semantic-compressor 2.1__py3-none-any.whl → 2.3__py3-none-any.whl

compressor/resources/nltk_data/tokenizers/punkt_tab/dutch/sent_starters.txt

@@ -0,0 +1,54 @@
+het
+daardoor
+de
+er
+hoewel
+wat
+urlings
+na
+ze
+alleen
+dat
+ik
+pijls
+wie
+daarna
+foto
+als
+boer
+hammes
+verder
+ook
+evers
+vandaar
+toen
+we
+langenberg
+naast
+want
+in
+wij
+zo
+hendrikx
+daar
+crouzen
+dit
+daarnaast
+anp
+zij
+behalve
+waarom
+daarom
+bovendien
+hij
+daarbij
+nee
+volgens
+daarmee
+bukkems
+dvnl
+eén
+pas
+tijdens
+vooral
+maar

compressor/resources/nltk_data/tokenizers/punkt_tab/english/abbrev_types.txt

@@ -0,0 +1,156 @@
+ct
+m.j
+t
+a.c
+n.h
+ms
+p.a.m
+dr
+pa
+p.m
+u.k
+st
+dec
+u.s.a
+lt
+g.k
+adm
+p
+h.m
+ga
+tenn
+yr
+sen
+n.c
+j.j
+d.h
+s.g
+inc
+vs
+s.p.a
+a.t
+n
+feb
+sr
+jan
+s.a.y
+n.y
+col
+g.f
+c.o.m.b
+d
+ft
+va
+r.k
+e.f
+chg
+r.i
+a.g
+minn
+a.h
+k
+n.j
+m
+l.f
+f.j
+gen
+i.m.s
+s.a
+aug
+j.p
+okla
+m.d.c
+ltd
+oct
+s
+vt
+r.a
+j.c
+ariz
+w.w
+b.v
+ore
+h
+w.r
+e.h
+mrs
+cie
+corp
+w
+n.v
+a.d
+r.j
+ok
+. . 
+e.m
+w.c
+ill
+nov
+u.s
+prof
+conn
+u.s.s.r
+mg
+f.g
+ph.d
+g
+calif
+messrs
+h.f
+wash
+tues
+sw
+bros
+u.n
+l
+wis
+mr
+sep
+d.c
+ave
+e.l
+co
+s.s
+reps
+c
+r.t
+h.c
+r
+wed
+a.s
+v
+fla
+jr
+r.h
+c.v
+m.b.a
+rep
+a.a
+e
+c.i.t
+l.a
+b.f
+j.b
+d.w
+j.k
+ala
+f
+w.va
+sept
+mich
+n.m
+j.r
+l.p
+s.c
+colo
+fri
+a.m
+g.d
+kan
+maj
+ky
+a.m.e
+n.d
+t.j
+cos
+nev

compressor/resources/nltk_data/tokenizers/punkt_tab/english/collocations.tab

@@ -0,0 +1,37 @@
+##number##	international
+##number##	rj
+##number##	commodities
+##number##	cooper
+b	stewart
+##number##	genentech
+##number##	wedgestone
+i	toussie
+##number##	pepper
+j	fialka
+o	ludcke
+##number##	insider
+##number##	aes
+i	magnin
+##number##	credit
+##number##	corrections
+##number##	financing
+##number##	henley
+##number##	business
+##number##	pay-fone
+b	wigton
+b	edelman
+b	levine
+##number##	leisure
+b	smith
+j	walter
+##number##	pegasus
+##number##	dividend
+j	aron
+##number##	review
+##number##	abreast
+##number##	who
+##number##	letters
+##number##	colgate
+##number##	cbot
+##number##	notable
+##number##	zimmer