1 |
33136
|
claudio.at
|
pace.conf {
|
2 |
|
|
clustering {
|
3 |
|
|
acronyms { fields = [title, desc], params = { max = 1, minLen = 2, maxLen = 4} },
|
4 |
|
|
ngrampairs { fields = [title], params = { max = 1, ngramLen = 3} },
|
5 |
|
|
suffixprefix { fields = [title], params = { max = 1, len = 3 } }
|
6 |
|
|
},
|
7 |
|
|
conditions {
|
8 |
|
|
titleVersionMatch { fields = [title] },
|
9 |
|
|
sizeMatch { fields = [authors] }
|
10 |
|
|
},
|
11 |
|
|
model {
|
12 |
|
|
title { algo = JaroWinkler, type = String, weight = 0.5, ignoreMissing = false, path = result/metadata/title/value },
|
13 |
|
|
authors { algo = SortedLevel2JaroWinkler, type = List, weight = 0.5, ignoreMissing = true, path = result/author/metadata/fullname/value }
|
14 |
|
|
},
|
15 |
|
|
blacklists = {
|
16 |
|
|
title = [
|
17 |
|
|
"^(Corpus Oral Dialectal \\(COD\\)\\.).*$",
|
18 |
|
|
"^(Kiri Karl Morgensternile).*$",
|
19 |
|
|
"^(\\[Eksliibris Aleksandr).*\\]$",
|
20 |
|
|
"^(\\[Eksliibris Aleksandr).*$",
|
21 |
|
|
"^(Eksliibris Aleksandr).*$",
|
22 |
|
|
"^(Kiri A\\. de Vignolles).*$",
|
23 |
|
|
"^(2 kirja Karl Morgensternile).*$",
|
24 |
|
|
"^(Pirita kloostri idaosa arheoloogilised).*$",
|
25 |
|
|
"^(Kiri tundmatule).*$",
|
26 |
|
|
"^(Kiri Jenaer Allgemeine Literaturzeitung toimetusele).*$",
|
27 |
|
|
"^(Eksliibris Nikolai Birukovile).*$",
|
28 |
|
|
"^(Eksliibris Nikolai Issakovile).*$",
|
29 |
|
|
"^(WHP Cruise Summary Information of section).*$",
|
30 |
|
|
"^(Measurement of the top quark\\-pair production cross section with ATLAS in pp collisions at).*$",
|
31 |
|
|
"^(Measurement of the spin\\-dependent structure function).*"
|
32 |
|
|
] }
|
33 |
|
|
}
|