Search

Annotating long-document question answering (long-document QA) pairs is time-consuming and expensive. To alleviate the problem, it might be possible to generate long-document QA pairs via unsupervised question answering (UQA) methods. However, existing UQA tasks are based on short documents, and can hardly incorporate long-range information. To tackle the problem, we propose a new task, named unsupervised long-document question answering (ULQA), aiming to generate high-quality long-document QA instances in an unsupervised manner. Besides, we propose AttenWalker, a novel unsupervised method to aggregate and generate answers with long-range dependency so as to construct long-document QA pairs. Specifically, AttenWalker is composed of three modules, i.e., span collector, span linker and answer aggregator. Firstly, the span collector takes advantage of constituent parsing and reconstruction loss to select informative candidate spans for constructing answers. Secondly, by going through the attention graph of a pre-trained long-document model, potentially interrelated text spans (that might be far apart) could be linked together via an attention-walking algorithm. Thirdly, in the answer aggregator, linked spans are aggregated into the final answer via the mask-filling ability of a pre-trained model. Extensive experiments show that AttenWalker outperforms previous methods on Qasper and NarrativeQA. In addition, AttenWalker also shows strong performance in the few-shot learning setting., Accepted to the Findings of ACL 2023

Open-retrieval conversational machine reading comprehension (OCMRC) simulates real-life conversational interaction scenes. Machines are required to make a decision of "Yes/No/Inquire" or generate a follow-up question when the decision is "Inquire" based on retrieved rule texts, user scenario, user question, and dialogue history. Recent studies explored the methods to reduce the information gap between decision-making and question generation and thus improve the performance of generation. However, the information gap still exists because these pipeline structures are still limited in decision-making, span extraction, and question rephrasing three stages. Decision-making and generation are reasoning separately, and the entailment reasoning utilized in decision-making is hard to share through all stages. To tackle the above problem, we proposed a novel one-stage end-to-end framework, called Entailment Fused-T5 (EFT), to bridge the information gap between decision-making and generation in a global understanding manner. The extensive experimental results demonstrate that our proposed framework achieves new state-of-the-art performance on the OR-ShARC benchmark.

Prompt learning recently become an effective linguistic tool to motivate the PLMs' knowledge on few-shot-setting tasks. However, studies have shown the lack of robustness still exists in prompt learning, since suitable initialization of continuous prompt and expert-first manual prompt are essential in fine-tuning process. What is more, human also utilize their comparative ability to motivate their existing knowledge for distinguishing different examples. Motivated by this, we explore how to use contrastive samples to strengthen prompt learning. In detail, we first propose our model ConsPrompt combining with prompt encoding network, contrastive sampling module, and contrastive scoring module. Subsequently, two sampling strategies, similarity-based and label-based strategies, are introduced to realize differential contrastive learning. The effectiveness of proposed ConsPrompt is demonstrated in five different few-shot learning tasks and shown the similarity-based sampling strategy is more effective than label-based in combining contrastive learning. Our results also exhibits the state-of-the-art performance and robustness in different few-shot settings, which proves that the ConsPrompt could be assumed as a better knowledge probe to motivate PLMs.

Pretraining-based (PT-based) automatic evaluation metrics (e.g., BERTScore and BARTScore) have been widely used in several sentence generation tasks (e.g., machine translation and text summarization) due to their better correlation with human judgments over traditional overlap-based methods. Although PT-based methods have become the de facto standard for training grammatical error correction (GEC) systems, GEC evaluation still does not benefit from pretrained knowledge. This paper takes the first step towards understanding and improving GEC evaluation with pretraining. We first find that arbitrarily applying PT-based metrics to GEC evaluation brings unsatisfactory correlation results because of the excessive attention to inessential systems outputs (e.g., unchanged parts). To alleviate the limitation, we propose a novel GEC evaluation metric to achieve the best of both worlds, namely PT-M2 which only uses PT-based metrics to score those corrected parts. Experimental results on the CoNLL14 evaluation task show that PT-M2 significantly outperforms existing methods, achieving a new state-of-the-art result of 0.949 Pearson correlation. Further analysis reveals that PT-M2 is robust to evaluate competitive GEC systems. Source code and scripts are freely available at https://github.com/pygongnlp/PT-M2., Accepted to EMNLP 2022

Recently, to improve the unsupervised image retrieval performance, plenty of unsupervised hashing methods have been proposed by designing a semantic similarity matrix, which is based on the similarities between image features extracted by a pre-trained CNN model. However, most of these methods tend to ignore high-level abstract semantic concepts contained in images. Intuitively, concepts play an important role in calculating the similarity among images. In real-world scenarios, each image is associated with some concepts, and the similarity between two images will be larger if they share more identical concepts. Inspired by the above intuition, in this work, we propose a novel Unsupervised Hashing with Semantic Concept Mining, called UHSCM, which leverages a VLP model to construct a high-quality similarity matrix. Specifically, a set of randomly chosen concepts is first collected. Then, by employing a vision-language pretraining (VLP) model with the prompt engineering which has shown strong power in visual representation learning, the set of concepts is denoised according to the training images. Next, the proposed method UHSCM applies the VLP model with prompting again to mine the concept distribution of each image and construct a high-quality semantic similarity matrix based on the mined concept distributions. Finally, with the semantic similarity matrix as guiding information, a novel hashing loss with a modified contrastive loss based regularization item is proposed to optimize the hashing network. Extensive experiments on three benchmark datasets show that the proposed method outperforms the state-of-the-art baselines in the image retrieval task.

Even as an important lexical information for Latin languages, word case is often ignored in machine translation. According to observations, the translation performance drops significantly when we introduce case-sensitive evaluation metrics. In this paper, we introduce two types of case-sensitive neural machine translation (NMT) approaches to alleviate the above problems: i) adding case tokens into the decoding sequence, and ii) adopting case prediction to the conventional NMT. Our proposed approaches incorporate case information to the NMT decoder by jointly learning target word generation and word case prediction. We compare our approaches with multiple kinds of baselines including NMT with naive case-restoration methods and analyze the impacts of various setups on our approaches. Experimental results on three typical translation tasks (Zh-En, En-Fr, En-De) show that our proposed methods lead to the improvements up to 2.5, 1.0 and 0.5 in case-sensitive BLEU scores respectively. Further analyses also illustrate the inherent reasons why our approaches lead to different improvements on different translation tasks.

Recent progress in deep learning has continuously improved the accuracy of dialogue response selection. In particular, sophisticated neural network architectures are leveraged to capture the rich interactions between dialogue context and response candidates. While remarkably effective, these models also bring in a steep increase in computational cost. Consequently, such models can only be used as a re-rank module in practice. In this study, we present a solution to directly select proper responses from a large corpus or even a nonparallel corpus that only consists of unpaired sentences, using a dense retrieval model. To push the limits of dense retrieval, we design an interaction layer upon the dense retrieval models and apply a set of tailor-designed learning strategies. Our model shows superiority over strong baselines on the conventional re-rank evaluation setting, which is remarkable given its efficiency. To verify the effectiveness of our approach in realistic scenarios, we also conduct full-rank evaluation, where the target is to select proper responses from a full candidate pool that may contain millions of candidates and evaluate them fairly through human annotations. Our proposed model notably outperforms pipeline baselines that integrate fast recall and expressive re-rank modules. Human evaluation results show that enlarging the candidate pool with nonparallel corpora improves response quality further., 11 pages, 4 figures, 6 tables

In this paper, we introduce ELECTRA-style tasks to cross-lingual language model pre-training. Specifically, we present two pre-training tasks, namely multilingual replaced token detection, and translation replaced token detection. Besides, we pretrain the model, named as XLM-E, on both multilingual and parallel corpora. Our model outperforms the baseline models on various cross-lingual understanding tasks with much less computation cost. Moreover, analysis shows that XLM-E tends to obtain better cross-lingual transferability., ACL-2022

Since the pre-trained language models are widely used, retrieval-based open-domain dialog systems, have attracted considerable attention from researchers recently. Most of the previous works select a suitable response only according to the matching degree between the query and each individual candidate response. Although good performance has been achieved, these recent works ignore the comparison among the candidate responses, which could provide rich information for selecting the most appropriate response. Intuitively, better decisions could be made when the models can get access to the comparison information among all the candidate responses. In order to leverage the comparison information among the candidate responses, in this paper, we propose a novel and plug-in Self-attention Comparison Module for retrieval-based open-domain dialog systems, called SCM. Extensive experiment results demonstrate that our proposed self-attention comparison module effectively boosts the performance of the existing retrieval-based open-domain dialog systems. Besides, we have publicly released our source codes for future research.

Due to their high retrieval efficiency and low storage cost for cross-modal search task, cross-modal hashing methods have attracted considerable attention. For the supervised cross-modal hashing methods, how to make the learned hash codes preserve semantic information sufficiently contained in the label of datapoints is the key to further enhance the retrieval performance. Hence, almost all supervised cross-modal hashing methods usually depends on defining a similarity between datapoints with the label information to guide the hashing model learning fully or partly. However, the defined similarity between datapoints can only capture the label information of datapoints partially and misses abundant semantic information, then hinders the further improvement of retrieval performance. Thus, in this paper, different from previous works, we propose a novel cross-modal hashing method without defining the similarity between datapoints, called Deep Cross-modal Hashing via \textit{Margin-dynamic-softmax Loss} (DCHML). Specifically, DCHML first trains a proxy hashing network to transform each category information of a dataset into a semantic discriminative hash code, called proxy hash code. Each proxy hash code can preserve the semantic information of its corresponding category well. Next, without defining the similarity between datapoints to supervise the training process of the modality-specific hashing networks , we propose a novel \textit{margin-dynamic-softmax loss} to directly utilize the proxy hashing codes as supervised information. Finally, by minimizing the novel \textit{margin-dynamic-softmax loss}, the modality-specific hashing networks can be trained to generate hash codes which can simultaneously preserve the cross-modal similarity and abundant semantic information well.

Currently, open-domain generative dialog systems have attracted considerable attention in academia and industry. Despite the success of single-turn dialog generation, multi-turn dialog generation is still a big challenge. So far, there are two kinds of models for open-domain multi-turn dialog generation: hierarchical and non-hierarchical models. Recently, some works have shown that the hierarchical models are better than non-hierarchical models under their experimental settings; meanwhile, some works also demonstrate the opposite conclusion. Due to the lack of adequate comparisons, it's not clear which kind of models are better in open-domain multi-turn dialog generation. Thus, in this paper, we will measure systematically nearly all representative hierarchical and non-hierarchical models over the same experimental settings to check which kind is better. Through extensive experiments, we have the following three important conclusions: (1) Nearly all hierarchical models are worse than non-hierarchical models in open-domain multi-turn dialog generation, except for the HRAN model. Through further analysis, the excellent performance of HRAN mainly depends on its word-level attention mechanism; (2) The performance of other hierarchical models will also obtain a great improvement if integrating the word-level attention mechanism into these models. The modified hierarchical models even significantly outperform the non-hierarchical models; (3) The reason why the word-level attention mechanism is so powerful for hierarchical models is because it can leverage context information more effectively, especially the fine-grained information. Besides, we have implemented all of the models and already released the codes.

Recently, open-domain dialogue systems have attracted growing attention. Most of them use the sequence-to-sequence (Seq2Seq) architecture to generate responses. However, traditional Seq2Seq-based open-domain dialogue models tend to generate generic and safe responses, which are less informative, unlike human responses. In this paper, we propose a simple but effective keywords-guided Sequence-to-Sequence model (KW-Seq2Seq) which uses keywords information as guidance to generate open-domain dialogue responses. Specifically, KW-Seq2Seq first uses a keywords decoder to predict some topic keywords, and then generates the final response under the guidance of them. Extensive experiments demonstrate that the KW-Seq2Seq model produces more informative, coherent and fluent responses, yielding substantive gain in both automatic and human evaluation metrics.

As a key component in a dialogue system, dialogue state tracking plays an important role. It is very important for dialogue state tracking to deal with the problem of unknown slot values. As far as we known, almost all existing approaches depend on pointer network to solve the unknown slot value problem. These pointer network-based methods usually have a hidden assumption that there is at most one out-of-vocabulary word in an unknown slot value because of the character of a pointer network. However, often, there are multiple out-of-vocabulary words in an unknown slot value, and it makes the existing methods perform bad. To tackle the problem, in this paper, we propose a novel Context-Sensitive Generation network (CSG) which can facilitate the representation of out-of-vocabulary words when generating the unknown slot value. Extensive experiments show that our proposed method performs better than the state-of-the-art baselines.

Relation ties, defined as the correlation and mutual exclusion between different relations, are critical for distant supervised relation extraction. Existing approaches model this property by greedily learning local dependencies. However, they are essentially limited by failing to capture the global topology structure of relation ties. As a result, they may easily fall into a locally optimal solution. To solve this problem, in this paper, we propose a novel force-directed graph based relation extraction model to comprehensively learn relation ties. Specifically, we first build a graph according to the global co-occurrence of relations. Then, we borrow the idea of Coulomb's Law from physics and introduce the concept of attractive force and repulsive force to this graph to learn correlation and mutual exclusion between relations. Finally, the obtained relation representations are applied as an inter-dependent relation classifier. Experimental results on a large scale benchmark dataset demonstrate that our model is capable of modeling global relation ties and significantly outperforms other baselines. Furthermore, the proposed force-directed graph can be used as a module to augment existing relation extraction systems and improve their performance., Learning Relation Ties

Back translation (BT) has been widely used and become one of standard techniques for data augmentation in Neural Machine Translation (NMT), BT has proven to be helpful for improving the performance of translation effectively, especially for low-resource scenarios. While most works related to BT mainly focus on European languages, few of them study languages in other areas around the world. In this paper, we investigate the impacts of BT on Asia language translations between the extremely low-resource Chinese and Vietnamese language pair. We evaluate and compare the effects of different sizes of synthetic data on both NMT and Statistical Machine Translation (SMT) models for Chinese to Vietnamese and Vietnamese to Chinese, with character-based and word-based settings. Some conclusions from previous works are partially confirmed and we also draw some other interesting findings and conclusions, which are beneficial to understand BT further.

Recently, Chinese word segmentation (CWS) methods using neural networks have made impressive progress. Most of them regard the CWS as a sequence labeling problem which construct models based on local features rather than considering global information of input sequence. In this paper, we cast the CWS as a sequence translation problem and propose a novel sequence-to-sequence CWS model with an attention-based encoder-decoder framework. The model captures the global information from the input and directly outputs the segmented sequence. It can also tackle other NLP tasks with CWS jointly in an end-to-end mode. Experiments on Weibo, PKU and MSRA benchmark datasets show that our approach has achieved competitive performances compared with state-of-the-art methods. Meanwhile, we successfully applied our proposed model to jointly learning CWS and Chinese spelling correction, which demonstrates its applicability of multi-task fusion., In proceedings of SMP 2017 (Chinese National Conference on Social Media Processing)

In this paper, we transform tag recommendation into a word-based text generation problem and introduce a sequence-to-sequence model. The model inherits the advantages of LSTM-based encoder for sequential modeling and attention-based decoder with local positional encodings for learning relations globally. Experimental results on Zhihu datasets illustrate the proposed model outperforms other state-of-the-art text classification based methods., This is a full length version of the paper in DASFAA 2019

Multilingual pretrained language models (such as multilingual BERT) have achieved impressive results for cross-lingual transfer. However, due to the constant model capacity, multilingual pre-training usually lags behind the monolingual competitors. In this work, we present two approaches to improve zero-shot cross-lingual classification, by transferring the knowledge from monolingual pretrained models to multilingual ones. Experimental results on two cross-lingual classification benchmarks show that our methods outperform vanilla multilingual fine-tuning., 5 pages

We introduce a new scientific named entity recognizer called SEPT, which stands for Span Extractor with Pre-trained Transformers. In recent papers, span extractors have been demonstrated to be a powerful model compared with sequence labeling models. However, we discover that with the development of pre-trained language models, the performance of span extractors appears to become similar to sequence labeling models. To keep the advantages of span representation, we modified the model by under-sampling to balance the positive and negative samples and reduce the search space. Furthermore, we simplify the origin network architecture to combine the span extractor with BERT. Experiments demonstrate that even simplified architecture achieves the same performance and SEPT achieves a new state of the art result in scientific named entity recognition even without relation information involved., This work is outdated. The result should not be trusted

The Hadamard matrix with orthogonality is a more important modulation matrix for computational ghost imaging (CGI), especially its optimized Hadamard matrix. However, as far as we know, little mention has been paid to efficient and convenient Hadamard matrix generation for CGI. The existing methods are to reconstruct any row of Hadamard matrix into two-dimensional matrix and then optimize it. In this work, we propose a Hadamard `pipeline' coding computational ghost imaging approach, which can directly generate two-dimensional Hadamard derived pattern and Hadamard optimization sequence, whereby both the memory consumption and the complexity of coding implementation for CGI can be significantly reduced. The optimization method of commonly used hadamard optimization sequence implementation is also discussed. This method provides a new approach for Hadamard sequence optimization and ghost imaging applications.

There is an increasing demand for task-oriented dialogue systems which can assist users in various activities such as booking tickets and restaurant reservations. In order to complete dialogues effectively, dialogue policy plays a key role in task-oriented dialogue systems. As far as we know, the existing task-oriented dialogue systems obtain the dialogue policy through classification, which can assign either a dialogue act and its corresponding parameters or multiple dialogue acts without their corresponding parameters for a dialogue action. In fact, a good dialogue policy should construct multiple dialogue acts and their corresponding parameters at the same time. However, it's hard for existing classification-based methods to achieve this goal. Thus, to address the issue above, we propose a novel generative dialogue policy learning method. Specifically, the proposed method uses attention mechanism to find relevant segments of given dialogue context and input utterance and then constructs the dialogue policy by a seq2seq way for task-oriented dialogue systems. Extensive experiments on two benchmark datasets show that the proposed model significantly outperforms the state-of-the-art baselines. In addition, we have publicly released our codes.

Second language acquisition (SLA) modeling is to predict whether second language learners could correctly answer the questions according to what they have learned. It is a fundamental building block of the personalized learning system and has attracted more and more attention recently. However, as far as we know, almost all existing methods cannot work well in low-resource scenarios due to lacking of training data. Fortunately, there are some latent common patterns among different language-learning tasks, which gives us an opportunity to solve the low-resource SLA modeling problem. Inspired by this idea, in this paper, we propose a novel SLA modeling method, which learns the latent common patterns among different language-learning datasets by multi-task learning and are further applied to improving the prediction performance in low-resource scenarios. Extensive experiments show that the proposed method performs much better than the state-of-the-art baselines in the low-resource scenario. Meanwhile, it also obtains improvement slightly in the non-low-resource scenario.

The task of table structure recognition aims to recognize the internal structure of a table, which is a key step to make machines understand tables. Currently, there are lots of studies on this task for different file formats such as ASCII text and HTML. It also attracts lots of attention to recognize the table structures in PDF files. However, it is hard for the existing methods to accurately recognize the structure of complicated tables in PDF files. The complicated tables contain spanning cells which occupy at least two columns or rows. To address the issue, we propose a novel graph neural network for recognizing the table structure in PDF files, named GraphTSR. Specifically, it takes table cells as input, and then recognizes the table structures by predicting relations among cells. Moreover, to evaluate the task better, we construct a large-scale table structure recognition dataset from scientific papers, named SciTSR, which contains 15,000 tables from PDF files and their corresponding structure labels. Extensive experiments demonstrate that our proposed model is highly effective for complicated tables and outperforms state-of-the-art baselines over a benchmark dataset and our new constructed dataset.