We propose a local coherence model based on a convolutional neural network that operates over the entity grid representation of a text. The model captures long range entity transitions along with entity-specific features without loosing generalization, thanks to the power of distributed representation. We present a pairwise ranking method to train the model in an end-to-end fashion on a task and learn task-specific high level features. Our evaluation on three different coherence assessment tasks demonstrates that our model achieves state of the art results outperforming existing models by a good margin.

@inproceedings{joty-nakov-marquez-jaradat-conll-17, abstract = {We propose a local coherence model based on a convolutional neural network that operates over the entity grid representation of a text. The model captures long range entity transitions along with entity-specific features without loosing generalization, thanks to the power of distributed representation. We present a pairwise ranking method to train the model in an end-to-end fashion on a task and learn task-specific high level features. Our evaluation on three different coherence assessment tasks demonstrates that our model achieves state of the art results outperforming existing models by a good margin.}, address = {Vancouver, Canada}, author = {Shafiq Joty and Preslav Nakov and Lluís Màrquez and Israa Jaradat}, booktitle = {Proceedings of The SIGNLL Conference on Computational Natural Language Learning}, month = {August}, pages = {226--237}, publisher = {Association for Computational Linguistics}, series = {CoNLL'17}, title = {Cross-language Learning with Adversarial Neural Networks: Application to Community Question Answering}, url = {papers/joty-nakov-marquez-jaradat-conll-17.pdf}, year = {2017} }