Exportar Publicação

Rodrigues, F., Martins, B. & Ribeiro, R. (2018). Neural methods for cross-lingual sentence compression. In van Genabith J.,Agre G.,Declerck T. (Ed.), 18th International Conference on Artificial Intelligence: Methodology, Systems, and Applications, AIMSA 2018. (pp. 104-114). Varna: Springer.

F. Rodrigues et al.,  "Neural methods for cross-lingual sentence compression", in 18th Int. Conf. on Artificial Intelligence: Methodology, Systems, and Applications, AIMSA 2018, van Genabith J.,Agre G.,Declerck T., Ed., Varna, Springer, 2018, vol. 11089, pp. 104-114

@inproceedings{rodrigues2018_1734955851163,
	author = "Rodrigues, F. and Martins, B. and Ribeiro, R.",
	title = "Neural methods for cross-lingual sentence compression",
	booktitle = "18th International Conference on Artificial Intelligence: Methodology, Systems, and Applications, AIMSA 2018",
	year = "2018",
	editor = "van Genabith J.,Agre G.,Declerck T.",
	volume = "11089",
	number = "",
	series = "",
	doi = "10.1007/978-3-319-99344-7_10",
	pages = "104-114",
	publisher = "Springer",
	address = "Varna",
	organization = "",
	url = "https://link.springer.com/chapter/10.1007%2F978-3-319-99344-7_10"
}

TY  - CPAPER
TI  - Neural methods for cross-lingual sentence compression
T2  - 18th International Conference on Artificial Intelligence: Methodology, Systems, and Applications, AIMSA 2018
VL  - 11089
AU  - Rodrigues, F.
AU  - Martins, B.
AU  - Ribeiro, R.
PY  - 2018
SP  - 104-114
SN  - 0302-9743
DO  - 10.1007/978-3-319-99344-7_10
CY  - Varna
UR  - https://link.springer.com/chapter/10.1007%2F978-3-319-99344-7_10
AB  - Sentence compression produces a shorter sentence by removing redundant information, preserving the grammaticality and the important content. We propose an improvement to current neural deletion systems. These systems output a binary sequence of labels for an input sentence: one indicates that the token from the source sentence remains in the compression, whereas zero indicates that the token should be removed. Our main improvement is the use of a Conditional Random Field as final layer, which benefits the decoding of the best global sequence of labels for a given input. In addition, we also evaluate the incorporation of syntactic features, which can improve grammaticality. Finally, this task is extended into a cross-lingual setting where the models are evaluated on English and Portuguese. The proposed architecture achieves better than or equal results to the current state-of-the-art systems, validating that the model benefits from the modification in both languages.
ER  -