From 91d10249fbf52bbd55750d1567c73cfcbc8c18ef Mon Sep 17 00:00:00 2001
From: BLADIER Tatiana <tatiana.bladier@lis-lab.fr>
Date: Fri, 16 May 2025 10:35:29 +0200
Subject: [PATCH] add readfile option
---
.../tania-some-other-metrics-checkpoint.ipynb | 241 ++++++++++++++----
tania_scripts/tania-some-other-metrics.ipynb | 241 ++++++++++++++----
2 files changed, 376 insertions(+), 106 deletions(-)
diff --git a/tania_scripts/.ipynb_checkpoints/tania-some-other-metrics-checkpoint.ipynb b/tania_scripts/.ipynb_checkpoints/tania-some-other-metrics-checkpoint.ipynb
index a918ace..0197b08 100644
--- a/tania_scripts/.ipynb_checkpoints/tania-some-other-metrics-checkpoint.ipynb
+++ b/tania_scripts/.ipynb_checkpoints/tania-some-other-metrics-checkpoint.ipynb
@@ -1015,7 +1015,7 @@
},
{
"cell_type": "code",
- "execution_count": 38,
+ "execution_count": 43,
"id": "673d6a76-42a0-4dcd-9c54-ae18710a032a",
"metadata": {},
"outputs": [
@@ -1023,12 +1023,6 @@
"name": "stdout",
"output_type": "stream",
"text": [
- "--\tponct\t<nul>@@<nul>\t0\n",
- "Le\tdet\tNP@@<nul>\t0\n",
- "parlement\tnc\t<nul>@@<nul>\t1\n",
- "...\tponct\t<nul>@@SENT\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" SENT \n",
@@ -1047,21 +1041,6 @@
" (ponct ...)\n",
" (</s> (</s> </s>))))\n",
"Tree depth: 5\n",
- "<s>\t<s>\t<s>\t1\n",
- "--\tponct\t<nul>@@<nul>\t0\n",
- "Vous\tpro\tVN@@<nul>\t0\n",
- ",\tponct\t<nul>@@<nul>\t1\n",
- "dit\tv\t<nul>@@<nul>\t1\n",
- "Athos\tnpp\tNP-OBJ@@Sint-MOD\t1\n",
- ",\tponct\t<nul>@@<nul>\t1\n",
- "je\tcls-suj\tVN@@<nul>\t1\n",
- "ne\tadv\t<nul>@@<nul>\t2\n",
- "me\tv\t<nul>@@<nul>\t2\n",
- "reconnais\tv\t<nul>@@<nul>\t2\n",
- "pas\tadv\t<nul>@@<nul>\t2\n",
- ".\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1086,14 +1065,6 @@
" (ponct .)\n",
" (</s> (</s> </s>))))\n",
"Tree depth: 5\n",
- "<s>\t<s>\t<s>\t1\n",
- "--\tponct\t<nul>@@<nul>\t1\n",
- "M.\tnpp\tNP-SUJ@@<nul>\t1\n",
- "d'\tp\tNPP+@@<nul>\t2\n",
- "Artagnan\tnpp\t<nul>@@<nul>\t3\n",
- ".\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1117,11 +1088,6 @@
" (ponct .)\n",
" (</s> (</s> </s>))))\n",
"Tree depth: 6\n",
- "<s>\t<s>\t<s>\t1\n",
- "Acté\tnc\tNP-OBJ@@<nul>\t1\n",
- "?\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1137,16 +1103,6 @@
"NLTK TREE (SENT\n",
" (<s> (<s> <s> (NP-OBJ (nc Acté))) (ponct ?) (</s> (</s> </s>))))\n",
"Tree depth: 5\n",
- "<s>\t<s>\t<s>\t1\n",
- "--\tponct\t<nul>@@<nul>\t1\n",
- "Oui\tadj\tAP-ATS@@<nul>\t1\n",
- ",\tponct\t<nul>@@<nul>\t1\n",
- "répondit\tv\tVN@@<nul>\t1\n",
- "le\tdet\tNP-OBJ@@Sint-MOD\t2\n",
- "mousquetaire\tnc\t<nul>@@<nul>\t3\n",
- ".\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1173,7 +1129,144 @@
" (NP-OBJ (det le) (nc mousquetaire))))\n",
" (ponct .)\n",
" (</s> (</s> </s>))))\n",
- "Tree depth: 6\n"
+ "Tree depth: 6\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " ________|_____________________________________________________________ \n",
+ " Sint-MOD | | \n",
+ " _______________________________|_____________________________________ | | \n",
+ " | | | | Ssub-OBJ | | \n",
+ " | | | | _____________|________ | | \n",
+ " <s> | | | | Sint | | \n",
+ " __________|__________ | | | | ________|_____________ | | \n",
+ " | | | | VN NP-OBJ | VN | VN | | </s>\n",
+ " | | | | | ____|_____ | _____|___ | _______|________________ | | | \n",
+ " | ponct adv ponct v det nc ponct cls-suj v cs cls-suj adv clo-obj v adv ponct </s>\n",
+ " | | | | | | | | | | | | | | | | | | \n",
+ "<s> -- Alors , dit le roi , il faut que vous ne le rendiez pas . </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (Sint-MOD\n",
+ " (<s> <s> (ponct --) (adv Alors) (ponct ,) (VN (v dit)))\n",
+ " (NP-OBJ (det le) (nc roi))\n",
+ " (ponct ,)\n",
+ " (VN (cls-suj il) (v faut))\n",
+ " (Ssub-OBJ\n",
+ " (cs que)\n",
+ " (Sint\n",
+ " (VN (cls-suj vous) (adv ne) (clo-obj le) (v rendiez))\n",
+ " (adv pas))))\n",
+ " (ponct .)\n",
+ " (</s> (</s> </s>))))\n",
+ "Tree depth: 7\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " _______|_________________________________________________________ \n",
+ " <s> | | \n",
+ " _________________________________________________|____________________ | | \n",
+ " | | | | VN | | \n",
+ " | | | | _____________________________________________|_________________ | | \n",
+ " | | | | | | | | | | | COORD | | \n",
+ " | | | | | | | | | | | ______|_____ | | \n",
+ " | | AP-ATS | | | AP-ATS | | NP-OBJ | | VN | </s>\n",
+ " | | | | | | | | | ______|_____ | | _____|________ | | \n",
+ " | ponct adv ponct cls-suj v adj ponct cls-suj clo-obj v ponct cc cls-suj clr v vpp ponct </s>\n",
+ " | | | | | | | | | | | | | | | | | | | \n",
+ "<s> -- Non , il est vrai , je le suis , mais il m' a semblé ... </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (<s>\n",
+ " <s>\n",
+ " (ponct --)\n",
+ " (AP-ATS (adv Non))\n",
+ " (ponct ,)\n",
+ " (VN\n",
+ " (cls-suj il)\n",
+ " (v est)\n",
+ " (AP-ATS (adj vrai))\n",
+ " (ponct ,)\n",
+ " (cls-suj je)\n",
+ " (NP-OBJ (clo-obj le) (v suis))\n",
+ " (ponct ,)\n",
+ " (COORD\n",
+ " (cc mais)\n",
+ " (VN (cls-suj il) (clr m') (v a) (vpp semblé)))))\n",
+ " (ponct ...)\n",
+ " (</s> (</s> </s>))))\n",
+ "Tree depth: 7\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " | \n",
+ " <s> \n",
+ " ____|____ \n",
+ " | | </s>\n",
+ " | | | \n",
+ " | ponct </s>\n",
+ " | | | \n",
+ "<s> -- </s>\n",
+ "\n",
+ "NLTK TREE (SENT (<s> (<s> <s> (ponct --) (</s> (</s> </s>)))))\n",
+ "Tree depth: 5\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " ___________________|________________________________ \n",
+ " Sint-MOD | Sint-MOD | | \n",
+ " _____________|______ | ______|_______ | | \n",
+ " | VN | VN NP-OBJ | </s>\n",
+ " | ______________|_____ | | | | | \n",
+ "<s> cls-suj v vpp ponct vinf adv ponct </s>\n",
+ " | | | | | | | | | \n",
+ "<s> Il a répondu : voilà tout . </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (Sint-MOD (<s> <s>) (VN (cls-suj Il) (v a) (vpp répondu)))\n",
+ " (ponct :)\n",
+ " (Sint-MOD (VN (vinf voilà)) (NP-OBJ (adv tout)))\n",
+ " (ponct .)\n",
+ " (</s> (</s> </s>))))\n",
+ "Tree depth: 5\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " | \n",
+ " Sint-MOD \n",
+ " _____________________________________|____________________ \n",
+ " | | | COORD \n",
+ " | | | __________________________|____________________ \n",
+ " | | | | | | | PP \n",
+ " | | | | | | | ______________|____ \n",
+ " | | | | | | | | VPinf \n",
+ " | | | | | | | | ____________|____ \n",
+ " | VN | | VN | AP-ATS | VN | </s>\n",
+ " | _____|________ | | ______|___ | | | | | | \n",
+ "<s> cls-suj adv adv v ponct cc cls-suj v adv adj p vinf ponct </s>\n",
+ " | | | | | | | | | | | | | | | \n",
+ "<s> Je ne puis dire , car il est bien facile à comprendre . </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (Sint-MOD\n",
+ " (<s> <s>)\n",
+ " (VN (cls-suj Je) (adv ne) (adv puis) (v dire))\n",
+ " (ponct ,)\n",
+ " (COORD\n",
+ " (cc car)\n",
+ " (VN (cls-suj il) (v est))\n",
+ " (adv bien)\n",
+ " (AP-ATS (adj facile))\n",
+ " (PP\n",
+ " (p à)\n",
+ " (VPinf (VN (vinf comprendre)) (ponct .) (</s> (</s> </s>))))))))\n",
+ "Tree depth: 8\n",
+ "-- Le parlement ... </s> <s> -- Vous , dit Athos , je ne me reconnais pas . </s> <s> -- M. d' Artagnan . </s> <s> Acté ? </s> <s> -- Oui , répondit le mousquetaire . </s> <s> -- Alors , dit le roi , il faut que vous ne le rendiez pas . </s> <s> -- Non , il est vrai , je le suis , mais il m' a semblé ... </s> <s> -- </s> <s> Il a répondu : voilà tout . </s> <s> Je ne puis dire , car il est bien facile à comprendre . </s> \n",
+ "0\n"
]
}
],
@@ -1182,15 +1275,21 @@
"\n",
"sents = readFile(file_conll)\n",
"exepts = []\n",
- "for sent in sents[:5]:\n",
+ "file_text = ''\n",
+ "for sent in sents[:10]:\n",
" sent_lst = ['\\t'.join(word)+'\\n' for word in sent]\n",
- " sent_str = ''.join(sent_lst)\n",
- " print(sent_str)\n",
+ " triple_str = ''.join(sent_lst)\n",
+ " sent_text = ' '.join([word[0] for word in sent]) + ' '\n",
+ " file_text += sent_text\n",
" try:\n",
- " visualize_const_prediction(sent_str)\n",
+ " visualize_const_prediction(triple_str)\n",
" except:\n",
- " exepts.append(sent_str)\n",
- "print(len(exepts))"
+ " exepts.append(triple_str)\n",
+ "print(file_text)\n",
+ "print(\"Number of sentences without a tree: \", len(exepts))\n",
+ "\n",
+ "for exept in exepts:\n",
+ " print(exept)"
]
},
{
@@ -1199,7 +1298,43 @@
"id": "729d7671-02be-443b-94ee-ddc79284aac4",
"metadata": {},
"outputs": [],
- "source": []
+ "source": [
+ "def prepare_file_to_evaluate(file_conll):\n",
+ " sents = readFile(file_conll)\n",
+ " file_text = ''\n",
+ " triple_strings = []\n",
+ " for sent in sents[:10]:\n",
+ " sent_lst = ['\\t'.join(word)+'\\n' for word in sent]\n",
+ " triple_str = ''.join(sent_lst)\n",
+ " sent_text = ' '.join([word[0] for word in sent]) + ' '\n",
+ " file_text += sent_text\n",
+ " triple_strings.append(triple_str)\n",
+ " return file_text, triple_strings\n",
+ " "
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "cefe6712-ab53-4589-9fb8-8d4c3fb00378",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "file_text, triple_strings = prepare_file_to_evaluate(file_conll)\n",
+ "\n",
+ "exepts = [] \n",
+ "\n",
+ "for triple_str in triple_strings:\n",
+ " try:\n",
+ " visualize_const_prediction(triple_str)\n",
+ " except:\n",
+ " exepts.append(triple_str)\n",
+ "print(\"Number of sentences without a tree: \", len(exepts))\n",
+ " \n",
+ "for exept in exepts:\n",
+ " print(exept)\n",
+ " \n"
+ ]
}
],
"metadata": {
diff --git a/tania_scripts/tania-some-other-metrics.ipynb b/tania_scripts/tania-some-other-metrics.ipynb
index a918ace..0197b08 100644
--- a/tania_scripts/tania-some-other-metrics.ipynb
+++ b/tania_scripts/tania-some-other-metrics.ipynb
@@ -1015,7 +1015,7 @@
},
{
"cell_type": "code",
- "execution_count": 38,
+ "execution_count": 43,
"id": "673d6a76-42a0-4dcd-9c54-ae18710a032a",
"metadata": {},
"outputs": [
@@ -1023,12 +1023,6 @@
"name": "stdout",
"output_type": "stream",
"text": [
- "--\tponct\t<nul>@@<nul>\t0\n",
- "Le\tdet\tNP@@<nul>\t0\n",
- "parlement\tnc\t<nul>@@<nul>\t1\n",
- "...\tponct\t<nul>@@SENT\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" SENT \n",
@@ -1047,21 +1041,6 @@
" (ponct ...)\n",
" (</s> (</s> </s>))))\n",
"Tree depth: 5\n",
- "<s>\t<s>\t<s>\t1\n",
- "--\tponct\t<nul>@@<nul>\t0\n",
- "Vous\tpro\tVN@@<nul>\t0\n",
- ",\tponct\t<nul>@@<nul>\t1\n",
- "dit\tv\t<nul>@@<nul>\t1\n",
- "Athos\tnpp\tNP-OBJ@@Sint-MOD\t1\n",
- ",\tponct\t<nul>@@<nul>\t1\n",
- "je\tcls-suj\tVN@@<nul>\t1\n",
- "ne\tadv\t<nul>@@<nul>\t2\n",
- "me\tv\t<nul>@@<nul>\t2\n",
- "reconnais\tv\t<nul>@@<nul>\t2\n",
- "pas\tadv\t<nul>@@<nul>\t2\n",
- ".\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1086,14 +1065,6 @@
" (ponct .)\n",
" (</s> (</s> </s>))))\n",
"Tree depth: 5\n",
- "<s>\t<s>\t<s>\t1\n",
- "--\tponct\t<nul>@@<nul>\t1\n",
- "M.\tnpp\tNP-SUJ@@<nul>\t1\n",
- "d'\tp\tNPP+@@<nul>\t2\n",
- "Artagnan\tnpp\t<nul>@@<nul>\t3\n",
- ".\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1117,11 +1088,6 @@
" (ponct .)\n",
" (</s> (</s> </s>))))\n",
"Tree depth: 6\n",
- "<s>\t<s>\t<s>\t1\n",
- "Acté\tnc\tNP-OBJ@@<nul>\t1\n",
- "?\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1137,16 +1103,6 @@
"NLTK TREE (SENT\n",
" (<s> (<s> <s> (NP-OBJ (nc Acté))) (ponct ?) (</s> (</s> </s>))))\n",
"Tree depth: 5\n",
- "<s>\t<s>\t<s>\t1\n",
- "--\tponct\t<nul>@@<nul>\t1\n",
- "Oui\tadj\tAP-ATS@@<nul>\t1\n",
- ",\tponct\t<nul>@@<nul>\t1\n",
- "répondit\tv\tVN@@<nul>\t1\n",
- "le\tdet\tNP-OBJ@@Sint-MOD\t2\n",
- "mousquetaire\tnc\t<nul>@@<nul>\t3\n",
- ".\tponct\t<nul>@@<nul>\t0\n",
- "</s>\t</s>\t</s>\t0\n",
- "\n",
" SENT \n",
" | \n",
" <s> \n",
@@ -1173,7 +1129,144 @@
" (NP-OBJ (det le) (nc mousquetaire))))\n",
" (ponct .)\n",
" (</s> (</s> </s>))))\n",
- "Tree depth: 6\n"
+ "Tree depth: 6\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " ________|_____________________________________________________________ \n",
+ " Sint-MOD | | \n",
+ " _______________________________|_____________________________________ | | \n",
+ " | | | | Ssub-OBJ | | \n",
+ " | | | | _____________|________ | | \n",
+ " <s> | | | | Sint | | \n",
+ " __________|__________ | | | | ________|_____________ | | \n",
+ " | | | | VN NP-OBJ | VN | VN | | </s>\n",
+ " | | | | | ____|_____ | _____|___ | _______|________________ | | | \n",
+ " | ponct adv ponct v det nc ponct cls-suj v cs cls-suj adv clo-obj v adv ponct </s>\n",
+ " | | | | | | | | | | | | | | | | | | \n",
+ "<s> -- Alors , dit le roi , il faut que vous ne le rendiez pas . </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (Sint-MOD\n",
+ " (<s> <s> (ponct --) (adv Alors) (ponct ,) (VN (v dit)))\n",
+ " (NP-OBJ (det le) (nc roi))\n",
+ " (ponct ,)\n",
+ " (VN (cls-suj il) (v faut))\n",
+ " (Ssub-OBJ\n",
+ " (cs que)\n",
+ " (Sint\n",
+ " (VN (cls-suj vous) (adv ne) (clo-obj le) (v rendiez))\n",
+ " (adv pas))))\n",
+ " (ponct .)\n",
+ " (</s> (</s> </s>))))\n",
+ "Tree depth: 7\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " _______|_________________________________________________________ \n",
+ " <s> | | \n",
+ " _________________________________________________|____________________ | | \n",
+ " | | | | VN | | \n",
+ " | | | | _____________________________________________|_________________ | | \n",
+ " | | | | | | | | | | | COORD | | \n",
+ " | | | | | | | | | | | ______|_____ | | \n",
+ " | | AP-ATS | | | AP-ATS | | NP-OBJ | | VN | </s>\n",
+ " | | | | | | | | | ______|_____ | | _____|________ | | \n",
+ " | ponct adv ponct cls-suj v adj ponct cls-suj clo-obj v ponct cc cls-suj clr v vpp ponct </s>\n",
+ " | | | | | | | | | | | | | | | | | | | \n",
+ "<s> -- Non , il est vrai , je le suis , mais il m' a semblé ... </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (<s>\n",
+ " <s>\n",
+ " (ponct --)\n",
+ " (AP-ATS (adv Non))\n",
+ " (ponct ,)\n",
+ " (VN\n",
+ " (cls-suj il)\n",
+ " (v est)\n",
+ " (AP-ATS (adj vrai))\n",
+ " (ponct ,)\n",
+ " (cls-suj je)\n",
+ " (NP-OBJ (clo-obj le) (v suis))\n",
+ " (ponct ,)\n",
+ " (COORD\n",
+ " (cc mais)\n",
+ " (VN (cls-suj il) (clr m') (v a) (vpp semblé)))))\n",
+ " (ponct ...)\n",
+ " (</s> (</s> </s>))))\n",
+ "Tree depth: 7\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " | \n",
+ " <s> \n",
+ " ____|____ \n",
+ " | | </s>\n",
+ " | | | \n",
+ " | ponct </s>\n",
+ " | | | \n",
+ "<s> -- </s>\n",
+ "\n",
+ "NLTK TREE (SENT (<s> (<s> <s> (ponct --) (</s> (</s> </s>)))))\n",
+ "Tree depth: 5\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " ___________________|________________________________ \n",
+ " Sint-MOD | Sint-MOD | | \n",
+ " _____________|______ | ______|_______ | | \n",
+ " | VN | VN NP-OBJ | </s>\n",
+ " | ______________|_____ | | | | | \n",
+ "<s> cls-suj v vpp ponct vinf adv ponct </s>\n",
+ " | | | | | | | | | \n",
+ "<s> Il a répondu : voilà tout . </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (Sint-MOD (<s> <s>) (VN (cls-suj Il) (v a) (vpp répondu)))\n",
+ " (ponct :)\n",
+ " (Sint-MOD (VN (vinf voilà)) (NP-OBJ (adv tout)))\n",
+ " (ponct .)\n",
+ " (</s> (</s> </s>))))\n",
+ "Tree depth: 5\n",
+ " SENT \n",
+ " | \n",
+ " <s> \n",
+ " | \n",
+ " Sint-MOD \n",
+ " _____________________________________|____________________ \n",
+ " | | | COORD \n",
+ " | | | __________________________|____________________ \n",
+ " | | | | | | | PP \n",
+ " | | | | | | | ______________|____ \n",
+ " | | | | | | | | VPinf \n",
+ " | | | | | | | | ____________|____ \n",
+ " | VN | | VN | AP-ATS | VN | </s>\n",
+ " | _____|________ | | ______|___ | | | | | | \n",
+ "<s> cls-suj adv adv v ponct cc cls-suj v adv adj p vinf ponct </s>\n",
+ " | | | | | | | | | | | | | | | \n",
+ "<s> Je ne puis dire , car il est bien facile à comprendre . </s>\n",
+ "\n",
+ "NLTK TREE (SENT\n",
+ " (<s>\n",
+ " (Sint-MOD\n",
+ " (<s> <s>)\n",
+ " (VN (cls-suj Je) (adv ne) (adv puis) (v dire))\n",
+ " (ponct ,)\n",
+ " (COORD\n",
+ " (cc car)\n",
+ " (VN (cls-suj il) (v est))\n",
+ " (adv bien)\n",
+ " (AP-ATS (adj facile))\n",
+ " (PP\n",
+ " (p à)\n",
+ " (VPinf (VN (vinf comprendre)) (ponct .) (</s> (</s> </s>))))))))\n",
+ "Tree depth: 8\n",
+ "-- Le parlement ... </s> <s> -- Vous , dit Athos , je ne me reconnais pas . </s> <s> -- M. d' Artagnan . </s> <s> Acté ? </s> <s> -- Oui , répondit le mousquetaire . </s> <s> -- Alors , dit le roi , il faut que vous ne le rendiez pas . </s> <s> -- Non , il est vrai , je le suis , mais il m' a semblé ... </s> <s> -- </s> <s> Il a répondu : voilà tout . </s> <s> Je ne puis dire , car il est bien facile à comprendre . </s> \n",
+ "0\n"
]
}
],
@@ -1182,15 +1275,21 @@
"\n",
"sents = readFile(file_conll)\n",
"exepts = []\n",
- "for sent in sents[:5]:\n",
+ "file_text = ''\n",
+ "for sent in sents[:10]:\n",
" sent_lst = ['\\t'.join(word)+'\\n' for word in sent]\n",
- " sent_str = ''.join(sent_lst)\n",
- " print(sent_str)\n",
+ " triple_str = ''.join(sent_lst)\n",
+ " sent_text = ' '.join([word[0] for word in sent]) + ' '\n",
+ " file_text += sent_text\n",
" try:\n",
- " visualize_const_prediction(sent_str)\n",
+ " visualize_const_prediction(triple_str)\n",
" except:\n",
- " exepts.append(sent_str)\n",
- "print(len(exepts))"
+ " exepts.append(triple_str)\n",
+ "print(file_text)\n",
+ "print(\"Number of sentences without a tree: \", len(exepts))\n",
+ "\n",
+ "for exept in exepts:\n",
+ " print(exept)"
]
},
{
@@ -1199,7 +1298,43 @@
"id": "729d7671-02be-443b-94ee-ddc79284aac4",
"metadata": {},
"outputs": [],
- "source": []
+ "source": [
+ "def prepare_file_to_evaluate(file_conll):\n",
+ " sents = readFile(file_conll)\n",
+ " file_text = ''\n",
+ " triple_strings = []\n",
+ " for sent in sents[:10]:\n",
+ " sent_lst = ['\\t'.join(word)+'\\n' for word in sent]\n",
+ " triple_str = ''.join(sent_lst)\n",
+ " sent_text = ' '.join([word[0] for word in sent]) + ' '\n",
+ " file_text += sent_text\n",
+ " triple_strings.append(triple_str)\n",
+ " return file_text, triple_strings\n",
+ " "
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "cefe6712-ab53-4589-9fb8-8d4c3fb00378",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "file_text, triple_strings = prepare_file_to_evaluate(file_conll)\n",
+ "\n",
+ "exepts = [] \n",
+ "\n",
+ "for triple_str in triple_strings:\n",
+ " try:\n",
+ " visualize_const_prediction(triple_str)\n",
+ " except:\n",
+ " exepts.append(triple_str)\n",
+ "print(\"Number of sentences without a tree: \", len(exepts))\n",
+ " \n",
+ "for exept in exepts:\n",
+ " print(exept)\n",
+ " \n"
+ ]
}
],
"metadata": {
--
GitLab