Shi, Huang, and Lee (2017) obtained state-of-the-art results for English and Chinese dependency parsing by combining dynamic-programming implementations of transition-based dependency parsers with a minimal set of bidirectional LSTM features. However, their results were limited to projective parsing. In this paper, we extend their approach to support non-projectivity by providing the first practical implementation of the MH4 algorithm, an O(n^4) mildly nonprojective dynamic-programming parser with very high coverage on non-projective treebanks. To make MH4 compatible with minimal transition-based feature sets, we introduce a transition-based interpretation of it in which parser items are mapped to sequences of transitions. We thus obtain the first implementation of global decoding for non-projective transition-based parsing, and demonstrate empirically that it is more effective than its projective counterpart in parsing a number of highly non-projective languages.
more »
« less
Improving Coverage and Runtime Complexity for Exact Inference in Non-Projective Transition-Based Dependency Parsers
We generalize Cohen, Gómez-Rodríguez, and Satta’s (2011) parser to a family of non-projective transition-based dependency parsers allowing polynomial-time exact inference. This includes novel parsers with better coverage than Cohen et al. (2011), and even a variant that reduces time complexity to O(n^6), improving on prior bounds. We hope that this piece of theoretical work inspires design of novel transition systems with better coverage and better run-time guarantees.
more »
« less
- Award ID(s):
- 1741441
- PAR ID:
- 10075259
- Date Published:
- Journal Name:
- Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
- Volume:
- 2 (short papers)
- Page Range / eLocation ID:
- 420 - 425
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)The student body in university science classrooms is increasingly diverse demographically and this change brings with it an increased chance of mismatch between professor’s expectations and students’ behaviors. Being aware of how cultural expectations influence teaching and learning is the first step in understanding and overcoming these mismatches in order to help all students succeed. This involves making expectations clear, particularly about homework requirements (Ludwig et al., 2011), and defining the line between collaboration and cheating (Craig et al., 2010). When possible, professors should be flexible regarding different cultures’ ideas of time (Hall, 1983), family obligations (Hoover, 2017), and the social power structure (Hofstede, 1986; Yoo, 2014). At the same time, professors should maintain high expectations of all students regardless of ethnic background (Rosenthal & Jacobson, 1968). Drawing from published research as well as interview and survey data, we highlight ways for both professors and students to create an atmosphere of belonging (Walton & Cohen, 2011) and an appreciation of people from all cultures (Museus et al., 2017).more » « less
-
null (Ed.)Relation Extraction (RE) is one of the fundamental tasks in Information Extraction. The goal of this task is to find the semantic relations between entity mentions in text. It has been shown in many previous work that the structure of the sentences (i.e., dependency trees) can provide important information/features for the RE models. However, the common limitation of the previous work on RE is the reliance on some external parsers to obtain the syntactic trees for the sentence structures. On the one hand, it is not guaranteed that the independent external parsers can offer the optimal sentence structures for RE and the customized structures for RE might help to further improve the performance. On the other hand, the quality of the external parsers might suffer when applied to different domains, thus also affecting the performance of the RE models on such domains. In order to overcome this issue, we introduce a novel method for RE that simultaneously induces the structures and predicts the relations for the input sentences, thus avoiding the external parsers and potentially leading to better sentence structures for RE. Our general strategy to learn the RE-specific structures is to apply two different methods to infer the structures for the input sentences (i.e., two views). We then introduce several mechanisms to encourage the structure and semantic consistencies between these two views so the effective structure and semantic representations for RE can emerge. We perform extensive experiments on the ACE 2005 and SemEval 2010 datasets to demonstrate the advantages of the proposed method, leading to the state-of-the-art performance on such datasets.more » « less
-
more » « less
Abstract The 1991–2020 climate normals for sea surface temperature (SST) are computed based on the NOAA Daily Optimum Interpolation SST dataset. This is the first time that high‐resolution SST normals with global coverage can be achieved in the satellite SST era. Normals are one of the fundamental parameters in describing and understanding weather and climate and provide decision‐making information to industry, public, and scientific communities. This product suite includes SST mean, standard deviation, count and extreme parameters at daily, monthly, seasonal and annual time scales on 0.25° spatial grids. The main feature of the SST mean state revealed by the normals is that in the Tropics, the Indo‐Pacific Ocean is dominated by the warm pool (SST ≥ 28°C) while the eastern Pacific is characterized by the cold tongue (SST ≤ 24°C); in the midlatitudes, SSTs are in zonal patterns with high meridional gradients. Daily SST standard deviations are generally small (<1.0°C) except in frontal zones (>1.5°C) mostly associated with ocean currents such as the Gulf Stream, Kuroshio and Equatorial Currents. Compared to the 1982–2011 climatology, the 1991–2020 mean SSTs increased over most global areas but obvious cooling is seen in the Southern Ocean, eastern tropical South Pacific Ocean and North Atlantic warming hole. The Indo‐Pacific warm pool (IPWP) is found to have strengthened in both intensity and coverage since 1982–2011. By a count parameter criterion of ≥300 days annually with SST ≥ 28°C, the IPWP coverage increased 33% from 1982–2011 to 1991–2020. The global mean SST of 1991–2020 is warmer than that of 1982–2011, and the warming rate over 1991–2020 doubles that over 1901–2020.
-
Most molecular diagram parsers recover chemical structure from raster images (e.g., PNGs). However, many PDFs include commands giving explicit locations and shapes for characters, lines, and polygons. We present a new parser that uses these born-digital PDF primitives as input. The parsing model is fast and accurate, and does not require GPUs, Optical Character Recognition (OCR), or vectorization. We use the parser to annotate raster images and then train a new multi-task neural network for recognizing molecules in raster images.We evaluate our parsers using SMILES and standard benchmarks, along with a novel evaluation protocol comparing molecular graphs directly that supports automatic error compilation and reveals errors missed by SMILES-based evaluation. On the synthetic USPTObenchmark, our born-digital parser obtains a recognition rate of 98.4% (1% higher than previous models) and our relatively simple neural parser for raster images obtains a rate of 85% using less training data than existing neural approaches (thousands vs. millions of molecules).more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.18653/v1/N18-2067
