany\"'), ('b / be-located-at-91', ':ARG1', 'p / \"Germany\"'))]\n",
+ "[]\n"
]
- },
- {
- "data": {
- "text/plain": [
- "([(('w / want-01', ':ARG0', 'p / \"Germany\"'),\n",
- " ('b / be-located-at-91', ':ARG1', 'p / \"Germany\"'))],\n",
- " [])"
- ]
- },
- "execution_count": 74,
- "metadata": {},
- "output_type": "execute_result"
}
],
"source": [
- "map_arguments(annotation)"
+ "argument_mapping, left_over = map_arguments(annotation)\n",
+ "print(argument_mapping)\n",
+ "print(left_over)"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "### In loop for every annotation of the edge"
]
},
{
"cell_type": "code",
- "execution_count": 67,
+ "execution_count": 23,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "annotations = [{'dimension': '1', 'quality': 3, 'worker_id': 152, 'rules': [[\"Addie's family drive to the zoo\", 'Causes/Enables', 'Addie is at the zoo'], [\"{Addie's family}_[subject] {drive}_[verb] {to}_[preposition1] {the zoo}_[object1]\", 'Causes/Enables', '{Addie}_[subject] {is}_[verb] {at}_[preposition1] {the zoo}_[object1]'], ['Some People_A drive to Somewhere_A', 'Causes/Enables', 'Someone_A (who is a member of Some People_A) is at Somewhere_A'], ['{Some People_A}_[subject] {drive}_[verb] {to}_[preposition1] {Somewhere_A}_[object1]', 'Causes/Enables', '{Someone_A ||who is a member of Some People_A||}_[subject] {is}_[verb] {at}_[preposition1] {Somewhere_A}_[object1]']], '2parse': [[\"Addie's family drive to the zoo\", 'Causes/Enables', 'Addie is at the zoo'], ['Some Germans drive to France', 'Causes/Enables', 'Max who is a member of Some Germans is at France']], 'general_0': ['Some people_a drive to somewhere_a', 'Causes/Enables', 'someone_a who is a member of Some people_a is at somewhere_a'], 'general_replaced': ['Some Germans drive to France', 'Causes/Enables', 'Max who is a member of Some Germans is at France'], 'general_no_brackets': ['Some Germans drive to France', 'Causes/Enables', 'Max who is a member of Some Germans is at France']}, {'dimension': '1', 'quality': 2, 'worker_id': 11, 'rules': [['Addie goes to the zoo with her family', 'Causes/Enables', 'Addie is at the zoo with her family'], ['{Addie}_[subject] {goes}_[verb] {to}_[preposition1] {the zoo}_[object1] {with}_[preposition2] {her family}_[object2]', 'Causes/Enables', '{Addie}_[subject] {is}_[verb] {at}_[preposition1] {the zoo}_[object1] {with}_[preposition2] {her family}_[object2]'], ['Someone_A goes to Somewhere_A with Some People_A', 'Causes/Enables', 'Someone_A is at Somewhere_A with Some People_A'], ['{Someone_A}_[subject] {goes}_[verb] {to}_[preposition1] {Somewhere_A}_[object1] {with}_[preposition2] {Some People_A}_[object2]', 'Causes/Enables', '{Someone_A}_[subject] {is}_[verb] {at}_[preposition1] {Somewhere_A}_[object1] {with}_[preposition2] {Some People_A}_[object2]']], '2parse': [['Addie goes to the zoo with her family', 'Causes/Enables', 'Addie is at the zoo with her family'], ['Max goes to France with Some Germans', 'Causes/Enables', 'Max is at France with Some Germans']], 'general_0': ['someone_a goes to somewhere_a with Some people_a', 'Causes/Enables', 'someone_a is at somewhere_a with Some people_a'], 'general_replaced': ['Max goes to France with Some Germans', 'Causes/Enables', 'Max is at France with Some Germans'], 'general_no_brackets': ['Max goes to France with Some Germans', 'Causes/Enables', 'Max is at France with Some Germans']}]"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 37,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "['\"Russia\"']\n",
- "['person\\n :mod (c / \"Russia\")']\n",
- "(p / possible-01\n",
- " :ARG1 (r / run-02\n",
- " :ARG0 (c / crown\n",
- " :ARG1-of (g / give-01\n",
- " :ARG0 (p2 / \"Max\")\n",
- " :ARG2 (s / sweater)))\n",
- " :ARG1 (a / around\n",
- " :op1 (c2 / \"France\"))\n",
- " :manner (t / throne))\n",
- " p / \"Russia\") \n",
- " c / \"Adidas\"\n"
+ "(('d / drive-01', ':ARG4', 'c2 / \"France\"'), ('b / be-located-at-91', ':ARG2', 'c / \"France\"'))\n",
+ "(('g / go-02', ':accompanier', 'p2 / \"Germany\"'), ('b / be-located-at-91', ':accompanier', 'p2 / \"Germany\"'))\n",
+ "(('g / go-02', ':ARG4', 'c / \"France\"'), ('b / be-located-at-91', ':ARG2', 'c / \"France\"'))\n",
+ "(('g / go-02', ':ARG0', 'p / \"Max\"'), ('b / be-located-at-91', ':ARG1', 'p / \"Max\"'))\n"
]
}
],
"source": [
- "parse =\"\"\"(f / feed-01\n",
- " :ARG0 (p / person\n",
- " :name (n / name\n",
- " :op1 \"Max\"))\n",
- " :ARG2 (c / crown))\"\"\"\n",
- "parse=\"\"\"(p / possible-01\n",
- " :ARG1 (r / run-02\n",
- " :ARG0 (c / crown\n",
- " :ARG1-of (g / give-01\n",
- " :ARG0 (p2 / person\n",
- " :name (n / name\n",
- " :op1 \"Max\"))\n",
- " :ARG2 (s / sweater)))\n",
- " :ARG1 (a / around\n",
- " :op1 (c2 / country\n",
- " :name (n2 / name\n",
- " :op1 \"France\")))\n",
- " :manner (t / throne))\n",
- " p / person\n",
- " :mod (c / country\n",
- " :name (n / name\n",
- " :op1 \"Russia\"))) \n",
- " c / company\n",
- " :name (n / name\n",
- " :op1 \"Adidas\")\"\"\"\n",
- "name = re.findall('[a-z]+[\\s\\n\\t]*:name \\(n[0-9]? / name[\\s\\n\\t]*:op1 (\"[A-Za-z]+\")\\)', parse)\n",
- "names = re.findall('[a-z]+[\\s\\n\\t]*:name \\(n[0-9]? / name[\\s\\n\\t]*:op1 \"[A-Za-z]+\"\\)', parse)\n",
- "for i,el in enumerate(names):\n",
- " parse = parse.replace(el,name[i])\n",
- "name = re.findall('[a-z]+[\\s\\n\\t]*:mod \\(c[0-9]? / (\"[A-Za-z]+\")\\)', parse)\n",
- "names = re.findall('[a-z]+[\\s\\n\\t]*:mod \\(c[0-9]? / \"[A-Za-z]+\"\\)', parse)\n",
- "for i,el in enumerate(names):\n",
- " parse = parse.replace(el,name[i])\n",
- "print(name)\n",
- "print(names)\n",
- "print(parse)"
+ "def get_argument_mappings(edge):\n",
+ " annotations = graph.edges[edge]['annotations']\n",
+ " argument_mappings = []\n",
+ " for annotation in annotations:\n",
+ " argument_mapping, left_over = map_arguments(annotation)\n",
+ " argument_mappings += argument_mapping\n",
+ " #print(left_over)\n",
+ " return(argument_mappings)\n",
+ "\n",
+ "edge = ('3500_EMO_0', '3500_0')\n",
+ "edge = ('3500_EMO_1', '3500_2')\n",
+ "edge = ('3500_EVENT_3', '3500_0')\n",
+ "argument_mappings = get_argument_mappings(edge)\n",
+ "for map in argument_mappings:\n",
+ " print(map)"
]
},
{
"cell_type": "code",
- "execution_count": 62,
+ "execution_count": 40,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "(p / possible-01\n",
- " :ARG1 (r / run-02\n",
- " :ARG0 (c / crown\n",
- " :ARG1-of (g / give-01\n",
- " :ARG0 (p2 / \"Max\")\n",
- " :ARG2 (s / sweater)))\n",
- " :ARG1 (a / around\n",
- " :op1 (c2 / \"France\"))\n",
- " :manner (t / throne))\n",
- " p / person\n",
- " :mod (c / \"Russia\")) \n",
- " (c / \"Adidas\"\n"
+ "['Some Germans drive to France', 'Causes/Enables', 'Max who is a member of Some Germans is at France']\n",
+ "['Max goes to France with Some Germans', 'Causes/Enables', 'Max is at France with Some Germans']\n",
+ "('p / \"Germany\"', ':quant', 's / some')\n",
+ "('g / go-02', ':ARG0', 'p / \"Max\"')\n",
+ "('n2 / name', ':op1', '\"Some\"')\n",
+ "('g / go-02', ':accompanier', 'p2 / \"Germany\"')\n",
+ "('o / organization', ':name', 'n2 / name')\n",
+ "('b / be-located-at-91', ':ARG2', 'c / \"France\"')\n",
+ "('p2 / \"Germany\"', ':quant', 's / some')\n",
+ "('h / have-org-role-91', ':ARG1', 'o / organization')\n",
+ "('b / be-located-at-91', ':ARG1', 'p / \"Max\"')\n",
+ "('h / have-org-role-91', ':ARG2', 'm / member')\n",
+ "('n2 / name', ':op2', '\"Germans\"')\n",
+ "('g / go-02', ':ARG4', 'c / \"France\"')\n",
+ "('b / be-located-at-91', ':accompanier', 'p2 / \"Germany\"')\n",
+ "('p / \"Max\"', ':ARG0-of', 'h / have-org-role-91')\n",
+ "('d / drive-01', ':ARG0', 'p / \"Germany\"')\n",
+ "('d / drive-01', ':ARG4', 'c2 / \"France\"')\n"
]
}
],
"source": [
- "for i,el in enumerate(names):\n",
- " parse = parse.replace(el,name[i])\n",
- "print(parse)"
+ "def unite_graph_triples(edge):\n",
+ " unite_triples = []\n",
+ " for annotation in graph.edges[edge]['annotations']:\n",
+ " general_rule = annotation['2parse'][1]\n",
+ " print(general_rule)\n",
+ " part1, part2 = general_rule[0], general_rule[2]\n",
+ " unite_triples += get_graph_triples(replace_names(amr_dict[part1]))+get_graph_triples(replace_names(amr_dict[part2]))\n",
+ " unite_triples = list(set(unite_triples))\n",
+ " return(unite_triples)\n",
+ "unite_triples = unite_graph_triples(edge)\n",
+ "for triple in unite_triples:\n",
+ " print(triple)\n",
+ " "
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": []
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "### Check :name and :op2"
]
},
{
@@ -241764,6 +241749,68 @@
"for el i"
]
},
+ {
+ "cell_type": "code",
+ "execution_count": 7,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "#for el in amr_dict.values():\n",
+ " #if ':op2' in el:\n",
+ " #print(el,'\\n\\n')"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 41,
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "(b / be-located-at-91\n",
+ " :ARG1 (p / person\n",
+ " :name (n / name\n",
+ " :op1 \"Max\")\n",
+ " :ARG0-of (h / have-org-role-91\n",
+ " :ARG1 (o / organization\n",
+ " :name (n2 / name\n",
+ " :op1 \"Some\"\n",
+ " :op2 \"Germans\"))\n",
+ " :ARG2 (m / member)))\n",
+ " :ARG2 (c / country\n",
+ " :name (n3 / name\n",
+ " :op1 \"France\")))\n"
+ ]
+ }
+ ],
+ "source": [
+ "print(amr_dict['Max who is a member of Some Germans is at France'])"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ " (b / be-located-at-91\n",
+ " :ARG1 (e / elephant\n",
+ " :ARG1-of (n / new-01))\n",
+ " :ARG2 (c / city\n",
+ " :name (n2 / name\n",
+ " :op1 \"Oklahoma\"))) \n",
+ " \n",
+ " (ii / include-91\n",
+ " :ARG1 (g / game\n",
+ " :name (n / name\n",
+ " :op1 \"Animal\"\n",
+ " :op2 \"Crossing\"))\n",
+ " :ARG2 (g2 / game\n",
+ " :ARG1-of (f / favor-01\n",
+ " :ARG0 (ii2 / i)\n",
+ " :time (k / kid)))) "
+ ]
+ },
{
"cell_type": "code",
"execution_count": null,
diff --git a/data/story_data_final_coref.json b/data/story_data_final_coref.json
index 49197eaf23662849c24e6d0359c54acfb3576842..2d7498059c07cbf3a43159647fcc87bc0c9b856b 100644
Binary files a/data/story_data_final_coref.json and b/data/story_data_final_coref.json differ
diff --git a/doc/22_01_12.md b/doc/22_01_12.md
index 20ae784254db6fd5a6e8ed4acdc6f8d8c1187127..eee0212d32b9833a85403a33494be01ee5466eaf 100644
--- a/doc/22_01_12.md
+++ b/doc/22_01_12.md
@@ -1,4 +1,11 @@
+PENMAN:
+=======
+
+- https://github.com/goodmami/penman/blob/main/docs/api-demo.ipynb
+- https://github.com/goodmami/penman
+- https://penman.readthedocs.io/en/latest/library.html
+
IDEAS:
======