cc-dbp/config.properties at master · IBM/cc-dbp · GitHub

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
#CommonCrawlConfig

language=en
minLanguageConfidence=0.8
numThreads=8
#save these types as offset annotation in the corpus
annotationTypes = [LinkAnnotation]
urlPrefix = https://commoncrawl.s3.amazonaws.com/

#support downloading only a portion of Common Crawl
#warcFileLimit=1000


#DBpediaKBConfig

dbpediaOwlUrl=http://downloads.dbpedia.org/2016-10/dbpedia_2016-10.owl
objectsUrl=http://downloads.dbpedia.org/2016-10/core-i18n/en/mappingbased_objects_en.ttl.bz2
literalsUrl=http://downloads.dbpedia.org/2016-10/core-i18n/en/mappingbased_literals_en.ttl.bz2
labelsUrl=http://downloads.dbpedia.org/2016-10/core-i18n/en/labels_en.ttl.bz2
typesUrl=http://downloads.dbpedia.org/2016-10/core-i18n/en/instance_types_transitive_en.ttl.bz2

#for ground truth
maxNodeCorpusCount = 500000
minNodeCorpusCount = 1
useRelationTaxonomy = True

#for the coarse-grained type system
minTypeSize = 3000
maxNumberOfTypes = 100

#for the type filter
minTypePairFreq = 1

noNodeCorpusCounts = False


#RelexConfig

documentSampleFraction = 1.0
negativeExampleSampleFraction = 0.05
targetNegativeToPositveRatio = -1
directionStyle = bothWays
titleContext = False
sectionContext = False
limitEntitiesToGroundTruth = False
gtTypes = False
vocabLimit = 2000000
vocabMinCount = 2
minMentionSet = 1
maxMentionSet = 100
maxMentionGroups = 5
maxPositionEmbeddings = 80
typeStyle = single
datasetSplitNames = [train, validate, test]
datasetSpitFractions = [0.8, 0.1, 0.1]
entityPairFilterClass = com.ibm.research.ai.ki.kbp.TypePairEntityPairFilter