Semantic-Textual-Similarity code

[issa....@gmail.com]

Dear Colleagues,

Is there any available source code that can be used to re-run STS systems and using the STS datasets.

re-implementing Semantic Text Similarity systems from scratch is a tedious task.

 

Issa

[hen...@mymail.vcu.edu]

Hi Issa,

   The source code is available for our system, Vrep.  It is written in Perl. The system performed just OK, but hopefully are be able to use some of the code. We competed in tasks 1 and 2, see below for links to the code:
http://www.people.vcu.edu/~henryst/#Downloads

And here is a link to our paper describing the system:
http://www.people.vcu.edu/~henryst/vrepAtSemeval2016.pdf

Please let me know if you have questions. Thank you,

   -Sam Henry

[Carmen Banea]

Hi Issa,

You can also try the DKPro Similarity package (https://dkpro.github.io/dkpro-similarity/) that has implementations for various similarity metrics.

Thanks,
Carmen

--
--
Website of task, http://alt.qcri.org/semeval2017/task1/
To post to this group, send email to sts-s...@googlegroups.com
To unsubscribe, send email to sts-semeval...@googlegroups.com
For more options, http://groups.google.com/group/sts-semeval?hl=en?hl=en
---
You received this message because you are subscribed to the Google Groups "STS SemEval" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sts-semeval...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Eneko]

Hi Issa,

you can also check the tensorflow Hub code at https://www.tensorflow.org/hub/overview (under Semantic Similarity).

best

eneko

[josh...@gmail.com]

Hi Issa,

This is a Tensorflow implementation of our paper that used CNN. In spite of the simplicity, it could achieve quite good performance on STSBenchmark dataset.

https://github.com/celarex/STS-CNN-STSbenchmark-Semantic-Similarity-Semantic-Textual-Similarity-CNN-HCTI-Tensorflow-Keras
Sincerely yours,

Yang SHAO