Segmentation Fault of nersuite tag

[tholor]

Hey, I am new to nersuite. I am following the steps from the "basic usage" doc, but I am getting a "Segementation fault". The first two steps of the pipeline seem to work:

nersuite_tokenizer < test.txt > test.tok
nersuite_gtagger -d ~/nersuite/models/gtagger < test.tok > test.gtag 
Loading morphdic...done.
Loading pos_models................done.
Loading chunk_models....done.

The created file(s) look fine to me:

cat test.gtag
0   3   ZAP ZAP NN  B-NP
3   4   -   -   HYPH    I-NP
4   6   70  70  CD  I-NP
7   15  tyrosine    tyrosine    NN  I-NP
16  22  kinase  kinase  NN  I-NP
22  23  ,   ,   ,   O
24  28  CD45    CD45    NN  B-NP
28  29  ,   ,   ,   O
30  33  and and CC  O
34  35  T   T   NN  B-NP
36  40  cell    cell    NN  I-NP
41  49  receptor    receptor    NN  I-NP
50  61  involvement involvement NN  I-NP
62  64  in  in  IN  B-PP
65  67  UV  UV  NN  B-NP
67  68  -   -   HYPH    B-NP
69  72  and and CC  I-NP
73  77  H2O2    H2O2    NN  I-NP
77  78  -   -   HYPH    O
78  85  induced induce  VBN B-NP
86  87  T   T   NN  I-NP
88  92  cell    cell    NN  I-NP
93  99  signal  signal  NN  I-NP
100 112 transduction    transduction    NN  I-NP
112 113 .   .   .   O

However the last step using nersuite tag fails

nersuite tag -m ~/nersuite/models/bc2gm/bc2gm.iob2.no_dic.m < test.gtag > test.ner

Segmentation fault (core dumped)

Any idea whats going wrong here?

[priancho]

Hi tholor,

Sorry for my late reply. I recently moved from Japan to South Korea and couldn't make time for debugging this problem.

The segfault error is from priancho/crfsuite. When I updated crfsuite by syncing it with its original repo (chokkan/crfsuite), some codes related to label bias option were removed accidentally. I made a bugfix and applied to priancho/crfsuite repo. Please try nersuite after re-installing nersuite with the new priancho/crfsuite.

Best wishes, Hancheol

[tholor]

Hi Hancheol,

thanks for the fix. Works like a charm now :+1:

Best, Malte