Skip to content

Commit

Permalink
a last pass at name filters...
Browse files Browse the repository at this point in the history
  • Loading branch information
mubaldino committed Mar 5, 2024
1 parent c74e9f0 commit 3cbccca
Showing 1 changed file with 36 additions and 22 deletions.
58 changes: 36 additions & 22 deletions solr/etc/gazetteer/filters/non-placenames.csv
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
exclusion,category
exclusion,category
a.k.a,abbrev
ab,abbrev
acme,abbrev
Expand Down Expand Up @@ -72,11 +72,11 @@ pers,abbrev
pharma,abbrev
pjs,abbrev
po,abbrev
pty,abbrev
pre,abbrev
proj,abbrev
pros,abbrev
pty,abbrev
pty,abbrev
re,abbrev
rec,abbrev
recap,abbrev
Expand Down Expand Up @@ -574,6 +574,7 @@ buona,adj.ita
de novo,adj.ita
dolce,adj.ita
forte,adj.ita
nazionale,adj.ita
novo,adj.ita
santi,adj.ita
boa,adj.por
Expand Down Expand Up @@ -674,8 +675,10 @@ who,article
moi,article.fre
qui,article.fre
sans,article.fre
c�,article.ita
che,article.ita
cé,article.ita
c'è,article.ita
cè,article.ita
gli,article.ita
lui,article.ita
tra,article.ita
Expand Down Expand Up @@ -918,6 +921,7 @@ blueberry,food
burgers,food
butternut,food
cake,food
caviar,food
chai,food
chili,food
chive,food
Expand All @@ -935,6 +939,7 @@ flour,food
food,food
fries,food
grape,food
honeycomb,food
jam,food
latte,food
lattes,food
Expand Down Expand Up @@ -1115,8 +1120,8 @@ kana,language
kanji,language
norsk,language
português,language
việt,language
viét,language
việt,language
cana,latin
canis,latin
coli,latin
Expand All @@ -1125,6 +1130,7 @@ datum,latin
deus,latin
gratis,latin
homo,latin
modus,latin
pas,latin
philo,latin
prima,latin
Expand Down Expand Up @@ -1457,6 +1463,7 @@ check,noun
cheek,noun
chef,noun
chemistry,noun
cheque,noun
chess board,noun
childs,noun
chimney,noun
Expand Down Expand Up @@ -1660,6 +1667,7 @@ elders,noun
electron,noun
elf,noun
embargo,noun
emblem,noun
eminence,noun
encore,noun
end,noun
Expand Down Expand Up @@ -1928,6 +1936,7 @@ labor,noun
laboratory,noun
labour,noun
lack,noun
lading,noun
lamps,noun
landfall,noun
landing,noun
Expand Down Expand Up @@ -1999,9 +2008,9 @@ mailbox,noun
main camp,noun
maker,noun
man,noun
mania,noun
management,noun
mange,noun
mania,noun
manifesto,noun
manifold,noun
manner,noun
Expand Down Expand Up @@ -2087,6 +2096,7 @@ motto,noun
mounts,noun
mouse,noun
mud,noun
mug,noun
mural,noun
muscle,noun
muse,noun
Expand Down Expand Up @@ -2468,6 +2478,7 @@ space,noun
spares,noun
sparks,noun
spat,noun
spate,noun
spear,noun
speculation,noun
speed,noun
Expand Down Expand Up @@ -2522,6 +2533,7 @@ survey,noun
suspension,noun
sweepstakes,noun
symphony,noun
syndicate,noun
system,noun
tab,noun
table,noun
Expand Down Expand Up @@ -2552,8 +2564,11 @@ tent,noun
term,noun
test,noun
testing,noun
the blockade,noun
the head,noun
the heads,noun
the opening,noun
the platform,noun
the quake,noun
the view,noun
the views,noun
Expand Down Expand Up @@ -2670,6 +2685,7 @@ wars,noun
washer,noun
watchman,noun
water tank,noun
watermark,noun
wave,noun
waves,noun
wax,noun
Expand Down Expand Up @@ -2819,7 +2835,6 @@ folks,person.label
fool,person.label
foreman,person.label
founder,person.label
the founder,person.label
freak,person.label
friar,person.label
friend,person.label
Expand Down Expand Up @@ -2999,6 +3014,7 @@ students,person.label
sultan,person.label
teen,person.label
tf,person.label
the founder,person.label
tourist,person.label
trader,person.label
traders,person.label
Expand Down Expand Up @@ -3255,6 +3271,7 @@ in an,phrase
int,phrase
je,phrase
jif,phrase
key point,phrase
kinda,phrase
knife point,phrase
kool,phrase
Expand Down Expand Up @@ -3320,6 +3337,7 @@ olive branch,phrase
on an island,phrase
on in,phrase
on point,phrase
on the rocks,phrase
on to,phrase
one man,phrase
one point,phrase
Expand Down Expand Up @@ -3369,6 +3387,7 @@ stopover,phrase
story building,phrase
ta,phrase
task force,phrase
the address,phrase
the adobe,phrase
the best,phrase
the black,phrase
Expand Down Expand Up @@ -3665,6 +3684,7 @@ landmark,place.civil
levy,place.civil
library,place.civil
lighthouse,place.civil
living quarters,place.civil
lodge,place.civil
lookout,place.civil
main entrance,place.civil
Expand Down Expand Up @@ -3774,6 +3794,7 @@ suite,place.civil
summer home,place.civil
supermarket,place.civil
supreme court,place.civil
tavern,place.civil
temple,place.civil
the capital,place.civil
the corner,place.civil
Expand All @@ -3789,13 +3810,15 @@ the village,place.civil
the well,place.civil
theater,place.civil
theme park,place.civil
toilets,place.civil
tower,place.civil
towers,place.civil
town hall,place.civil
townhouse,place.civil
trade school,place.civil
traffic stop,place.civil
trail,place.civil
trails,place.civil
train station,place.civil
trial court,place.civil
tunnel,place.civil
Expand Down Expand Up @@ -4141,6 +4164,7 @@ thus,stopword
bonjour,stopword.fre
tre,stopword.fre
vous,stopword.fre
shi,stopword.ja
het,stopword.nld
tao,stopword.por
#on,stopword.wordnet
Expand Down Expand Up @@ -4528,6 +4552,7 @@ experiment,verb
express,verb
fail,verb
favor,verb
fearing,verb
fed,verb
feed,verb
fell,verb
Expand Down Expand Up @@ -4673,8 +4698,8 @@ oversee,verb
owning,verb
packed,verb
pan,verb
parse,verb
pardon,verb
parse,verb
parting,verb
passing,verb
paste,verb
Expand Down Expand Up @@ -5087,15 +5112,19 @@ shaq,well-known.person
uncle sam,well-known.person
bleach,well-known.science
botany,well-known.science
botany,well-known.science
bromide,well-known.science
edema,well-known.science
latitude,well-known.science
lipid,well-known.science
longitude,well-known.science
magnesium,well-known.science
magnet,well-known.science
magnet,well-known.science
mir,well-known.science
ozone,well-known.science
proton,well-known.science
proton,well-known.science
pyro,well-known.science
quartz,well-known.science
radon,well-known.science
Expand All @@ -5111,18 +5140,3 @@ uranium,well-known.science
vapor,well-known.science
volt,well-known.science
x-ray,well-known.science
botany,well-known.science
proton,well-known.science
magnet,well-known.science
magnesium,well-known.science
spate,noun
honeycomb,food
the opening,noun
the blockade,noun
the platform,noun
on the rocks,phrase
the address,phrase
shi,stopword.ja
fearing,verb
nazionale,adj.it

0 comments on commit 3cbccca

Please sign in to comment.