Project

General

Profile

1
{ 
2
	"wf" : { 
3
        "threshold" : "0.99", 
4
        "run" : "001", 
5
        "entityType" : "result", 
6
        "orderField" : "title", 
7
        "queueMaxSize" : "2000",
8
        "groupMaxSize" : "10",
9
        "slidingWindowSize" : "200",
10
        "rootBuilder" : [ "result" ],
11
        "includeChildren" : "true" 
12
    },
13
	"pace" : {		
14
		"clustering" : [
15
			{ "name" : "acronyms", "fields" : [ "title" ], "params" : { "max" : "1", "minLen" : "2", "maxLen" : "4"} },
16
			{ "name" : "ngrampairs", "fields" : [ "title" ], "params" : { "max" : "1", "ngramLen" : "3"} },
17
			{ "name" : "suffixprefix", "fields" : [ "title" ], "params" : { "max" : "1", "len" : "3" } } 
18
		],		
19
		"strictConditions" : [
20
  			{ "name" : "exactMatch", "fields" : [ "pid" ] }
21
  		], 
22
  		"conditions" : [ 
23
  			{ "name" : "yearMatch", "fields" : [ "dateofacceptance" ] },
24
  			{ "name" : "titleVersionMatch", "fields" : [ "title" ] },
25
  			{ "name" : "sizeMatch", "fields" : [ "authors" ] } 
26
  		],		
27
		"model" : [
28
			{ "name" : "pid", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "pid[qualifier#classid = {doi}]/value", "overrideMatch" : "true" }, 	
29
			{ "name" : "title", "algo" : "JaroWinkler", "type" : "String", "weight" : "1.0", "ignoreMissing" : "false", "path" : "result/metadata/title[qualifier#classid = {main title}]/value" },
30
			{ "name" : "dateofacceptance", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "result/metadata/dateofacceptance/value" } ,
31
			{ "name" : "authors", "algo" : "Null", "type" : "List", "weight" : "0.0", "ignoreMissing" : "true", "path" : "result/author/metadata/fullname/value" },
32
		  {
33
			"name": "anchors",
34
			"algo": "PersonCoAnchorsDistance",
35
			"type": "JSON",
36
			"weight": "0.0",
37
			"ignoreMissing": "true",
38
			"path": "person",
39
			"params": {
40
			  "common.anchors": "1"
41
			}
42
		  },
43
		  {
44
			"name": "coauthor",
45
			"algo": "PersonCoAuthorSurnamesDistance",
46
			"type": "JSON",
47
			"weight": "0.0",
48
			"ignoreMissing": "true",
49
			"path": "person",
50
			"params": {
51
			  "common.surnames": "2"
52
			}
53
		  },
54
		  {
55
			"name": "person",
56
			"algo": "PersonDistance",
57
			"type": "JSON",
58
			"weight": "0.0",
59
			"ignoreMissing": "true",
60
			"path": "person",
61
			"params": {
62
			  "common.surnames": "2"
63
			}
64
		  }
65
		],
66
		"blacklists" : {
67
			"title" : [
68
				"^(Corpus Oral Dialectal \\(COD\\)\\.).*$",
69
				"^(Kiri Karl Morgensternile).*$",
70
				"^(\\[Eksliibris Aleksandr).*\\]$",
71
				"^(\\[Eksliibris Aleksandr).*$",
72
				"^(Eksliibris Aleksandr).*$",
73
				"^(Kiri A\\. de Vignolles).*$",
74
				"^(2 kirja Karl Morgensternile).*$",
75
				"^(Pirita kloostri idaosa arheoloogilised).*$",
76
				"^(Kiri tundmatule).*$",
77
				"^(Kiri Jenaer Allgemeine Literaturzeitung toimetusele).*$",
78
				"^(Eksliibris Nikolai Birukovile).*$",
79
				"^(Eksliibris Nikolai Issakovile).*$",
80
				"^(WHP Cruise Summary Information of section).*$",
81
				"^(Measurement of the top quark\\-pair production cross section with ATLAS in pp collisions at).*$",
82
				"^(Measurement of the spin\\-dependent structure function).*"
83
			] } 		
84
	}
85

    
86
}
(4-4/5)