This repository contains the code required to reproduce the results reported in
- Sina Zarrieß, David Schlangen. 2017. Obtaining referential word meanings from visual and distributional information: Experiments on object naming. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2017), Vancouver, Canada.*
The work reported here is based on image and referring expression corpora collected by other groups (the images themselves are from the "IAPR TC-12" dataset ( http://www.imageclef.org/photodata), the referring expressions have been collected using the ReferIt game by Tamara Berg and colleagues).
The work is also based on procedures for data preprocessing used for other papers, see our repository on the words-as-classifiers model. For convenience, the various files obtained from preprocessing including pre-trained models can be found here.
The structure of the folder should look like this:
├── evaluation
├── indata
├── linmodels
├── logmodels
├── preproc
└── training
You can do two things now:
-
retrain your own models, see training/
- linmap.py: train transfer models
- linwac.py: train words-as-classifiers with distributional supervision signal
- logwac.py: train standard words-as-classifiers (with a lot of negative samples)
- zero_shot_models.py: train models for zero-shot object naming
-
evaluate the models, see models/
Sina Zarrieß, 2017-10-20