forked from yago-naga/yago-4.5
-
Notifications
You must be signed in to change notification settings - Fork 0
/
06-make-statistics.py
252 lines (223 loc) · 11 KB
/
06-make-statistics.py
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
242
243
244
245
246
247
248
249
250
251
252
"""
Produces statistics about YAGO entities and predicates, and extracts samples
CC-BY 2022 Fabian M. Suchanek
Input:
- 01-yago-final-schema.ttl
- 05-yago-final-beyond-wikipedia.tsv
- 05-yago-final-wikipedia.tsv
- 05-yago-final-taxonomy.tsv
Output:
- 06-statistics.txt
- 06-taxonomy.html
- 06-upper-taxonomy.html
- 06-sample-entities.ttl
Algorithm:
- load taxonomy
- run through yago-final-full
- update statistics
- sample entities
- print statistics and trees
"""
TEST=False
FOLDER="test-data/06-make-statistics/" if TEST else "yago-data/"
##########################################################################
# Booting
##########################################################################
import sys
import glob
import re
import evaluator
import itertools
import TurtleUtils
import TsvUtils
import random
import os
import Prefixes
from collections import defaultdict
# Predicates that are excluded for fact counting
excludePredicates=["rdfs:label", "rdfs:comment", "rdf:type", "schema:mainEntityOfPage", "owl:sameAs", "schema:alternateName"]
##########################################################################
# Full Taxonomy as HTML
##########################################################################
def getSuperClasses(cls, classes, yagoTaxonomyUp, pathsToRoot):
"""Adds all superclasses of a class <cls> (including <cls>) to the set <classes>"""
classes.add(cls)
# Make a check before because it's a defaultdict,
# which would create cls if it's not there
if cls==Prefixes.schemaThing:
pathsToRoot[0]+=1
if cls in yagoTaxonomyUp:
for sc in yagoTaxonomyUp[cls]:
getSuperClasses(sc, classes, yagoTaxonomyUp, pathsToRoot)
def printTaxonomy2(writer, cls=Prefixes.schemaThing):
""" Prints the taxonomy to the writer. <cls> is the class to start with, i.e., the top-level class. """
if cls not in yagoTaxonomyDown:
writer.write(f"<li>{cls.replace('yago:','y:')}: {str(classStats.get(cls,0))}\n")
return
writer.write(f"<li><details style='margin-left: 2em'><summary style='margin-left: -2em'>{cls.replace('yago:','y:')}: {str(classStats.get(cls,0))}</summary><ul>\n")
for subclass in yagoTaxonomyDown.get(cls, []):
printTaxonomy2(writer, subclass)
writer.write("</ul></details>\n")
def printTaxonomy(file):
""" Prints the full taxonomy to the file """
with open(file, "wt", encoding="UTF-8") as writer:
writer.write("""
<!DOCTYPE html>
<html>
<head>
<meta charset=utf-8>
<meta name=viewport content="width=device-width, initial-scale=1.0">
<title>
YAGO Taxonomy
</title>
<style>
ul {list-style-type:none}
</style>
</head>
<body>
<h1>YAGO Taxonomy</h1>
<ul>\n""")
printTaxonomy2(writer)
writer.write("</ul></body>\n</html>")
##########################################################################
# Top-level taxonomy as HTML
##########################################################################
def printUpperTaxonomy(file):
""" Visualizes the top-level taxonomy as an HTML document"""
with open(file, "wt", encoding="UTF-8") as writer:
writer.write("""
<h1>YAGO Schema</h1>
This is the top-level taxonomy of classes of YAGO 4.5, together with their properties.
<ul style='list-style-type: none'>
""")
def add_node(cls):
if not yagoSchema.objects(cls):
return
writer.write(f"<li><details style='margin-left: 2em'{' open' if cls=='schema:Thing' else ''}><summary style='font-weight:bold; margin-left: -2em'>{cls}</summary><details style='margin-left: 2em'><summary style='margin-left: -2em'>Outgoing properties</summary><ul style='list-style-type: none'>\n")
for blank in sorted(yagoSchema.objects(cls,Prefixes.shaclProperty), key=lambda b: yagoSchema.objects(b, Prefixes.shaclPath)):
writer.write(f'<li>- {yagoSchema.objects(blank, Prefixes.shaclPath)[0]}')
if yagoSchema.objects(blank, Prefixes.shaclNode):
ran=yagoSchema.objects(blank, Prefixes.shaclNode)[0]
elif yagoSchema.objects(blank, Prefixes.shaclOr):
ran=", ".join([target for el in yagoSchema.getList(yagoSchema.objects(blank, Prefixes.shaclOr)[0]) for target in yagoSchema.objects(el, Prefixes.shaclNode) + yagoSchema.objects(el, Prefixes.shaclDatatype)])
elif yagoSchema.objects(blank, Prefixes.shaclDatatype):
ran=yagoSchema.objects(blank, Prefixes.shaclDatatype)[0]
if ran:
writer.write(" →")
maxCount=yagoSchema.objects(blank, Prefixes.shaclMaxCount)
if maxCount:
writer.write("<sup>"+maxCount[0]+"</sup>")
elif yagoSchema.objects(blank, Prefixes.shaclUniqueLang):
writer.write("<sup>1 per language</sup>")
writer.write(" "+ran+"\n")
writer.write("</ul></details>\n<details style='margin-left: 2em'><summary style='margin-left: -2em'>Incoming properties</summary><ul style='list-style-type: none'>\n")
for blank in sorted(yagoSchema.subjects(Prefixes.shaclNode, cls)):
p=yagoSchema.objects(blank, Prefixes.shaclPath)
c=yagoSchema.subjects(Prefixes.shaclProperty,blank)
if p and c:
writer.write(f'<li>- ({c[0]}) {p[0]}')
writer.write(f"</ul></details><details style='margin-left: 2em'{' open' if cls=='schema:Thing' else ''}><summary style='margin-left: -2em'>Subclasses</summary><ul style='list-style-type: none'>\n")
for subclass in sorted(yagoSchema.subjects(Prefixes.rdfsSubClassOf, cls)):
add_node(subclass)
writer.write("</ul></details></details>\n")
add_node("schema:Thing")
writer.write("</ul>")
##########################################################################
# Main
##########################################################################
with TsvUtils.Timer("Step 06: Collecting YAGO statistics"):
# Load YAGO schema
yagoSchema = TurtleUtils.Graph()
yagoSchema.loadTurtleFile(FOLDER+"01-yago-final-schema.ttl", " Loading YAGO schema")
# Load YAGO taxonomy
yagoTaxonomyDown=defaultdict(set)
yagoTaxonomyUp=defaultdict(set)
for triple in TsvUtils.tsvTuples(FOLDER+"05-yago-final-taxonomy.tsv", " Loading YAGO taxonomy"):
if len(triple)>3:
yagoTaxonomyUp[triple[0]].add(triple[2])
yagoTaxonomyDown[triple[2]].add(triple[0])
# Initialize counters
predicateStats=defaultdict(int)
classStats=defaultdict(int)
samples=[]
entities=0
totalPathsToRoot=0
totalClassesPerInstance=0
humanReadableNames=0
# Initialize predicateStats with predicates from schema, same for classes
for s, p, o in yagoSchema.triplesWithPredicate(Prefixes.shaclPath):
predicateStats[o]=0
for s, p, o in yagoSchema.triplesWithPredicate("rdfs:subClassOf"):
classStats[s]=0
genericInstancesCount=0
# Run through the facts
for entityFacts in itertools.chain(TurtleUtils.tsvEntities(FOLDER+"05-yago-final-wikipedia.tsv", " Parsing YAGO Wikipedia"), TurtleUtils.tsvEntities(FOLDER+"05-yago-final-beyond-wikipedia.tsv", " Parsing YAGO beyond Wikipedia")):
classes=set()
subject=None
for s, p, o in entityFacts:
predicateStats[p]+=1
if p=='rdf:type':
classes.add(o)
if s.endswith("_generic_instance"):
genericInstancesCount+=1
subject=s
if subject is None or "rdfs:Class" in classes:
continue
entities+=1
superClasses=set()
pathsToRoot=[0]
if not re.match(r"yago:Q[0-9]+", subject):
humanReadableNames+=1
for c in classes:
getSuperClasses(c, superClasses, yagoTaxonomyUp, pathsToRoot)
for c in superClasses:
classStats[c]+=1
totalClassesPerInstance+=len(superClasses)
totalPathsToRoot+=pathsToRoot[0]
if (len(samples)<100 or (len(samples)==100 and random.random()<0.01)):
for c in superClasses:
entityFacts.add((subject, 'rdf:type', c))
if len(samples)<100:
samples.append(entityFacts)
else:
samples[int(random.random()*99)]=entityFacts
print(" Writing out sample entities... ",end="",flush=True)
with open(FOLDER+"06-sample-entities.ttl", "wt", encoding="UTF-8") as sampleFile:
for sample in samples:
sample.printToWriter(sampleFile)
print("done")
metaFacts=0
for triple in TsvUtils.tsvTuples(FOLDER+"05-yago-final-meta.tsv", " Counting meta facts"):
metaFacts += 1
print(" Computing dump size... ",end="",flush=True)
dumpSize=0
for f in glob.glob(FOLDER+"*final*.tsv"):
dumpSize+=os.path.getsize(f)
print("done")
print(" Writing out statistics... ",end="",flush=True)
with open(FOLDER+"06-statistics.txt", "wt", encoding="UTF-8") as writer:
writer.write("YAGO 4.5 statistics\n\n")
writer.write("Dump size: "+str(dumpSize/1024/1024/1024)+" GB\n\n")
writer.write("Total number of entities: "+str(entities)+"\n\n")
writer.write(" ... of which generic: "+str(genericInstancesCount)+"\n\n")
writer.write("Total number of classes: "+str(len(yagoTaxonomyUp))+"\n\n")
writer.write("Disjointness statements: "+str(len(yagoSchema.triplesWithPredicate(Prefixes.owlDisjointWith)))+"\n\n")
writer.write("Avg number of paths to root: "+str(totalPathsToRoot/entities)+"\n\n")
writer.write("Avg number of classes per instance: "+str(totalClassesPerInstance/entities)+"\n\n")
writer.write("Human-readable names: "+str(humanReadableNames*100.0/entities)+"%\n\n")
writer.write("Total number of facts (excluding labels etc.): "+str(sum([predicateStats[p] for p in predicateStats if p not in excludePredicates]))+"\n\n")
writer.write("Avg number of facts (excluding labels etc.) per entity: "+str(sum([predicateStats[p] for p in predicateStats if p not in excludePredicates])/entities)+"\n\n")
writer.write("Total number of meta facts: "+str(metaFacts)+"\n\n")
writer.write("Total number of predicates: "+str(len(predicateStats))+"\n\n")
writer.write("Predicates:\n")
for pred in sorted(predicateStats.items(), key=lambda x:-x[1]):
writer.write(" "+pred[0]+": "+str(pred[1])+"\n")
print("done")
print(" Writing out taxonomy... ",end="",flush=True)
printTaxonomy(FOLDER+"06-taxonomy.html")
printUpperTaxonomy(FOLDER+"06-upper-taxonomy.html")
print("done")
if TEST:
evaluator.compare(FOLDER+"06-statistics.txt", FOLDER+"06-statistics-gold.txt")
evaluator.compare(FOLDER+"06-taxonomy.html", FOLDER+"06-taxonomy-gold.html")
evaluator.compare(FOLDER+"06-upper-taxonomy.html", FOLDER+"06-upper-taxonomy-gold.html")