Zobrazeno 1 - 10
of 10
pro vyhledávání: '"Benjamin Heinzerling"'
Publikováno v:
Journal of Natural Language Processing. 29:1138-1164
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::c888635ca5fef1f769cf72877af4899f
https://doi.org/10.5715/jnlp.29.1138
https://doi.org/10.5715/jnlp.29.1138
Publikováno v:
Information Processing & Management. 60:103272
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::ae3d39f54beb1ac432681ac3cb0f7da5
https://doi.org/10.1016/j.ipm.2023.103272
https://doi.org/10.1016/j.ipm.2023.103272
Publikováno v:
NAACL-HLT
Improving model generalization on held-out data is one of the core objectives in common- sense reasoning. Recent work has shown that models trained on the dataset with superficial cues tend to perform well on the easy test set with superficial cues b
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::0ae8a3565144e866cfb0b557e14f1caa
http://arxiv.org/abs/2104.11514
http://arxiv.org/abs/2104.11514
Publikováno v:
Proceedings of the First Workshop on Commonsense Inference in Natural Language Processing.
Pretrained language models, such as BERT and RoBERTa, have shown large improvements in the commonsense reasoning benchmark COPA. However, recent work found that many improvements in benchmarks of natural language understanding are not due to models l
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::24b74f9405636475e9f036479868181d
https://doi.org/10.18653/v1/d19-6004
https://doi.org/10.18653/v1/d19-6004
Publikováno v:
Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)
CoNLL
CoNLL
Recent work has validated the importance of subword information for word representation learning. Since subwords increase parameter sharing ability in neural models, their value should be even more pronounced in low-data regimes. In this work, we the
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::9ec1d2d686233ab2c99d60c34fd7c7bb
Publikováno v:
RepL4NLP@ACL
How can we represent hierarchical information present in large type inventories for entity typing? We study the ability of hyperbolic embeddings to capture hierarchical relations between mentions in context and their target types in a shared vector s
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::817bdf1d8ee7984c6275c8305a4d12ae
https://doi.org/10.18653/v1/w19-4319
https://doi.org/10.18653/v1/w19-4319
Autor:
Benjamin Heinzerling, Michael Strube
Publikováno v:
ACL (1)
Pretrained contextual and non-contextual subword embeddings have become available in over 250 languages, allowing massively multilingual NLP. However, while there is no dearth of pretrained embeddings, the distinct lack of systematic evaluations make
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::45d4f2049a24a41d18bffb7acbb49042
Publikováno v:
EMNLP
Selectional preferences have long been claimed to be essential for coreference resolution. However, they are mainly modeled only implicitly by current coreference resolvers. We propose a dependency-based embedding model of selectional preferences whi
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::9d7016da34adb234ede057bdc37bf4ce
http://arxiv.org/abs/1707.06456
http://arxiv.org/abs/1707.06456
Publikováno v:
EACL (1)
We introduce automatic verification as a post-processing step for entity linking (EL). The proposed method \emph{trusts} EL system results collectively, by assuming entity mentions are mostly linked correctly, in order to create a semantic profile of
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::e54c034003ddf54008e2b523d1f63dfe
https://doi.org/10.18653/v1/e17-1078
https://doi.org/10.18653/v1/e17-1078
Autor:
Benjamin Heinzerling, Michael Strube
Publikováno v:
ACL (System Demonstrations)
We present the Visual Entity Explorer (VEX), an interactive tool for visually exploring and analyzing the output of entity linking systems. VEX is designed to aid developers in improving their systems by visualizing system results, gold annotations,
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::d8313c841c689ea0473cf4ac40bf066a
https://doi.org/10.3115/v1/p15-4007
https://doi.org/10.3115/v1/p15-4007