ImanAndrea/citation_annotations · Datasets at Hugging Face

file stringclasses 58 values	start int64 14 8.27k	end int64 49 8.32k	label stringclasses 4 values	user stringclasses 4 values	text stringlengths 2 4.16k
paper_11.txt	78	562	Coherence	Ed	Cross-lingual information retrieval (CLIR) (Braschler et al., 1999;Shakery and Zhai, 2013;Jiang et al., 2020;Asai et al., 2021a), for example, can find relevant text in a high-resource language such as English even when the query is posed in a different, possibly low-resource, language. In this work, we develop useful CLIR models for this constrained, yet important, setting where a retrieval corpus is available only in a single high-resource language (English in our experiments).
paper_11.txt	797	960	Unsupported claim	Ed	alternative end-to-end approach that can tackle the problem purely cross-lingually, i.e., without involving MT, would clearly be more efficient and cost-effective
paper_11.txt	2,235	2,456	Lacks synthesis	Ed	KD (Hinton et al., 2014) is a powerful supervision technique typically used to distill the knowledge of a large teacher model about some task into a smaller student model (Mukherjee and Awadallah, 2020;Turc et al., 2020)
paper_13.txt	14	444	Lacks synthesis	Ed	Few-shot learning is the problem of learning classifiers with only a few training examples. Zero-shot learning (Larochelle et al., 2008), also known as dataless classification (Chang et al., 2008), is the extreme case, in which no labeled data is used. For text data, this is usually accomplished by representing the labels of the task in a textual form, which can either be the name of the label or a concise textual description.
paper_13.txt	1,321	2,033	Lacks synthesis	Ed	These models embed both input and label texts into a common vector space. The similarity of the two items can then be computed using a similarity function such as the dot product. The advantage is that input and label text are encoded independently, which means that the label embeddings can be pre-computed. Therefore, at inference time, only a single call to the model per input is needed. In contrast, the models typically applied in the entailment approach are Cross Attention (CA) models which need to be executed for every combination of text and label. On the other hand, they allow for interaction between the tokens of label and input, so that in theory they should be superior in classification accurac
paper_13.txt	1,713	1,880	Unsupported claim	Ed	In contrast, the models typically applied in the entailment approach are Cross Attention (CA) models which need to be executed for every combination of text and label.
paper_13.txt	3,365	3,395	Unsupported claim	Ed	In contrast to most prior work
paper_14.txt	182	310	Unsupported claim	Ed	Unfortunately, for many languages, and especially low-resource languages, such taskspecific labelled data is often not available
paper_14.txt	2,549	2,644	Unsupported claim	Ed	as this is the only task for which high-quality data is available in a large number of language
paper_14.txt	2,833	2,980	Unsupported claim	Ed	a base understanding of syntactic structure in both the source and target language is necessary for any meaningful natural language processing task
paper_15.txt	897	902	Format	Ed	2021)
paper_16.txt	14	414	Coherence	Ed	To facilitate the study of text summarization, earlier datasets are mostly in the news domain with relatively short input passages, such as NYT (Sandhaus, 2008), Gigaword (Napoles et al., 2012), CNN/Daily Mail (Hermann et al., 2015), NEWSROOM (Grusky et al., 2018) and XSUM (Narayan et al., 2018). Datasets for long docu-ments include Sharma et al. (2019), Cohan et al. (2018), andFisas et al. (2016)
paper_16.txt	14	414	Lacks synthesis	Ed	To facilitate the study of text summarization, earlier datasets are mostly in the news domain with relatively short input passages, such as NYT (Sandhaus, 2008), Gigaword (Napoles et al., 2012), CNN/Daily Mail (Hermann et al., 2015), NEWSROOM (Grusky et al., 2018) and XSUM (Narayan et al., 2018). Datasets for long docu-ments include Sharma et al. (2019), Cohan et al. (2018), andFisas et al. (2016)
paper_16.txt	713	1,264	Lacks synthesis	Ed	Researchers recently explore the peer review domain data for a few tasks, such as PeerRead (Kang et al., 2018) for paper decision predictions, AM-PERE for proposition classification in reviews, and RR (Cheng et al., 2020) for paired-argument extraction from review-rebuttal pairs. Additionally, a meta-review dataset is introduced by Bhatia et al. (2020) without any annotation. There are also some explorations on research articles (Teufel et al., 1999;Liakata et al., 2010;Lauscher et al., 2018), which differ in nature from the peer review domain.
paper_17.txt	14	1,286	Lacks synthesis	Ed	Fully supervised event extraction. Event extraction has been studied for over a decade (Ahn, 2006;Ji and Grishman, 2008) and most traditional event extraction works follow the fully supervised setting (Nguyen et al., 2016;Sha et al., 2018;Nguyen and Nguyen, 2019;Yang et al., 2019;Lin et al., 2020;Li et al., 2020). Many of them use classification-based models and use pipeline-style frameworks to extract events (Nguyen et al., 2016;Yang et al., 2019;Wadden et al., 2019). To better leverage shared knowledge in event triggers and arguments, some works propose to incorporate global features to jointly decide triggers and arguments (Lin et al., 2020;Li et al., 2013;Yang and Mitchell, 2016). Recently, few generation-based event extraction models have been proposed. TANL (Paolini et al., 2021) treats event extraction as translation tasks between augmented natural languages. Their predicted targetaugmented language embed labels into the input passage via using brackets and vertical bar symbols, hindering the model from fully leveraging label semantics. BART-Gen is also a generation-based model focusing on documentlevel event argument extraction. Yet, similar to TANL, they solve event extraction with a pipeline, which prevents knowledge sharing across subtasks.
paper_17.txt	1,726	2,044	Lacks synthesis	Ed	Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence
paper_17.txt	1,726	2,044	Coherence	Ed	Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence
paper_18.txt	952	970	Format	Ed	(Li et al., 2020a;
paper_18.txt	3,495	3,574	Unsupported claim	Ed	three large-scale benchmark datasets (OntoNotes V4.0, OntoNotes V5.0, and MSRA)
paper_19.txt	1,566	1,654	Unsupported claim	Ed	shown promising for AL in NLP due to its good qualitative and computational performance
paper_19.txt	1,801	1,824	Format	Ed	Shelmanov et al. (2021
paper_20.txt	252	631	Coherence	Ed	Following Chen et al. (2020c), other works adopt PLMs for few-shot D2T generation (Chang et al., 2021b;Su et al., 2021a). Kale and Rastogi (2020b) and Ribeiro et al. (2020) showed that PLMs using linearized representations of data can outperform graph neural networks on graph-to-text datasets, recently surpassed again by graph-based models (Ke et al., 2021;Chen et al., 2020a)
paper_20.txt	1,781	1,797	Format	Ed	Recently, have
paper_20.txt	3,514	3,533	Format	Ed	Jiang et al., 2020)
paper_37.txt	1,213	1,280	Format	Ed	Radford et al., 2021;Schick and Schütze, 2020a,b;Brown et al., 2020
paper_37.txt	1,544	1,572	Format	Ed	Schick and Schütze, 2020a,b)
paper_37.txt	992	1,071	Unsupported claim	Ed	they are impractical to use in real-world applications due to their model sizes
paper_37.txt	1,100	1,692	Lacks synthesis	Ed	Providing prompts or task descriptions play an vital role in improving pre-trained language models in many tasks Radford et al., 2021;Schick and Schütze, 2020a,b;Brown et al., 2020). Among them, GPT models (Radford et al., 2019;Brown et al., 2020) achieved great success in prompting or task demonstrations in NLP tasks. In light of this direction, prompt-based approaches improve small pre-trained models in few-shot text classification tasks Schick and Schütze, 2020a,b). CLIP (Radford et al., 2021) also explores prompt templates for image classification which affect zero-shot performance
paper_37.txt	49	932	Lacks synthesis	Ed	Recently, several few-shot learners on vision-language tasks were proposed including GPT (Radford et al., 2019;Brown et al., 2020), Frozen (Tsimpoukelli et al., 2021), PICa , and SimVLM . Frozen (Tsimpoukelli et al., 2021) is a large language model based on GPT-2 (Radford et al., 2019), and is transformed into a multimodal few-shot learner by extending the soft prompting to incorporate a set of images and text. Their approach shows the fewshot capability on visual question answering and image classification tasks. Similarly, PICa uses GPT-3 (Brown et al., 2020) to solve VQA tasks in a few-shot manner by providing a few in-context VQA examples. It converts images into textual descriptions so that GPT-3 can understand the images. SimVLM is trained with prefix language modeling on weakly-supervised datasets. It demonstrates its effectiveness on a zero-shot captioning task
paper_38.txt	24	794	Lacks synthesis	Ed	pre-training a transformer model on a large corpus with language modeling tasks and finetuning it on different downstream tasks has become the main transfer learning paradigm in natural language processing (Devlin et al., 2019). Notably, this paradigm requires updating and storing all the model parameters for every downstream task. As the model size proliferates (e.g., 330M parameters for BERT (Devlin et al., 2019) and 175B for GPT-3 (Brown et al., 2020)), it becomes computationally expensive and challenging to fine-tune the entire pre-trained language model (LM). Thus, it is natural to ask the question of whether we can transfer the knowledge of a pre-trained LM into downstream tasks by tuning only a small portion of its parameters with most of them freezing.
paper_38.txt	872	1,390	Lacks synthesis	Ed	One line of research (Li and Liang, 2021) suggests to augment the model with a few small trainable mod-ules and freeze the original transformer weight. Take Adapter (Houlsby et al., 2019;Pfeiffer et al., 2020a,b) and Compacter (Mahabadi et al., 2021) for example, both of them insert a small set of additional modules between each transformer layer. During fine-tuning, only these additional and taskspecific modules are trained, reducing the trainable parameters to ∼ 1-3% of the original transformer model per task.
paper_38.txt	1,434	1,921	Lacks synthesis	Ed	The GPT-3 models (Brown et al., 2020;Schick and Schütze, 2020) find that with proper manual prompts, a pre-trained LM can successfully match the fine-tuning performance of BERT models. LM-BFF (Gao et al., 2020), EFL (Wang et al., 2021), and AutoPrompt (Shin et al., 2020) further this direction by insert prompts in the input embedding layer. However, these methods rely on grid-search for a natural language-based prompt from a large search space, resulting in difficulties to optimize.
paper_38.txt	2,461	2,602	Unsupported claim	Ed	all existing prompt-tuning methods have thus far focused on task-specific prompts, making them incompatible with the traditional LM objective
paper_38.txt	2,617	2,711	Unsupported claim	Ed	it is unlikely to see many different sentences with the same prefix in the pre-training corpus
paper_39.txt	129	213	Format	Ed	(Eric et al., 2017;Wu et al., 2019; and collections of largescale annotation corpora
paper_39.txt	355	377	Format	Ed	(El Asri et al., 2017
paper_39.txt	530	533	Unsupported claim	Ed	SGD
paper_39.txt	874	909	Format	Ed	Quan et al., 2020;Lin et al., 2021)
paper_39.txt	998	1,151	Unsupported claim	Ed	vast majority of existing multilingual ToD datasets do not consider the real use cases when using a ToD system to search for local entities in a country.
paper_40.txt	396	437	Format	Ed	[Levy et al., 2017, Elsahar et al., 2018
paper_41.txt	1,890	2,370	Lacks synthesis	Ed	Previous work has shown that SNLI (Bowman et al., 2015) and MNLI (Williams et al., 2018) have annotation artifacts (e.g., negation is a strong indicator of contradictions) (Gururangan et al., 2018). The literature has also shown that simple adversarial attacks including negation cues are very effective (Naik et al., 2018;Wallace et al., 2019). Kovatchev et al. (2019) analyze 11 paraphrasing systems and show that they obtain substantially worse results when negation is present
paper_41.txt	3,201	3,272	Format	Ed	Bar-Haim et al., 2006;Giampiccolo et al., 2007;Bentivogli et al., 2009)
paper_43.txt	465	524	Unsupported claim	Ed	Machine Translation (MT) is the mainstream approach for GEC
paper_43.txt	918	966	Unsupported claim	Ed	recent powerful Transformer-based Seq2Seq model
paper_100.txt	594	825	Coherence	Ekaterina	Zhang et al. (2019) improves an LSTM- based encoder-decoder model with online vocabulary adaptation. For abbreviated pinyin, CoCAT (Huang et al., 2015) uses machine translation technology to reduce the number of the typing letters.
paper_100.txt	968	1,179	Coherence	Ekaterina	Our work differs from existing works in that we are the first one to exploit GPT and verify the pros and cons of GPT in different situations. In addition, there are some works handling pinyin with typing errors.
paper_100.txt	1,763	2,087	Coherence	Ekaterina	Sun et al. (2021) propose a general-purpose Chinese BERT with new embedding layers to inject pinyin and glyph information of characters. There are also task-specific BERT models, especially for the task of grammatical error correction since an important type of error is caused by characters pronounced with the same pinyin.
paper_11.txt	1,025	1,042	Unsupported claim	Ekaterina	multilingual BERT
paper_11.txt	1,485	1,487	Format	Ekaterina	1
paper_13.txt	14	105	Unsupported claim	Ekaterina	Few-shot learning is the problem of learning classifiers with only a few training examples.
paper_13.txt	267	444	Unsupported claim	Ekaterina	For text data, this is usually accomplished by representing the labels of the task in a textual form, which can either be the name of the label or a concise textual description.
paper_13.txt	1,713	1,880	Unsupported claim	Ekaterina	In contrast, the models typically applied in the entailment approach are Cross Attention (CA) models which need to be executed for every combination of text and label.
paper_13.txt	1,881	2,119	Unsupported claim	Ekaterina	On the other hand, they allow for interaction between the tokens of label and input, so that in theory they should be superior in classification accuracy. However, in this work we show that in practice, the difference in quality is small.
paper_13.txt	2,121	2,383	Unsupported claim	Ekaterina	Both CA and SNs also support the few-shot learning setup by fine-tuning the models on a small number of labeled examples. This is usually done by updating all parameters of the model, which in turn makes it impossible to share the models between different tasks.
paper_13.txt	3,365	3,483	Unsupported claim	Ekaterina	In contrast to most prior work, we also show that these results can also be achieved for languages other than English.
paper_14.txt	14	181	Unsupported claim	Ekaterina	At present, for a large majority of natural language processing tasks, the most successful approach is fine-tuning pre-trained models with task-specific labelled data.
paper_15.txt	649	715	Unsupported claim	Ekaterina	Researchers also realize that the vision modality maybe redundant.
paper_15.txt	897	902	Format	Ekaterina	2021)
paper_15.txt	864	956	Coherence	Ekaterina	Encouraging results appeared in 2021) proposed a cross-lingual visual pretraining approach.
paper_15.txt	14	1,184	Lacks synthesis	Ekaterina	Multimodal machine translation is a cross-domain task in the filed of machine translation. Early attempts mainly focused on enhancing the MMT model by better incorporation of the vision features (Calixto and Liu, 2017;Elliott and Kádár, 2017;Delbrouck and Dupont, 2017). However, directly encoding the whole image feature brings additional noise to the text (Yao and Wan, 2020;Liu et al., 2021a). To address the above issue, Yao and Wan (2020) proposed a multimodal self-attention to consider the relative difference of information between two modalities. Similarly, Liu et al. (2021a) used a Gumbel Softmax to achieve the same goal. Researchers also realize that the vision modality maybe redundant. Irrelevant images have little impact on the translation quality, and no significant BLEU drop is observed even the image is absent (Elliott, 2018). Encouraging results appeared in 2021) proposed a cross-lingual visual pretraining approach. In this work, we make a systematic study on whether stronger vision features are helpful. We also extend the research to enhanced features, such as object-detection and image captioning, which is complementary to previous work.
paper_16.txt	856	906	Unsupported claim	Ekaterina	AM-PERE for proposition classification in reviews
paper_16.txt	1,093	1,621	Coherence	Ekaterina	There are also some explorations on research articles (Teufel et al., 1999;Liakata et al., 2010;Lauscher et al., 2018), which differ in nature from the peer review domain. A wide range of control perspectives has been explored in controllable generation, including style control (e.g., sentiments (Duan et al., 2020), politeness (Madaan et al., 2020), formality , domains (Takeno et al., 2017) and persona ) and content control (e.g., length (Duan et al., 2020), entities (Fan et al., 2018a), and keywords (Tang et al., 2019)).
paper_17.txt	1,074	1,169	Unsupported claim	Ekaterina	BART-Gen is also a generation-based model focusing on documentlevel event argument extraction.
paper_17.txt	1,397	1,586	Unsupported claim	Ekaterina	However, their designs are not specific for low-resource scenarios, hence, these models can not enjoy all the benefits that DEGREE obtains for low-resource event extraction at the same time
paper_17.txt	1,726	2,044	Coherence	Ekaterina	Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence
paper_17.txt	2,261	2,403	Coherence	Ekaterina	Another thread of works are using meta-learning to deal with the less label challenge (Deng et al., 2020;Shen et al., 2021;Cong et al., 2021).
paper_17.txt	1,619	2,259	Lacks synthesis	Ekaterina	Low-resource event extraction. It has been a rising interest in event extraction under less data scenario. Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence. Text2Event's unnatural output format hinders the model from fully leveraging pre-trained knowledge. Hence, their model falls short on the cases with only extremely low data being available (as shown in Section 3).
paper_18.txt	952	970	Format	Ekaterina	(Li et al., 2020a;
paper_18.txt	1,391	1,496	Unsupported claim	Ekaterina	Nevertheless, building the lexicon is time-consuming and the quality of the lexicon may not be satisfied.
paper_18.txt	2,131	2,253	Unsupported claim	Ekaterina	However, too immersed regularity leads to unfavorable boundary detection of entities and disturbing character composition.
paper_18.txt	3,533	3,547	Unsupported claim	Ekaterina	OntoNotes V4.0
paper_18.txt	3,549	3,563	Unsupported claim	Ekaterina	OntoNotes V5.0
paper_18.txt	3,569	3,573	Unsupported claim	Ekaterina	MSRA
paper_18.txt	3,576	3,729	Unsupported claim	Ekaterina	The results show that RICON achieves considerable improvements compared to the state-of-the-art models, even outperforming existing lexicon-based models.
paper_18.txt	3,757	3,792	Unsupported claim	Ekaterina	a practical medical dataset (CBLUE)
paper_19.txt	14	174	Unsupported claim	Ekaterina	Deep learning, to a large extent, has freed data scientists from doing feature engineering, which has been one of the essential obstacles to annotation with AL.
paper_19.txt	723	900	Unsupported claim	Ekaterina	In our work, we take MNLP as a query strategy for experiments on sequence tagging tasks since it has demonstrated a good trade-off between quality and computational performance.
paper_19.txt	1,464	1,655	Unsupported claim	Ekaterina	We continue this line of works by relying on pre-trained Transformers since this architecture has been shown promising for AL in NLP due to its good qualitative and computational performance.
paper_19.txt	1,802	1,824	Format	Ekaterina	Shelmanov et al. (2021
paper_19.txt	2,873	3,303	Lacks synthesis	Ekaterina	Recently proposed alternatives to uncertaintybased query strategies leverage reinforcement learning and imitation learning (Fang et al., 2017;Liu et al., 2018;Vu et al., 2019;Brantley et al., 2020). This series of works aims at constructing trainable policy-based query strategies. However, this requires an excessive amount of computation while the transferability of learned policies across domains and tasks is underresearched.
paper_19.txt	3,692	3,717	Format	Ekaterina	(Shelmanov et al., 2021).
paper_19.txt	3,422	3,433	Unsupported claim	Ekaterina	ASM problem
paper_20.txt	1,003	1,049	Format	Ekaterina	(Heidari et al., 2021;Kale and Rastogi, 2020a;
paper_20.txt	791	1,331	Coherence	Ekaterina	Generation Using simple handcrafted templates for individual keys or predicates is an efficient way of introducing domain knowledge while preventing text-to-text models from overfitting to a specific data format (Heidari et al., 2021;Kale and Rastogi, 2020a;. Transforming individual triples to text is also used in Laha et al. (2020) whose work is the most similar to ours. They also build a three-step pipeline for zero-shot D2T generation, but they use handcrafted rules for producing the output text and do not address content planning.
paper_20.txt	14	760	Lacks synthesis	Ekaterina	D2T Generation with PLMs Large neural language models pretrained on self-supervised tasks (Lewis et al., 2020;Liu et al., 2019;Devlin et al., 2019) have recently gained a lot of traction in D2T generation research (Ferreira et al., 2020). Following Chen et al. (2020c), other works adopt PLMs for few-shot D2T generation (Chang et al., 2021b;Su et al., 2021a). Kale and Rastogi (2020b) and Ribeiro et al. (2020) showed that PLMs using linearized representations of data can outperform graph neural networks on graph-to-text datasets, recently surpassed again by graph-based models (Ke et al., 2021;Chen et al., 2020a). Although the models make use of general-domain pretraining tasks, all of them are eventually finetuned on domain-specific data.
paper_20.txt	1,583	1,870	Coherence	Ekaterina	As previously demonstrated, using a content plan in neural D2T generation has important impact on the overall text quality (Moryossef et al., 2019a,b;Puduppully et al., 2019;Trisedya et al., 2020). Recently, have shown that using a content plan leads to improved quality of PLM outputs.
paper_20.txt	2,366	2,388	Format	Ekaterina	Li and Jurafsky, 2017)
paper_20.txt	3,514	3,533	Format	Ekaterina	Jiang et al., 2020)
paper_20.txt	4,026	4,046	Format	Ekaterina	(Botha et al., 2018;
paper_20.txt	3,615	4,047	Coherence	Ekaterina	In contrast to sentence fusion (Geva et al., 2019;Barzilay and McKeown, 2005) or sentence compression (Filippova and Altun, 2013), we operate in the context of multiple sentences in a paragraph. The task is the central focus of our WIKIFLUENT corpus ( §4), which we synthesize using a model for the reverse task, split-andrephrase, i.e. splitting a complex sentence into simpler ones while preserving semantics (Botha et al., 2018;.
paper_53.txt	225	268	Format	Ekaterina	(Davison et al., 2019;Petroni et al., 2019;
paper_53.txt	364	385	Format	Ekaterina	(Trinh and Le, 2018;)
paper_53.txt	3,472	3,609	Unsupported claim	Ekaterina	DynaGen uses pretrained commonsense models to generate implications of a question and expands the inference input with these generations
paper_54.txt	441	509	Format	Ekaterina	(Zhang et al., 2020;Aghajanyan et al., 2020Aghajanyan et al., , 2021
paper_54.txt	1,943	2,069	Coherence	Ekaterina	inspired by Li et al. (2020), to train a machine learning model to smooth out such noise and to speed up the inference process
paper_54.txt	2,071	2,213	Lacks synthesis	Ekaterina	Different from Li et al. (2020), we propose to utilize non-autoregressive text generators, which generate all tokens in the output in parallel
paper_54.txt	3,248	3,521	Unsupported claim	Ekaterina	Regarding inference efficiency, our NAUS with truncating is 1000 times more efficient than the search approach; even with dynamic programming for length control, NAUS is still 100 times more efficient than search and several times more efficient than autoregressive models.
paper_55.txt	2,878	3,084	Unsupported claim	Ekaterina	Lastly, our dataset yields a novel extractive summarization dataset, providing a benchmark for studying domain transfer in summarization and enabling QA models to provide concise answers to complex queries.
paper_56.txt	1,389	1,551	Unsupported claim	Ekaterina	On the other hand, neural conversational models are not necessarily designed to generate faithful outputs, but to mimic the distributional properties of the data.
paper_56.txt	2,734	2,863	Unsupported claim	Ekaterina	we annotate responses generated by several state-of-the-art models, including ones that are designed to alleviate hallucinations.

paper_11.txt

78

562

Coherence

Ed

Cross-lingual information retrieval (CLIR) (Braschler et al., 1999;Shakery and Zhai, 2013;Jiang et al., 2020;Asai et al., 2021a), for example, can find relevant text in a high-resource language such as English even when the query is posed in a different, possibly low-resource, language. In this work, we develop useful CLIR models for this constrained, yet important, setting where a retrieval corpus is available only in a single high-resource language (English in our experiments).

paper_11.txt

797

960

Unsupported claim

Ed

alternative end-to-end approach that can tackle the problem purely cross-lingually, i.e., without involving MT, would clearly be more efficient and cost-effective

paper_11.txt

2,235

2,456

Lacks synthesis

Ed

KD (Hinton et al., 2014) is a powerful supervision technique typically used to distill the knowledge of a large teacher model about some task into a smaller student model (Mukherjee and Awadallah, 2020;Turc et al., 2020)

paper_13.txt

14

444

Lacks synthesis

Ed

Few-shot learning is the problem of learning classifiers with only a few training examples. Zero-shot learning (Larochelle et al., 2008), also known as dataless classification (Chang et al., 2008), is the extreme case, in which no labeled data is used. For text data, this is usually accomplished by representing the labels of the task in a textual form, which can either be the name of the label or a concise textual description.

paper_13.txt

1,321

2,033

Lacks synthesis

Ed

These models embed both input and label texts into a common vector space. The similarity of the two items can then be computed using a similarity function such as the dot product. The advantage is that input and label text are encoded independently, which means that the label embeddings can be pre-computed. Therefore, at inference time, only a single call to the model per input is needed. In contrast, the models typically applied in the entailment approach are Cross Attention (CA) models which need to be executed for every combination of text and label. On the other hand, they allow for interaction between the tokens of label and input, so that in theory they should be superior in classification accurac

paper_13.txt

1,713

1,880

Unsupported claim

Ed

In contrast, the models typically applied in the entailment approach are Cross Attention (CA) models which need to be executed for every combination of text and label.

paper_13.txt

3,365

3,395

Unsupported claim

Ed

In contrast to most prior work

paper_14.txt

182

310

Unsupported claim

Ed

Unfortunately, for many languages, and especially low-resource languages, such taskspecific labelled data is often not available

paper_14.txt

2,549

2,644

Unsupported claim

Ed

as this is the only task for which high-quality data is available in a large number of language

paper_14.txt

2,833

2,980

Unsupported claim

Ed

a base understanding of syntactic structure in both the source and target language is necessary for any meaningful natural language processing task

paper_15.txt

897

902

Format

Ed

2021)

paper_16.txt

14

414

Coherence

Ed

To facilitate the study of text summarization, earlier datasets are mostly in the news domain with relatively short input passages, such as NYT (Sandhaus, 2008), Gigaword (Napoles et al., 2012), CNN/Daily Mail (Hermann et al., 2015), NEWSROOM (Grusky et al., 2018) and XSUM (Narayan et al., 2018). Datasets for long docu-ments include Sharma et al. (2019), Cohan et al. (2018), andFisas et al. (2016)

paper_16.txt

14

414

Lacks synthesis

Ed

To facilitate the study of text summarization, earlier datasets are mostly in the news domain with relatively short input passages, such as NYT (Sandhaus, 2008), Gigaword (Napoles et al., 2012), CNN/Daily Mail (Hermann et al., 2015), NEWSROOM (Grusky et al., 2018) and XSUM (Narayan et al., 2018). Datasets for long docu-ments include Sharma et al. (2019), Cohan et al. (2018), andFisas et al. (2016)

paper_16.txt

713

1,264

Lacks synthesis

Ed

Researchers recently explore the peer review domain data for a few tasks, such as PeerRead (Kang et al., 2018) for paper decision predictions, AM-PERE for proposition classification in reviews, and RR (Cheng et al., 2020) for paired-argument extraction from review-rebuttal pairs. Additionally, a meta-review dataset is introduced by Bhatia et al. (2020) without any annotation. There are also some explorations on research articles (Teufel et al., 1999;Liakata et al., 2010;Lauscher et al., 2018), which differ in nature from the peer review domain.

paper_17.txt

14

1,286

Lacks synthesis

Ed

Fully supervised event extraction. Event extraction has been studied for over a decade (Ahn, 2006;Ji and Grishman, 2008) and most traditional event extraction works follow the fully supervised setting (Nguyen et al., 2016;Sha et al., 2018;Nguyen and Nguyen, 2019;Yang et al., 2019;Lin et al., 2020;Li et al., 2020). Many of them use classification-based models and use pipeline-style frameworks to extract events (Nguyen et al., 2016;Yang et al., 2019;Wadden et al., 2019). To better leverage shared knowledge in event triggers and arguments, some works propose to incorporate global features to jointly decide triggers and arguments (Lin et al., 2020;Li et al., 2013;Yang and Mitchell, 2016). Recently, few generation-based event extraction models have been proposed. TANL (Paolini et al., 2021) treats event extraction as translation tasks between augmented natural languages. Their predicted targetaugmented language embed labels into the input passage via using brackets and vertical bar symbols, hindering the model from fully leveraging label semantics. BART-Gen is also a generation-based model focusing on documentlevel event argument extraction. Yet, similar to TANL, they solve event extraction with a pipeline, which prevents knowledge sharing across subtasks.

paper_17.txt

1,726

2,044

Lacks synthesis

Ed

Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence

paper_17.txt

1,726

2,044

Coherence

Ed

Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence

paper_18.txt

952

970

Format

Ed

(Li et al., 2020a;

paper_18.txt

3,495

3,574

Unsupported claim

Ed

three large-scale benchmark datasets (OntoNotes V4.0, OntoNotes V5.0, and MSRA)

paper_19.txt

1,566

1,654

Unsupported claim

Ed

shown promising for AL in NLP due to its good qualitative and computational performance

paper_19.txt

1,801

1,824

Format

Ed

Shelmanov et al. (2021

paper_20.txt

252

631

Coherence

Ed

Following Chen et al. (2020c), other works adopt PLMs for few-shot D2T generation (Chang et al., 2021b;Su et al., 2021a). Kale and Rastogi (2020b) and Ribeiro et al. (2020) showed that PLMs using linearized representations of data can outperform graph neural networks on graph-to-text datasets, recently surpassed again by graph-based models (Ke et al., 2021;Chen et al., 2020a)

paper_20.txt

1,781

1,797

Format

Ed

Recently, have

paper_20.txt

3,514

3,533

Format

Ed

Jiang et al., 2020)

paper_37.txt

1,213

1,280

Format

Ed

Radford et al., 2021;Schick and Schütze, 2020a,b;Brown et al., 2020

paper_37.txt

1,544

1,572

Format

Ed

Schick and Schütze, 2020a,b)

paper_37.txt

992

1,071

Unsupported claim

Ed

they are impractical to use in real-world applications due to their model sizes

paper_37.txt

1,100

1,692

Lacks synthesis

Ed

Providing prompts or task descriptions play an vital role in improving pre-trained language models in many tasks Radford et al., 2021;Schick and Schütze, 2020a,b;Brown et al., 2020). Among them, GPT models (Radford et al., 2019;Brown et al., 2020) achieved great success in prompting or task demonstrations in NLP tasks. In light of this direction, prompt-based approaches improve small pre-trained models in few-shot text classification tasks Schick and Schütze, 2020a,b). CLIP (Radford et al., 2021) also explores prompt templates for image classification which affect zero-shot performance

paper_37.txt

49

932

Lacks synthesis

Ed

Recently, several few-shot learners on vision-language tasks were proposed including GPT (Radford et al., 2019;Brown et al., 2020), Frozen (Tsimpoukelli et al., 2021), PICa , and SimVLM . Frozen (Tsimpoukelli et al., 2021) is a large language model based on GPT-2 (Radford et al., 2019), and is transformed into a multimodal few-shot learner by extending the soft prompting to incorporate a set of images and text. Their approach shows the fewshot capability on visual question answering and image classification tasks. Similarly, PICa uses GPT-3 (Brown et al., 2020) to solve VQA tasks in a few-shot manner by providing a few in-context VQA examples. It converts images into textual descriptions so that GPT-3 can understand the images. SimVLM is trained with prefix language modeling on weakly-supervised datasets. It demonstrates its effectiveness on a zero-shot captioning task

paper_38.txt

24

794

Lacks synthesis

Ed

pre-training a transformer model on a large corpus with language modeling tasks and finetuning it on different downstream tasks has become the main transfer learning paradigm in natural language processing (Devlin et al., 2019). Notably, this paradigm requires updating and storing all the model parameters for every downstream task. As the model size proliferates (e.g., 330M parameters for BERT (Devlin et al., 2019) and 175B for GPT-3 (Brown et al., 2020)), it becomes computationally expensive and challenging to fine-tune the entire pre-trained language model (LM). Thus, it is natural to ask the question of whether we can transfer the knowledge of a pre-trained LM into downstream tasks by tuning only a small portion of its parameters with most of them freezing.

paper_38.txt

872

1,390

Lacks synthesis

Ed

One line of research (Li and Liang, 2021) suggests to augment the model with a few small trainable mod-ules and freeze the original transformer weight. Take Adapter (Houlsby et al., 2019;Pfeiffer et al., 2020a,b) and Compacter (Mahabadi et al., 2021) for example, both of them insert a small set of additional modules between each transformer layer. During fine-tuning, only these additional and taskspecific modules are trained, reducing the trainable parameters to ∼ 1-3% of the original transformer model per task.

paper_38.txt

1,434

1,921

Lacks synthesis

Ed

The GPT-3 models (Brown et al., 2020;Schick and Schütze, 2020) find that with proper manual prompts, a pre-trained LM can successfully match the fine-tuning performance of BERT models. LM-BFF (Gao et al., 2020), EFL (Wang et al., 2021), and AutoPrompt (Shin et al., 2020) further this direction by insert prompts in the input embedding layer. However, these methods rely on grid-search for a natural language-based prompt from a large search space, resulting in difficulties to optimize.

paper_38.txt

2,461

2,602

Unsupported claim

Ed

all existing prompt-tuning methods have thus far focused on task-specific prompts, making them incompatible with the traditional LM objective

paper_38.txt

2,617

2,711

Unsupported claim

Ed

it is unlikely to see many different sentences with the same prefix in the pre-training corpus

paper_39.txt

129

213

Format

Ed

(Eric et al., 2017;Wu et al., 2019; and collections of largescale annotation corpora

paper_39.txt

355

377

Format

Ed

(El Asri et al., 2017

paper_39.txt

530

533

Unsupported claim

Ed

SGD

paper_39.txt

874

909

Format

Ed

Quan et al., 2020;Lin et al., 2021)

paper_39.txt

998

1,151

Unsupported claim

Ed

vast majority of existing multilingual ToD datasets do not consider the real use cases when using a ToD system to search for local entities in a country.

paper_40.txt

396

437

Format

Ed

[Levy et al., 2017, Elsahar et al., 2018

paper_41.txt

1,890

2,370

Lacks synthesis

Ed

Previous work has shown that SNLI (Bowman et al., 2015) and MNLI (Williams et al., 2018) have annotation artifacts (e.g., negation is a strong indicator of contradictions) (Gururangan et al., 2018). The literature has also shown that simple adversarial attacks including negation cues are very effective (Naik et al., 2018;Wallace et al., 2019). Kovatchev et al. (2019) analyze 11 paraphrasing systems and show that they obtain substantially worse results when negation is present

paper_41.txt

3,201

3,272

Format

Ed

Bar-Haim et al., 2006;Giampiccolo et al., 2007;Bentivogli et al., 2009)

paper_43.txt

465

524

Unsupported claim

Ed

Machine Translation (MT) is the mainstream approach for GEC

paper_43.txt

918

966

Unsupported claim

Ed

recent powerful Transformer-based Seq2Seq model

paper_100.txt

594

825

Coherence

Ekaterina

Zhang et al. (2019) improves an LSTM- based encoder-decoder model with online vocabulary adaptation. For abbreviated pinyin, CoCAT (Huang et al., 2015) uses machine translation technology to reduce the number of the typing letters.

paper_100.txt

968

1,179

Coherence

Ekaterina

Our work differs from existing works in that we are the first one to exploit GPT and verify the pros and cons of GPT in different situations. In addition, there are some works handling pinyin with typing errors.

paper_100.txt

1,763

2,087

Coherence

Ekaterina

Sun et al. (2021) propose a general-purpose Chinese BERT with new embedding layers to inject pinyin and glyph information of characters. There are also task-specific BERT models, especially for the task of grammatical error correction since an important type of error is caused by characters pronounced with the same pinyin.

paper_11.txt

1,025

1,042

Unsupported claim

Ekaterina

multilingual BERT

paper_11.txt

1,485

1,487

Format

Ekaterina

1

paper_13.txt

14

105

Unsupported claim

Ekaterina

Few-shot learning is the problem of learning classifiers with only a few training examples.

paper_13.txt

267

444

Unsupported claim

Ekaterina

For text data, this is usually accomplished by representing the labels of the task in a textual form, which can either be the name of the label or a concise textual description.

paper_13.txt

1,713

1,880

Unsupported claim

Ekaterina

In contrast, the models typically applied in the entailment approach are Cross Attention (CA) models which need to be executed for every combination of text and label.

paper_13.txt

1,881

2,119

Unsupported claim

Ekaterina

On the other hand, they allow for interaction between the tokens of label and input, so that in theory they should be superior in classification accuracy. However, in this work we show that in practice, the difference in quality is small.

paper_13.txt

2,121

2,383

Unsupported claim

Ekaterina

Both CA and SNs also support the few-shot learning setup by fine-tuning the models on a small number of labeled examples. This is usually done by updating all parameters of the model, which in turn makes it impossible to share the models between different tasks.

paper_13.txt

3,365

3,483

Unsupported claim

Ekaterina

In contrast to most prior work, we also show that these results can also be achieved for languages other than English.

paper_14.txt

14

181

Unsupported claim

Ekaterina

At present, for a large majority of natural language processing tasks, the most successful approach is fine-tuning pre-trained models with task-specific labelled data.

paper_15.txt

649

715

Unsupported claim

Ekaterina

Researchers also realize that the vision modality maybe redundant.

paper_15.txt

897

902

Format

Ekaterina

2021)

paper_15.txt

864

956

Coherence

Ekaterina

Encouraging results appeared in 2021) proposed a cross-lingual visual pretraining approach.

paper_15.txt

14

1,184

Lacks synthesis

Ekaterina

Multimodal machine translation is a cross-domain task in the filed of machine translation. Early attempts mainly focused on enhancing the MMT model by better incorporation of the vision features (Calixto and Liu, 2017;Elliott and Kádár, 2017;Delbrouck and Dupont, 2017). However, directly encoding the whole image feature brings additional noise to the text (Yao and Wan, 2020;Liu et al., 2021a). To address the above issue, Yao and Wan (2020) proposed a multimodal self-attention to consider the relative difference of information between two modalities. Similarly, Liu et al. (2021a) used a Gumbel Softmax to achieve the same goal. Researchers also realize that the vision modality maybe redundant. Irrelevant images have little impact on the translation quality, and no significant BLEU drop is observed even the image is absent (Elliott, 2018). Encouraging results appeared in 2021) proposed a cross-lingual visual pretraining approach. In this work, we make a systematic study on whether stronger vision features are helpful. We also extend the research to enhanced features, such as object-detection and image captioning, which is complementary to previous work.

paper_16.txt

856

906

Unsupported claim

Ekaterina

AM-PERE for proposition classification in reviews

paper_16.txt

1,093

1,621

Coherence

Ekaterina

There are also some explorations on research articles (Teufel et al., 1999;Liakata et al., 2010;Lauscher et al., 2018), which differ in nature from the peer review domain. A wide range of control perspectives has been explored in controllable generation, including style control (e.g., sentiments (Duan et al., 2020), politeness (Madaan et al., 2020), formality , domains (Takeno et al., 2017) and persona ) and content control (e.g., length (Duan et al., 2020), entities (Fan et al., 2018a), and keywords (Tang et al., 2019)).

paper_17.txt

1,074

1,169

Unsupported claim

Ekaterina

BART-Gen is also a generation-based model focusing on documentlevel event argument extraction.

paper_17.txt

1,397

1,586

Unsupported claim

Ekaterina

However, their designs are not specific for low-resource scenarios, hence, these models can not enjoy all the benefits that DEGREE obtains for low-resource event extraction at the same time

paper_17.txt

1,726

2,044

Coherence

Ekaterina

Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence

paper_17.txt

2,261

2,403

Coherence

Ekaterina

Another thread of works are using meta-learning to deal with the less label challenge (Deng et al., 2020;Shen et al., 2021;Cong et al., 2021).

paper_17.txt

1,619

2,259

Lacks synthesis

Ekaterina

Low-resource event extraction. It has been a rising interest in event extraction under less data scenario. Liu et al. (2020) uses a machine reading comprehension formulation to conduct event extraction in a low-resource regime. Text2Event (Lu et al., 2021), a sequence-to-structure generation paradigm, first presents events in a linearized format, and then trains a generative model to generate the linearized event sequence. Text2Event's unnatural output format hinders the model from fully leveraging pre-trained knowledge. Hence, their model falls short on the cases with only extremely low data being available (as shown in Section 3).

paper_18.txt

952

970

Format

Ekaterina

(Li et al., 2020a;

paper_18.txt

1,391

1,496

Unsupported claim

Ekaterina

Nevertheless, building the lexicon is time-consuming and the quality of the lexicon may not be satisfied.

paper_18.txt

2,131

2,253

Unsupported claim

Ekaterina

However, too immersed regularity leads to unfavorable boundary detection of entities and disturbing character composition.

paper_18.txt

3,533

3,547

Unsupported claim

Ekaterina

OntoNotes V4.0

paper_18.txt

3,549

3,563

Unsupported claim

Ekaterina

OntoNotes V5.0

paper_18.txt

3,569

3,573

Unsupported claim

Ekaterina

MSRA

paper_18.txt

3,576

3,729

Unsupported claim

Ekaterina

The results show that RICON achieves considerable improvements compared to the state-of-the-art models, even outperforming existing lexicon-based models.

paper_18.txt

3,757

3,792

Unsupported claim

Ekaterina

a practical medical dataset (CBLUE)

paper_19.txt

14

174

Unsupported claim

Ekaterina

Deep learning, to a large extent, has freed data scientists from doing feature engineering, which has been one of the essential obstacles to annotation with AL.

paper_19.txt

723

900

Unsupported claim

Ekaterina

In our work, we take MNLP as a query strategy for experiments on sequence tagging tasks since it has demonstrated a good trade-off between quality and computational performance.

paper_19.txt

1,464

1,655

Unsupported claim

Ekaterina

We continue this line of works by relying on pre-trained Transformers since this architecture has been shown promising for AL in NLP due to its good qualitative and computational performance.

paper_19.txt

1,802

1,824

Format

Ekaterina

Shelmanov et al. (2021

paper_19.txt

2,873

3,303

Lacks synthesis

Ekaterina

Recently proposed alternatives to uncertaintybased query strategies leverage reinforcement learning and imitation learning (Fang et al., 2017;Liu et al., 2018;Vu et al., 2019;Brantley et al., 2020). This series of works aims at constructing trainable policy-based query strategies. However, this requires an excessive amount of computation while the transferability of learned policies across domains and tasks is underresearched.

paper_19.txt

3,692

3,717

Format

Ekaterina

(Shelmanov et al., 2021).

paper_19.txt

3,422

3,433

Unsupported claim

Ekaterina

ASM problem

paper_20.txt

1,003

1,049

Format

Ekaterina

(Heidari et al., 2021;Kale and Rastogi, 2020a;

paper_20.txt

791

1,331

Coherence

Ekaterina

Generation Using simple handcrafted templates for individual keys or predicates is an efficient way of introducing domain knowledge while preventing text-to-text models from overfitting to a specific data format (Heidari et al., 2021;Kale and Rastogi, 2020a;. Transforming individual triples to text is also used in Laha et al. (2020) whose work is the most similar to ours. They also build a three-step pipeline for zero-shot D2T generation, but they use handcrafted rules for producing the output text and do not address content planning.

paper_20.txt

14

760

Lacks synthesis

Ekaterina

D2T Generation with PLMs Large neural language models pretrained on self-supervised tasks (Lewis et al., 2020;Liu et al., 2019;Devlin et al., 2019) have recently gained a lot of traction in D2T generation research (Ferreira et al., 2020). Following Chen et al. (2020c), other works adopt PLMs for few-shot D2T generation (Chang et al., 2021b;Su et al., 2021a). Kale and Rastogi (2020b) and Ribeiro et al. (2020) showed that PLMs using linearized representations of data can outperform graph neural networks on graph-to-text datasets, recently surpassed again by graph-based models (Ke et al., 2021;Chen et al., 2020a). Although the models make use of general-domain pretraining tasks, all of them are eventually finetuned on domain-specific data.

paper_20.txt

1,583

1,870

Coherence

Ekaterina

As previously demonstrated, using a content plan in neural D2T generation has important impact on the overall text quality (Moryossef et al., 2019a,b;Puduppully et al., 2019;Trisedya et al., 2020). Recently, have shown that using a content plan leads to improved quality of PLM outputs.

paper_20.txt

2,366

2,388

Format

Ekaterina

Li and Jurafsky, 2017)

paper_20.txt

3,514

3,533

Format

Ekaterina

Jiang et al., 2020)

paper_20.txt

4,026

4,046

Format

Ekaterina

(Botha et al., 2018;

paper_20.txt

3,615

4,047

Coherence

Ekaterina

In contrast to sentence fusion (Geva et al., 2019;Barzilay and McKeown, 2005) or sentence compression (Filippova and Altun, 2013), we operate in the context of multiple sentences in a paragraph. The task is the central focus of our WIKIFLUENT corpus ( §4), which we synthesize using a model for the reverse task, split-andrephrase, i.e. splitting a complex sentence into simpler ones while preserving semantics (Botha et al., 2018;.

paper_53.txt

225

268

Format

Ekaterina

(Davison et al., 2019;Petroni et al., 2019;

paper_53.txt

364

385

Format

Ekaterina

(Trinh and Le, 2018;)

paper_53.txt

3,472

3,609

Unsupported claim

Ekaterina

DynaGen uses pretrained commonsense models to generate implications of a question and expands the inference input with these generations

paper_54.txt

441

509

Format

Ekaterina

(Zhang et al., 2020;Aghajanyan et al., 2020Aghajanyan et al., , 2021

paper_54.txt

1,943

2,069

Coherence

Ekaterina

inspired by Li et al. (2020), to train a machine learning model to smooth out such noise and to speed up the inference process

paper_54.txt

2,071

2,213

Lacks synthesis

Ekaterina

Different from Li et al. (2020), we propose to utilize non-autoregressive text generators, which generate all tokens in the output in parallel

paper_54.txt

3,248

3,521

Unsupported claim

Ekaterina

Regarding inference efficiency, our NAUS with truncating is 1000 times more efficient than the search approach; even with dynamic programming for length control, NAUS is still 100 times more efficient than search and several times more efficient than autoregressive models.

paper_55.txt

2,878

3,084

Unsupported claim

Ekaterina

Lastly, our dataset yields a novel extractive summarization dataset, providing a benchmark for studying domain transfer in summarization and enabling QA models to provide concise answers to complex queries.

paper_56.txt

1,389

1,551

Unsupported claim

Ekaterina

On the other hand, neural conversational models are not necessarily designed to generate faithful outputs, but to mimic the distributional properties of the data.

paper_56.txt

2,734

2,863

Unsupported claim

Ekaterina

we annotate responses generated by several state-of-the-art models, including ones that are designed to alleviate hallucinations.

Datasets:

ImanAndrea
/

citation_annotations

Space using ImanAndrea/citation_annotations 1