one token aligned multiple-times

[matyaskopp]

Tokens to align:

$ grep ps2013-001-01-002-002.u1.p5.s1.w18 /net/work/people/kopp/ParCzech/audio-alignment/Data/audio-corresp-tsv-in/www.psp.cz/eknih/2013ps/audio/2013/11/25/2013112513581412.tsv 
se	ps2013-001-01-002-002.u1.p5.s1.w18	MiroslavaNemcova.1952

Alignment output (multiple alignments of one token):

$ grep ps2013-001-01-002-002.u1.p5.s1.w18 /net/work/people/kopp/ParCzech/audio-alignment/Data/audio-align-token/2013112513581412.tsv
se	-	False	ps2013-001-01-002-002.u1.p5.s1.w18	False	-	-	-	-	-	-
se	-	False	ps2013-001-01-002-002.u1.p5.s1.w18	False	-	-	-	-	-	-
se	-	False	ps2013-001-01-002-002.u1.p5.s1.w18	False	-	-	-	-	-	-
se	to	False	ps2013-001-01-002-002.u1.p5.s1.w18	True	2	1.000	612760.0	612910.0	150.0	75.000
se	se	False	ps2013-001-01-002-002.u1.p5.s1.w18	True	0	0.000	743260.0	743420.0	160.0	80.000
se	-	False	ps2013-001-01-002-002.u1.p5.s1.w18	False	-	-	-	-	-	-

The first match in audio-align-token/2013112513581412.tsv is aligned

<anchor synch="#ps2013-001-01-002-002.u1.p5.s1.w18.ab"/>
<w xml:id="ps2013-001-01-002-002.u1.p5.s1.w18" lemma="se" pos="PRON" msd="UPosTag=PRON|Case=Acc|PronType=Prs|Reflex=Yes|Variant=Short" ana="pdt:P7-X4----------">se</w>
<anchor synch="#ps2013-001-01-002-002.u1.p5.s1.w18.ae"/>
<!-- .... -->
<when xml:id="ps2013-001-01-002-002.u1.p5.s1.w18.ab" interval="612760.0" since="#ps2013-001-01-002-002.audio1.origin"/>
<when xml:id="ps2013-001-01-002-002.u1.p5.s1.w18.ae" interval="612910.0" since="#ps2013-001-01-002-002.audio1.origin"/>

---

$ grep ps2013-001-01-002-002.u1.p3.s1.w26 audio-corresp-tsv-in/www.psp.cz/eknih/2013ps/audio/2013/11/25/2013112513581412.tsv 
ten	ps2013-001-01-002-002.u1.p3.s1.w26	MiroslavaNemcova.1952

$ grep ps2013-001-01-002-002.u1.p3.s1.w26 /net/work/people/kopp/ParCzech/audio-alignment/Data/audio-align-token/2013112513581412.tsv 
ten	ten	False	ps2013-001-01-002-002.u1.p3.s1.w26	True	0	0.000	516570.0	516820.0	250.0	83.333
ten	-	False	ps2013-001-01-002-002.u1.p3.s1.w26	False	-	-	-	-	-	-
ten	-	False	ps2013-001-01-002-002.u1.p3.s1.w26	False	-	-	-	-	-	-
ten	-	False	ps2013-001-01-002-002.u1.p3.s1.w26	False	-	-	-	-	-	-
ten	ten	False	ps2013-001-01-002-002.u1.p3.s1.w26	True	0	0.000	693800.0	694080.0	280.0	93.333
ten	-	False	ps2013-001-01-002-002.u1.p3.s1.w26	False	-	-	-	-	-	-

[matyaskopp]

ParCzech 4.0:

/net/work/people/kopp/ParCzech/audio-alignment/Data/audio-align-token$ ls | xargs grep '^[^\t]*\t[^-]*\t'|cut -f4|grep -v CONTEXT|sort | uniq -c|sort -n| grep -v '^ *1 ' > ~/double-aligned-tokens.log

Affected sentences:

$ cat ~/double-aligned-tokens.log | sed "s/.* //;s/.w.*//"|sort|uniq|wc -l
32464