search

starling-lab / BoostSRL

BoostSRL: "Boosting for Statistical Relational Learning." A gradient-boosting based approach for learning different types of SRL models.
https://starling.utdallas.edu
GNU General Public License v3.0
32 stars 21 forks source link

Null point exception during FACT algorithm (-mln -approxCount) for WebKB data #35

Closed SeongwooLimKR closed 4 years ago

SeongwooLimKR commented 4 years ago

Dear all,

When I run following code,

java -cp boostsrl_v1.1.1.jar edu.wisc.cs.will.Boosting.MLN.RunBoostedMLN -l -train WebKB/train/ -model WebKB/model/ -target faculty -trees 10 -mln -approxCount -aucJarPath .

, following error ocuurs.

image

How can I fix this problem?

I tested on both of BoostSRL source (stable/nightly) and I used v1.1.1 jar of BoostSRL. I downloaded the dataset from https://starling.utdallas.edu/datasets/webkb/ I also used openjdk 11.0.6

Thank you, -Seongwoo

harshakokel commented 4 years ago

Hello Seongwoo,

Thank you for bringing this issue to our notice. I ran the above instruction and found that the approximate counting code is expecting a space after "mode:" in the bk file. We will push a fix soon.

Meanwhile, could you rerun the command with the updated webkb_bk.txt file attached here and let us if that works for you?

Thanks, Harsha Kokel.

webkb_bk.txt

SeongwooLimKR commented 4 years ago

Hello Harsha,

After update the webkb_bk.txt, the command works well!

Thank you! -Seongwoo