We present Sailor, a family of open language models ranging from 0.5B to 14B parameters, tailored for South-East Asian (SEA) languages. From Qwen1.5, Sailor models accept 200B to 400B tokens during continual pre-training, primarily covering the languages of English, Chinese, Vietnamese, Thai, Indonesian, Malay, and Lao. The training leverages several techniques, including BPE dropout for improving the model robustness, aggressive data cleaning and deduplication, and small proxy models to optimize the data mixture. Experimental results on four typical tasks indicate that Sailor models demonstrate strong performance across different benchmarks, including commonsense reasoning, question answering, reading comprehension and examination. We share our insights to spark a wider interest in developing large language models for multilingual use cases.

Evaluating large language models (LLMs) is fundamental, particularly in the context of practical applications. Conventional evaluation methods, typically designed primarily for LLM development, yield numerical scores that ignore the user experience. Therefore, our study shifts the focus from model-centered to human-centered evaluation in the context of AI-powered writing assistance applications. Our proposed metric, termed “Revision Distance,” utilizes LLMs to suggest revision edits that mimic the human writing process. It is determined by counting the revision edits generated by LLMs. Benefiting from the generated revision edit details, our metric can provide a self-explained text evaluation result in a human-understandable manner beyond the context-independent score. Our results show that for the easy-writing task, “Revision Distance” is consistent with established metrics (ROUGE, Bert-score, and GPT-score), but offers more insightful, detailed feedback and better distinguishes between texts. Moreover, in the context of challenging academic writing tasks, our metric still delivers reliable evaluations where other metrics tend to struggle. Furthermore, our metric also holds significant potential for scenarios lacking reference texts.

In non-autoregressive translation (NAT), directed acyclic Transformers (DAT) have demonstrated their ability to achieve comparable performance to the autoregressive Transformers.In this paper, we first show that DAT is essentially a fully connected left-to-right Hidden Markov Model (HMM), with the source and target sequences being observations and the token positions being latent states.Even though generative models like HMM do not suffer from label bias in traditional task settings (e.g., sequence labeling), we argue here that the left-to-right HMM in NAT may still encounter this issue due to the missing observations at the inference stage.To combat label bias, we propose two constrained HMMs: 1) Adaptive Window HMM, which explicitly balances the number of outgoing transitions at different states; 2) Bi-directional HMM, i.e., a combination of left-to-right and right-to-left HMMs, whose uni-directional components can implicitly regularize each other’s biases via shared parameters.Experimental results on WMT’14 EnDe and WMT’17 ZhEn demonstrate that our methods can achieve better or comparable performance to the original DAT using various decoding methods.We also demonstrate that our methods effectively reduce the impact of label bias.

Teaching small-scale language models to perform math reasoning is a valuable yet challenging task. Besides obtaining labeled data from human experts, one of the most common ways to collect high-quality data is by sampling from a larger and more powerful language model. Although previous works have demonstrated the effectiveness of this method, such a knowledge distillation paradigm can be costly and unstable, especially considering that many large language models, such as GPT-4, are closed-sourced, proprietary, and their behaviors are unpredictable. In this work, to avoid relying on outputs from large models, we demonstrate that the reasoning abilities of small-scale language models can be enhanced through self-training, which involves training models with their own outputs. We also show that the vanilla self-training can be further augmented by an alignment algorithm, direct preference optimization (DPO). We empirically found that models trained with the DPO objective are capable of making better generations that largely benefit multi-turn self-training. The experiments show our models outperform the state-of-the-art models with comparable sizes on a series of downstream math reasoning tasks with minimal resource requirements.

The widespread use of pre-trained language models (PLMs) in natural language processing (NLP) has greatly improved performance outcomes. However, these models’ vulnerability to adversarial attacks (e.g., camouflaged hints from drug dealers), particularly in the Chinese language with its rich character diversity/variation and complex structures, hatches vital apprehension. In this study, we propose a novel method, CHinese vAriatioN Graph Enhancement (CHANGE), to increase the robustness of PLMs against character variation attacks in Chinese content. CHANGE presents a novel approach to incorporate a Chinese character variation graph into the PLMs. Through designing different supplementary tasks utilizing the graph structure, CHANGE essentially enhances PLMs’ interpretation of adversarially manipulated text. Experiments conducted in a multitude of NLP tasks show that CHANGE outperforms current language models in combating against adversarial attacks and serves as a valuable contribution to robust language model research. Moreover, these findings highlight the substantial potential of graph-guided pre-training strategies for real-world applications.

Recent advancements in pre-trained language models (PLMs) have demonstrated that these models possess some degree of syntactic awareness. To leverage this knowledge, we propose a novel chart-based method for extracting parse trees from masked language models (LMs) without the need to train separate parsers. Our method computes a score for each span based on the distortion of contextual representations resulting from linguistic perturbations. We design a set of perturbations motivated by the linguistic concept of constituency tests, and use these to score each span by aggregating the distortion scores. To produce a parse tree, we use chart parsing to find the tree with the minimum score. Our method consistently outperforms previous state-of-the-art methods on English with masked LMs, and also demonstrates superior performance in a multilingual setting, outperforming the state-of-the-art in 6 out of 8 languages. Notably, although our method does not involve parameter updates or extensive hyperparameter search, its performance can even surpass some unsupervised parsing methods that require fine-tuning. Our analysis highlights that the distortion of contextual representation resulting from syntactic perturbation can serve as an effective indicator of constituency across languages.

Fine-tuning pre-trained language models for multiple tasks can be expensive in terms of storage. Parameter-efficient transfer learning (PETL) methods have been proposed to address this issue, but they still require a significant number of parameters when being applied to broader ranges of tasks. To achieve even greater storage reduction, we propose ProPETL, a novel method that enables efficient sharing of a single prototype PETL network (e.g. adapter, LoRA, and prefix-tuning) across layers and tasks. We learn binary masks to select different sub-networks from the prototype network and apply them as PETL modules into different layers. We find that the binary masks can determine crucial structural information from the network, which is often ignored in previous studies. Our work can also be seen as a type of pruning method, where we find that overparameterization also exists in the seemingly small PETL modules. We evaluate ProPETL on various downstream tasks and show that it can outperform other PETL methods with around 10% parameters required by the latter.

We consider the task of generating designs directly from natural language descriptions, and consider floor plan generation as the initial research area. Language conditional generative models have recently been very successful in generating high-quality artistic images. However, designs must satisfy different constraints that are not present in generating artistic images, particularly spatial and relational constraints. We make multiple contributions to initiate research on this task. First, we introduce a novel dataset, Tell2Design (T2D), which contains more than 80k floor plan designs associated with natural language instructions. Second, we propose a Sequence-to-Sequence model that can serve as a strong baseline for future research. Third, we benchmark this task with several text-conditional image generation models. We conclude by conducting human evaluations on the generated samples and providing an analysis of human performance. We hope our contributions will propel the research on language-guided design generation forward.

Mathematical reasoning is regarded as a necessary ability for Language Models (LMs). Recent works demonstrate large LMs’ impressive performance in solving math problems. The success is attributed to their Chain-of-Thought (CoT) reasoning abilities, i.e., the ability to decompose complex questions into step-by-step reasoning chains, but such ability seems only to emerge from models with abundant parameters. This work investigates how to incorporate relatively small LMs with the capabilities of multi-step reasoning. We propose to inject such abilities by continually pre-training LMs on a synthetic dataset MsAT which is composed of Multi-step Arithmetic Tasks. Our experiments on four math word problem datasets show the effectiveness of the proposed method in enhancing LMs’ math reasoning abilities.

Distantly supervised named entity recognition (DS-NER) has been proposed to exploit the automatically labeled training data instead of human annotations. The distantly annotated datasets are often noisy and contain a considerable number of false negatives. The recent approach uses a weighted sampling approach to select a subset of negative samples for training. However, it requires a good classifier to assign weights to the negative samples. In this paper, we propose a simple and straightforward approach for selecting the top negative samples that have high similarities with all the positive samples for training. Our method achieves consistent performance improvements on four distantly supervised NER datasets. Our analysis also shows that it is critical to differentiate the true negatives from the false negatives.

The chain-of-though (CoT) prompting methods were successful in various natural language processing (NLP) tasks thanks to their ability to unveil the underlying complex reasoning processes. Such reasoning processes typically exhibit highly structured steps. Recent efforts also started investigating methods to encourage more structured reasoning procedures to be captured (cite least to most).In this work, we propose Tab-CoT, a novel tabular-format CoT prompting method, which allows the complex reasoning process to be explicitly modeled in a highly structured manner. Despite its simplicity, we show that our approach is capable of performing reasoning across multiple dimensions (i.e., both rows and columns).We demonstrate our approach’s strong zero-shot and few-shot capabilities through extensive experiments on a range of reasoning tasks.

Chain-of-thought (CoT) prompting with large language models has proven effective in numerous natural language process tasks, but designing prompts that generalize well to diverse problem types can be challenging CITATION, especially in the context of math word problem solving. Additionally, it is common to have a large amount of training data that have a better diversity coverage but CoT annotations are not available, which limits the use of supervised learning techniques. To address these issues, we investigate two approaches to leverage the training data in few-shot prompting scenario: dynamic program prompting and program distillation.Our approach is largely inspired by CITATION where they proposed to replace the CoT with the programs as the intermediate reasoning step. Such a prompting strategy allows us to accurately verify the answer correctness through program execution in MWP solving.Our dynamic program prompting involves annotating the training data by sampling correct programs from a large language model, while program distillation involves adapting a smaller model to the program-annotated training data.Our experiments on three standard MWP datasets demonstrate the effectiveness of these approaches, yielding significant improvements over previous baselines for prompting and fine-tuning.Our results suggest that leveraging a large amount of training data can improve the generalization ability of prompts and boost the performance of fine-tuned smaller models in MWP solving.

While prompt tuning approaches have achieved competitive performance with high efficiency, we observe that they invariably employ the same initialization process, wherein the soft prompt is either randomly initialized or derived from an existing embedding vocabulary. In contrast to these conventional methods, this study aims to investigate an alternative way to derive soft prompt. Our empirical studies show that the soft prompt typically exhibits a low “intrinsic rank” characteristic. With such observations, we propose decomposed prompt tuning, a novel approach that utilizes low-rank matrices to initialize the soft prompt. Through the low-rank reparameterization, our method significantly reduces the number of trainable parameters while maintaining effectiveness. Experimental results on the SuperGLUE benchmark in both high-resource and low-resource scenarios demonstrate the effectiveness of the proposed method.

Instruction tuning of open-source large language models (LLMs) like LLaMA, using direct outputs from more powerful LLMs such as Instruct-GPT and GPT-4, has proven to be a cost-effective way to align model behaviors with human preferences. However, the instruction-tuned model has only seen one response per instruction, lacking the knowledge of potentially better responses. In this paper, we propose finetuning an instruction-tuned LLM using our novel probabilistic ranking and contextual ranking approaches to increase the likelihood of generating better responses. Probabilistic ranking enables the instruction-tuned model to inherit the relative rankings of high-quality and low-quality responses from the teacher LLM. On the other hand, learning with contextual ranking allows the model to refine its own response distribution using the contextual understanding ability of stronger LLMs. Furthermore, we apply probabilistic ranking and contextual ranking sequentially to the instruction-tuned LLM. The resulting model, which we call Tuna, consistently improves the performance on Super Natural Instructions (119 test tasks), LMentry (25 test tasks), Vicuna QA, and can even obtain better results than several strong reinforcement learning baselines. Our code and data are available at https://github.com/microsoft/LMOps.

The underlying mechanism of neural networks in capturing precise knowledge has been the subject of consistent research efforts. In this work, we propose a theoretical approach based on Neural Tangent Kernels (NTKs) to investigate such mechanisms. Specifically, considering the infinite network width, we hypothesize the learning dynamics of target models may intuitively unravel the features they acquire from training data, deepening our insights into their internal mechanisms. We apply our approach to several fundamental models and reveal how these models leverage statistical features during gradient descent and how they are integrated into final decisions. We also discovered that the choice of activation function can affect feature extraction. For instance, the use of the ReLU activation function could potentially introduce a bias in features, providing a plausible explanation for its replacement with alternative functions in recent pre-trained language models. Additionally, we find that while self-attention and CNN models may exhibit limitations in learning n-grams, multiplication-based models seem to excel in this area. We verify these theoretical findings through experiments and find that they can be applied to analyze language modeling tasks, which can be regarded as a special variant of classification. Our work may offer insights into the roles and capacities of fundamental modules within deep neural networks including large language models.

Although self-attention based models such as Transformers have achieved remarkable successes on natural language processing (NLP)tasks, recent studies reveal that they have limitations on modeling sequential transformations (Hahn, 2020), which may promptre-examinations of recurrent neural networks (RNNs) that demonstrated impressive results on handling sequential data. Despite manyprior attempts to interpret RNNs, their internal mechanisms have not been fully understood, and the question on how exactly they capturesequential features remains largely unclear. In this work, we present a study that shows there actually exist some explainable componentsthat reside within the hidden states, which are reminiscent of the classical n-grams features. We evaluated such extracted explainable features from trained RNNs on downstream sentiment analysis tasks and found they could be used to model interesting linguistic phenomena such as negation and intensification. Furthermore, we examined the efficacy of using such n-gram components alone as encoders on tasks such as sentiment analysis and language modeling, revealing they could be playing important roles in contributing to the overall performance of RNNs. We hope our findings could add interpretability to RNN architectures, and also provide inspirations for proposing new architectures for sequential data.

Solving math word problems requires deductive reasoning over the quantities in the text. Various recent research efforts mostly relied on sequence-to-sequence or sequence-to-tree models to generate mathematical expressions without explicitly performing relational reasoning between quantities in the given context. While empirically effective, such approaches typically do not provide explanations for the generated expressions. In this work, we view the task as a complex relation extraction problem, proposing a novel approach that presents explainable deductive reasoning steps to iteratively construct target expressions, where each step involves a primitive operation over two quantities defining their relation. Through extensive experiments on four benchmark datasets, we show that the proposed model significantly outperforms existing strong baselines. We further demonstrate that the deductive procedure not only presents more explainable steps but also enables us to make more accurate predictions on questions that require more complex reasoning.

Few-shot relation extraction aims to learn to identify the relation between two entities based on very limited training examples. Recent efforts found that textual labels (i.e., relation names and relation descriptions) could be extremely useful for learning class representations, which will benefit the few-shot learning task. However, what is the best way to leverage such label information in the learning process is an important research question. Existing works largely assume such textual labels are always present during both learning and prediction. In this work, we argue that such approaches may not always lead to optimal results. Instead, we present a novel approach called label prompt dropout, which randomly removes label descriptions in the learning process. Our experiments show that our approach is able to lead to improved class representations, yielding significantly better results on the few-shot relation extraction task.

Fine-tuning a pre-trained language model via the contrastive learning framework with a large amount of unlabeled sentences or labeled sentence pairs is a common way to obtain high-quality sentence representations. Although the contrastive learning framework has shown its superiority on sentence representation learning over previous methods, the potential of such a framework is under-explored so far due to the simple method it used to construct positive pairs. Motivated by this, we propose a method that makes hard positives from the original training examples. A pivotal ingredient of our approach is the use of prefix that attached to a pre-trained language model, which allows for differentiable data augmentation during contrastive learning. Our method can be summarized in two steps: supervised prefix-tuning followed by joint contrastive fine-tuning with unlabeled or labeled examples. Our experiments confirm the effectiveness of our data augmentation approach. The proposed method yields significant improvements over existing methods under both semi-supervised and supervised settings. Our experiments under a low labeled data setting also show that our method is more label-efficient than the state-of-the-art contrastive learning methods.

Training a good deep learning model requires substantial data and computing resources, which makes the resulting neural model a valuable intellectual property. To prevent the neural network from being undesirably exploited, non-transferable learning has been proposed to reduce the model generalization ability in specific target domains. However, existing approaches require labeled data for the target domain which can be difficult to obtain. Furthermore, they do not have the mechanism to still recover the model’s ability to access the target domain.In this paper, we propose a novel unsupervised non-transferable learning method for the text classification task that does not require annotated target domain data. We further introduce a secret key component in our approach for recovering the access to the target domain, where we design both an explicit and an implicit method for doing so. Extensive experiments demonstrate the effectiveness of our approach.

Using prompts to explore the knowledge contained within pre-trained language models for downstream tasks has now become an active topic. Current prompt tuning methods mostly convert the downstream tasks to masked language modeling problems by adding cloze-style phrases and mapping all labels to verbalizations with fixed length, which has proven effective for tasks with simple label spaces. However, when applied to relation classification exhibiting complex label spaces, vanilla prompt tuning methods may struggle with label verbalizations with arbitrary lengths due to rigid prompt restrictions. Inspired by the text infilling task for pre-training generative models that can flexibly predict missing spans, we propose a novel generative prompt tuning method to reformulate relation classification as an infilling problem, which frees our approach from limitations of current prompt based approaches and thus fully exploits rich semantics of entity and relation types. In addition, we design entity-guided decoding and discriminative relation scoring to generate and align relations effectively and efficiently during inference. Extensive experiments under fully supervised settings and low-resource settings demonstrate the effectiveness of our approach.

Multi-modal named entity recognition (NER) and relation extraction (RE) aim to leverage relevant image information to improve the performance of NER and RE. Most existing efforts largely focused on directly extracting potentially useful information from images (such as pixel-level features, identified objects, and associated captions).However, such extraction processes may not be knowledge aware, resulting in information that may not be highly relevant.In this paper, we propose a novel Multi-modal Retrieval based framework (MoRe).MoRe contains a text retrieval module and an image-based retrieval module, which retrieve related knowledge of the input text and image in the knowledge corpus respectively.Next, the retrieval results are sent to the textual and visual models respectively for predictions.Finally, a Mixture of Experts (MoE) module combines the predictions from the two models to make the final decision.Our experiments show that both our textual model and visual model can achieve state-of-the-art performance on four multi-modal NER datasets and one multi-modal RE dataset.With MoE, the model performance can be further improved and our analysis demonstrates the benefits of integrating both textual and visual cues for such tasks.

The MultiCoNER shared task aims at detecting semantically ambiguous and complex named entities in short and low-context settings for multiple languages. The lack of contexts makes the recognition of ambiguous named entities challenging. To alleviate this issue, our team DAMO-NLP proposes a knowledge-based system, where we build a multilingual knowledge base based on Wikipedia to provide related context information to the named entity recognition (NER) model. Given an input sentence, our system effectively retrieves related contexts from the knowledge base. The original input sentences are then augmented with such context information, allowing significantly better contextualized token representations to be captured. Our system wins 10 out of 13 tracks in the MultiCoNER shared task.

It has been shown that named entity recognition (NER) could benefit from incorporating the long-distance structured information captured by dependency trees. We believe this is because both types of features - the contextual information captured by the linear sequences and the structured information captured by the dependency trees may complement each other. However, existing approaches largely focused on stacking the LSTM and graph neural networks such as graph convolutional networks (GCNs) for building improved NER models, where the exact interaction mechanism between the two types of features is not very clear, and the performance gain does not appear to be significant. In this work, we propose a simple and robust solution to incorporate both types of features with our Synergized-LSTM (Syn-LSTM), which clearly captures how the two types of features interact. We conduct extensive experiments on several standard datasets across four languages. The results demonstrate that the proposed model achieves better performance than previous approaches while requiring fewer parameters. Our further analysis demonstrates that our model can capture longer dependencies compared with strong baselines.

Few-shot relation extraction (FSRE) focuses on recognizing novel relations by learning with merely a handful of annotated instances. Meta-learning has been widely adopted for such a task, which trains on randomly generated few-shot tasks to learn generic data representations. Despite impressive results achieved, existing models still perform suboptimally when handling hard FSRE tasks, where the relations are fine-grained and similar to each other. We argue this is largely because existing models do not distinguish hard tasks from easy ones in the learning process. In this paper, we introduce a novel approach based on contrastive learning that learns better representations by exploiting relation label information. We further design a method that allows the model to adaptively learn how to focus on hard tasks. Experiments on two standard datasets demonstrate the effectiveness of our method.

Dependency parse trees are helpful for discovering the opinion words in aspect-based sentiment analysis (ABSA) (CITATION). However, the trees obtained from off-the-shelf dependency parsers are static, and could be sub-optimal in ABSA. This is because the syntactic trees are not designed for capturing the interactions between opinion words and aspect words. In this work, we aim to shorten the distance between aspects and corresponding opinion words by learning an aspect-centric tree structure. The aspect and opinion words are expected to be closer along such tree structure compared to the standard dependency parse tree. The learning process allows the tree structure to adaptively correlate the aspect and opinion words, enabling us to better identify the polarity in the ABSA task. We conduct experiments on five aspect-based sentiment datasets, and the proposed model significantly outperforms recent strong baselines. Furthermore, our thorough analysis demonstrates the average distance between aspect and opinion words are shortened by at least 19% on the standard SemEval Restaurant14 (CITATION) dataset.

Information Extraction (IE) aims to extract structural information from unstructured texts. In practice, long-tailed distributions caused by the selection bias of a dataset may lead to incorrect correlations, also known as spurious correlations, between entities and labels in the conventional likelihood models. This motivates us to propose counterfactual IE (CFIE), a novel framework that aims to uncover the main causalities behind data in the view of causal inference. Specifically, 1) we first introduce a unified structural causal model (SCM) for various IE tasks, describing the relationships among variables; 2) with our SCM, we then generate counterfactuals based on an explicit language structure to better calculate the direct causal effect during the inference stage; 3) we further propose a novel debiasing approach to yield more robust predictions. Experiments on three IE tasks across five public datasets show the effectiveness of our CFIE model in mitigating the spurious correlation issues.

Chatbot is increasingly thriving in different domains, however, because of unexpected discourse complexity and training data sparseness, its potential distrust hatches vital apprehension. Recently, Machine-Human Chatting Handoff (MHCH), predicting chatbot failure and enabling human-algorithm collaboration to enhance chatbot quality, has attracted increasing attention from industry and academia. In this study, we propose a novel model, Role-Selected Sharing Network (RSSN), which integrates both dialogue satisfaction estimation and handoff prediction in one multi-task learning framework. Unlike prior efforts in dialog mining, by utilizing local user satisfaction as a bridge, global satisfaction detector and handoff predictor can effectively exchange critical information. Specifically, we decouple the relation and interaction between the two tasks by the role information after the shared encoder. Extensive experiments on two public datasets demonstrate the effectiveness of our model.

Entities can be found in various text genres, ranging from tweets and web pages to user queries submitted to web search engines. Existing research either considers all entities in the text equally important, or heuristics are used to measure their salience. We believe that a key reason for the relatively limited work on entity salience is the lack of appropriate datasets. To support research on entity salience, we present a new dataset, the WikiNews Salience dataset (WN-Salience), which can be used to benchmark tasks such as entity salience detection and salient entity linking. WN-Salience is built on top of Wikinews, a Wikimedia project whose mission is to present reliable news articles. Entities in Wikinews articles are identified by the authors of the articles and are linked to Wikinews categories when they are salient or to Wikipedia pages otherwise. The dataset is built automatically, and consists of approximately 7,000 news articles, and 90,000 in-text entity annotations. We compare the WN-Salience dataset against existing datasets on the task and analyze their differences. Furthermore, we conduct experiments on entity salience detection; the results demonstrate that WN-Salience is a challenging testbed that is complementary to existing ones.

Document-level relation extraction requires integrating information within and across multiple sentences of a document and capturing complex interactions between inter-sentence entities. However, effective aggregation of relevant information in the document remains a challenging research question. Existing approaches construct static document-level graphs based on syntactic trees, co-references or heuristics from the unstructured text to model the dependencies. Unlike previous methods that may not be able to capture rich non-local interactions for inference, we propose a novel model that empowers the relational reasoning across sentences by automatically inducing the latent document-level graph. We further develop a refinement strategy, which enables the model to incrementally aggregate relevant information for multi-hop reasoning. Specifically, our model achieves an F1 score of 59.05 on a large-scale document-level dataset (DocRED), significantly improving over the previous results, and also yields new state-of-the-art results on the CDR and GDA dataset. Furthermore, extensive analyses show that the model is able to discover more accurate inter-sentence relations.

Attention has been proven successful in many natural language processing (NLP) tasks. Recently, many researchers started to investigate the interpretability of attention on NLP tasks. Many existing approaches focused on examining whether the local attention weights could reflect the importance of input representations. In this work, we present a study on understanding the internal mechanism of attention by looking into the gradient update process, checking its behavior when approaching a local minimum during training. We propose to analyze for each word token the following two quantities: its polarity score and its attention score, where the latter is a global assessment on the token’s significance. We discuss conditions under which the attention mechanism may become more (or less) interpretable, and show how the interplay between the two quantities can contribute towards model performance.

We introduce SmartCiteCon (SCC), a Java API for extracting both explicit and implicit citation context from academic literature in English. The tool is built on a Support Vector Machine (SVM) model trained on a set of 7,058 manually annotated citation context sentences, curated from 34,000 papers from the ACL Anthology. The model with 19 features achieves F1=85.6%. SCC supports PDF, XML, and JSON files out-of-box, provided that they are conformed to certain schemas. The API supports single document processing and batch processing in parallel. It takes about 12–45 seconds on average depending on the format to process a document on a dedicated server with 6 multithreaded cores. Using SCC, we extracted 11.8 million citation context sentences from ~33.3k PMC papers in the CORD-19 dataset, released on June 13, 2020. We will provide continuous supplementary data contribution to the CORD-19 and other datasets. The source code is released at https://gitee.com/irlab/SmartCiteCon.

Previous works on knowledge-to-text generation take as input a few RDF triples or key-value pairs conveying the knowledge of some entities to generate a natural language description. Existing datasets, such as WIKIBIO, WebNLG, and E2E, basically have a good alignment between an input triple/pair set and its output text. However, in practice, the input knowledge could be more than enough, since the output description may only cover the most significant knowledge. In this paper, we introduce a large-scale and challenging dataset to facilitate the study of such a practical scenario in KG-to-text. Our dataset involves retrieving abundant knowledge of various types of main entities from a large knowledge graph (KG), which makes the current graph-to-sequence models severely suffer from the problems of information loss and parameter explosion while generating the descriptions. We address these challenges by proposing a multi-graph structure that is able to represent the original graph information more comprehensively. Furthermore, we also incorporate aggregation methods that learn to extract the rich graph information. Extensive experiments demonstrate the effectiveness of our model architecture.

Named entity recognition and relation extraction are two important fundamental problems. Joint learning algorithms have been proposed to solve both tasks simultaneously, and many of them cast the joint task as a table-filling problem. However, they typically focused on learning a single encoder (usually learning representation in the form of a table) to capture information required for both tasks within the same space. We argue that it can be beneficial to design two distinct encoders to capture such two different types of information in the learning process. In this work, we propose the novel table-sequence encoders where two different encoders – a table encoder and a sequence encoder are designed to help each other in the representation learning process. Our experiments confirm the advantages of having two encoders over one encoder. On several standard datasets, our model shows significant improvements over existing approaches.

AMR-to-text generation is used to transduce Abstract Meaning Representation structures (AMR) into text. A key challenge in this task is to efficiently learn effective graph representations. Previously, Graph Convolution Networks (GCNs) were used to encode input AMRs, however, vanilla GCNs are not able to capture non-local information and additionally, they follow a local (first-order) information aggregation scheme. To account for these issues, larger and deeper GCN models are required to capture more complex interactions. In this paper, we introduce a dynamic fusion mechanism, proposing Lightweight Dynamic Graph Convolutional Networks (LDGCNs) that capture richer non-local interactions by synthesizing higher order information from the input graphs. We further develop two novel parameter saving strategies based on the group graph convolutions and weight tied convolutions to reduce memory usage and model complexity. With the help of these strategies, we are able to train a model with fewer parameters while maintaining the model capacity. Experiments demonstrate that LDGCNs outperform state-of-the-art models on two benchmark datasets for AMR-to-text generation with significantly fewer parameters.

Aspect Sentiment Triplet Extraction (ASTE) is the task of extracting the triplets of target entities, their associated sentiment, and opinion spans explaining the reason for the sentiment. Existing research efforts mostly solve this problem using pipeline approaches, which break the triplet extraction process into several stages. Our observation is that the three elements within a triplet are highly related to each other, and this motivates us to build a joint model to extract such triplets using a sequence tagging approach. However, how to effectively design a tagging approach to extract the triplets that can capture the rich interactions among the elements is a challenging research question. In this work, we propose the first end-to-end model with a novel position-aware tagging scheme that is capable of jointly extracting the triplets. Our experimental results on several existing datasets show that jointly capturing elements in the triplet using our approach leads to improved performance over the existing approaches. We also conducted extensive experiments to investigate the model effectiveness and robustness.

Aspect based sentiment analysis, predicting sentiment polarity of given aspects, has drawn extensive attention. Previous attention-based models emphasize using aspect semantics to help extract opinion features for classification. However, these works are either not able to capture opinion spans as a whole, or not able to capture variable-length opinion spans. In this paper, we present a neat and effective structured attention model by aggregating multiple linear-chain CRFs. Such a design allows the model to extract aspect-specific opinion spans and then evaluate sentiment polarity by exploiting the extracted opinion features. The experimental results on four datasets demonstrate the effectiveness of the proposed model, and our analysis demonstrates that our model can capture aspect-specific opinion spans.

Abstractive document summarization is usually modeled as a sequence-to-sequence (SEQ2SEQ) learning problem. Unfortunately, training large SEQ2SEQ based summarization models on limited supervised summarization data is challenging. This paper presents three sequence-to-sequence pre-training (in shorthand, STEP) objectives which allow us to pre-train a SEQ2SEQ based abstractive summarization model on unlabeled text. The main idea is that, given an input text artificially constructed from a document, a model is pre-trained to reinstate the original document. These objectives include sentence reordering, next sentence generation and masked document generation, which have close relations with the abstractive document summarization task. Experiments on two benchmark summarization datasets (i.e., CNN/DailyMail and New York Times) show that all three objectives can improve performance upon baselines. Compared to models pre-trained on large-scale data (larger than 160GB), our method, with only 19GB text for pre-training, achieves comparable results, which demonstrates its effectiveness.

In existing sophisticated text-to-SQL models, schema linking is often considered as a simple, minor component, belying its importance. By providing a schema linking corpus based on the Spider text-to-SQL dataset, we systematically study the role of schema linking. We also build a simple BERT-based baseline, called Schema-Linking SQL (SLSQL) to perform a data-driven study. We find when schema linking is done well, SLSQL demonstrates good performance on Spider despite its structural simplicity. Many remaining errors are attributable to corpus noise. This suggests schema linking is the crux for the current text-to-SQL task. Our analytic studies provide insights on the characteristics of schema linking for future developments of text-to-SQL tasks.

Peer review and rebuttal, with rich interactions and argumentative discussions in between, are naturally a good resource to mine arguments. However, few works study both of them simultaneously. In this paper, we introduce a new argument pair extraction (APE) task on peer review and rebuttal in order to study the contents, the structure and the connections between them. We prepare a challenging dataset that contains 4,764 fully annotated review-rebuttal passage pairs from an open review platform to facilitate the study of this task. To automatically detect argumentative propositions and extract argument pairs from this corpus, we cast it as the combination of a sequence labeling task and a text relation classification task. Thus, we propose a multitask learning framework based on hierarchical LSTM networks. Extensive experiments and analysis demonstrate the effectiveness of our multi-task framework, and also show the challenges of the new task as well as motivate future research directions.

Dependency trees convey rich structural information that is proven useful for extracting relations among entities in text. However, how to effectively make use of relevant information while ignoring irrelevant information from the dependency trees remains a challenging research question. Existing approaches employing rule based hard-pruning strategies for selecting relevant partial dependency structures may not always yield optimal results. In this work, we propose Attention Guided Graph Convolutional Networks (AGGCNs), a novel model which directly takes full dependency trees as inputs. Our model can be understood as a soft-pruning approach that automatically learns how to selectively attend to the relevant sub-structures useful for the relation extraction task. Extensive results on various tasks including cross-sentence n-ary relation extraction and large-scale sentence-level relation extraction show that our model is able to better leverage the structural information of the full dependency trees, giving significantly better results than previous approaches.

Gazetteers were shown to be useful resources for named entity recognition (NER). Many existing approaches to incorporating gazetteers into machine learning based NER systems rely on manually defined selection strategies or handcrafted templates, which may not always lead to optimal effectiveness, especially when multiple gazetteers are involved. This is especially the case for the task of Chinese NER, where the words are not naturally tokenized, leading to additional ambiguities. To automatically learn how to incorporate multiple gazetteers into an NER system, we propose a novel approach based on graph neural networks with a multi-digraph structure that captures the information that the gazetteers offer. Experiments on various datasets show that our model is effective in incorporating rich gazetteer information while resolving ambiguities, outperforming previous approaches.

In this paper, we investigate the importance of social network information compared to content information in the prediction of a Twitter user’s occupational class. We show that the content information of a user’s tweets, the profile descriptions of a user’s follower/following community, and the user’s social network provide useful information for classifying a user’s occupational group. In our study, we extend an existing data set for this problem, and we achieve significantly better performance by using social network homophily that has not been fully exploited in previous work. In our analysis, we found that by using the graph convolutional network to exploit social homophily, we can achieve competitive performance on this data set with just a small fraction of the training data.

An arithmetic word problem typically includes a textual description containing several constant quantities. The key to solving the problem is to reveal the underlying mathematical relations (such as addition and subtraction) among quantities, and then generate equations to find solutions. This work presents a novel approach, Quantity Tagger, that automatically discovers such hidden relations by tagging each quantity with a sign corresponding to one type of mathematical operation. For each quantity, we assume there exists a latent, variable-sized quantity span surrounding the quantity token in the text, which conveys information useful for determining its sign. Empirical results show that our method achieves 5 and 8 points of accuracy gains on two datasets respectively, compared to prior approaches.

We focus on graph-to-sequence learning, which can be framed as transducing graph structures to sequences for text generation. To capture structural information associated with graphs, we investigate the problem of encoding graphs using graph convolutional networks (GCNs). Unlike various existing approaches where shallow architectures were used for capturing local structural information only, we introduce a dense connection strategy, proposing a novel Densely Connected Graph Convolutional Network (DCGCN). Such a deep architecture is able to integrate both local and non-local features to learn a better structural representation of a graph. Our model outperforms the state-of-the-art neural models significantly on AMR-to-text generation and syntax-based neural machine translation.

Supervised approaches to named entity recognition (NER) are largely developed based on the assumption that the training data is fully annotated with named entity information. However, in practice, annotated data can often be imperfect with one typical issue being the training data may contain incomplete annotations. We highlight several pitfalls associated with learning under such a setup in the context of NER and identify limitations associated with existing approaches, proposing a novel yet easy-to-implement approach for recognizing named entities with incomplete data annotations. We demonstrate the effectiveness of our approach through extensive experiments.

A pun is a form of wordplay for an intended humorous or rhetorical effect, where a word suggests two or more meanings by exploiting polysemy (homographic pun) or phonological similarity to another word (heterographic pun). This paper presents an approach that addresses pun detection and pun location jointly from a sequence labeling perspective. We employ a new tagging scheme such that the model is capable of performing such a joint task, where useful structural information can be properly captured. We show that our proposed model is effective in handling both homographic and heterographic puns. Empirical results on the benchmark datasets demonstrate that our approach can achieve new state-of-the-art results.

This paper introduces a new task – Chinese address parsing – the task of mapping Chinese addresses into semantically meaningful chunks. While it is possible to model this problem using a conventional sequence labelling approach, our observation is that there exist complex dependencies between labels that cannot be readily captured by a simple linear-chain structure. We investigate neural structured prediction models with latent variables to capture such rich structural information within Chinese addresses. We create and publicly release a new dataset consisting of 15K Chinese addresses, and conduct extensive experiments on the dataset to investigate the model effectiveness and robustness. We release our code and data at http://statnlp.org/research/sp.

Dependency tree structures capture long-distance and syntactic relationships between words in a sentence. The syntactic relations (e.g., nominal subject, object) can potentially infer the existence of certain named entities. In addition, the performance of a named entity recognizer could benefit from the long-distance dependencies between the words in dependency trees. In this work, we propose a simple yet effective dependency-guided LSTM-CRF model to encode the complete dependency trees and capture the above properties for the task of named entity recognition (NER). The data statistics show strong correlations between the entity types and dependency relations. We conduct extensive experiments on several standard datasets and demonstrate the effectiveness of the proposed model in improving NER and achieving state-of-the-art performance. Our analysis reveals that the significant improvements mainly result from the dependency relations and long-distance interactions provided by dependency trees.

Multilingual knowledge graphs (KGs), such as YAGO and DBpedia, represent entities in different languages. The task of cross-lingual entity alignment is to match entities in a source language with their counterparts in target languages. In this work, we investigate embedding-based approaches to encode entities from multilingual KGs into the same vector space, where equivalent entities are close to each other. Specifically, we apply graph convolutional networks (GCNs) to combine multi-aspect information of entities, including topological connections, relations, and attributes of entities, to learn entity embeddings. To exploit the literal descriptions of entities expressed in different languages, we propose two uses of a pretrained multilingual BERT model to bridge cross-lingual gaps. We further propose two strategies to integrate GCN-based and BERT-based modules to boost performance. Extensive experiments on two benchmark datasets demonstrate that our method significantly outperforms existing systems.

We propose Text2Math, a model for semantically parsing text into math expressions. The model can be used to solve different math related problems including arithmetic word problems and equation parsing problems. Unlike previous approaches, we tackle the problem from an end-to-end structured prediction perspective where our algorithm aims to predict the complete math expression at once as a tree structure, where minimal manual efforts are involved in the process. Empirical results on benchmark datasets demonstrate the efficacy of our approach.

Targeted sentiment analysis is the task of jointly predicting target entities and their associated sentiment information. Existing research efforts mostly regard this joint task as a sequence labeling problem, building models that can capture explicit structures in the output space. However, the importance of capturing implicit global structural information that resides in the input space is largely unexplored. In this work, we argue that both types of information (implicit and explicit structural information) are crucial for building a successful targeted sentiment analysis model. Our experimental results show that properly capturing both information is able to lead to better performance than competitive existing approaches. We also conduct extensive experiments to investigate our model’s effectiveness and robustness.

In medical documents, it is possible that an entity of interest not only contains a discontiguous sequence of words but also overlaps with another entity. Entities of such structures are intrinsically hard to recognize due to the large space of possible entity combinations. In this work, we propose a neural two-stage approach to recognizing discontiguous and overlapping entities by decomposing this problem into two subtasks: 1) it first detects all the overlapping spans that either form entities on their own or present as segments of discontiguous entities, based on the representation of segmental hypergraph, 2) next it learns to combine these segments into discontiguous entities with a classifier, which filters out other incorrect combinations of segments. Two neural components are designed for these subtasks respectively and they are learned jointly using a shared encoder for text. Our model achieves the state-of-the-art performance in a standard dataset, even in the absence of external features that previous methods used.

Existing works, including ELMO and BERT, have revealed the importance of pre-training for NLP tasks. While there does not exist a single pre-training model that works best in all cases, it is of necessity to develop a framework that is able to deploy various pre-training models efficiently. For this purpose, we propose an assemble-on-demand pre-training toolkit, namely Universal Encoder Representations (UER). UER is loosely coupled, and encapsulated with rich modules. By assembling modules on demand, users can either reproduce a state-of-the-art pre-training model or develop a pre-training model that remains unexplored. With UER, we have built a model zoo, which contains pre-trained models based on different corpora, encoders, and targets (objectives). With proper pre-trained models, we could achieve new state-of-the-art results on a range of downstream datasets.

We report an empirical study on the task of negation scope extraction given the negation cue. Our key observation is that certain useful information such as features related to negation cue, long-distance dependencies as well as some latent structural information can be exploited for such a task. We design approaches based on conditional random fields (CRF), semi-Markov CRF, as well as latent-variable CRF models to capture such information. Extensive experiments on several standard datasets demonstrate that our approaches are able to achieve better results than existing approaches reported in the literature.

With the development of several multilingual datasets used for semantic parsing, recent research efforts have looked into the problem of learning semantic parsers in a multilingual setup. However, how to improve the performance of a monolingual semantic parser for a specific language by leveraging data annotated in different languages remains a research question that is under-explored. In this work, we present a study to show how learning distributed representations of the logical forms from data annotated in different languages can be used for improving the performance of a monolingual semantic parser. We extend two existing monolingual semantic parsers to incorporate such cross-lingual distributed logical representations as features. Experiments show that our proposed approach is able to yield improved semantic parsing results on the standard multilingual GeoQuery dataset.

In this work, we propose a novel segmental hypergraph representation to model overlapping entity mentions that are prevalent in many practical datasets. We show that our model built on top of such a new representation is able to capture features and interactions that cannot be captured by previous models while maintaining a low time complexity for inference. We also present a theoretical analysis to formally assess how our representation is better than alternative representations reported in the literature in terms of representational power. Coupled with neural networks for feature learning, our model achieves the state-of-the-art performance in three benchmark datasets annotated with overlapping mentions.

It is common that entity mentions can contain other mentions recursively. This paper introduces a scalable transition-based method to model the nested structure of mentions. We first map a sentence with nested mentions to a designated forest where each mention corresponds to a constituent of the forest. Our shift-reduce based system then learns to construct the forest structure in a bottom-up manner through an action sequence whose maximal length is guaranteed to be three times of the sentence length. Based on Stack-LSTM which is employed to efficiently and effectively represent the states of the system in a continuous space, our system is further incorporated with a character-based component to capture letter-level patterns. Our model gets the state-of-the-art performances in ACE datasets, showing its effectiveness in detecting nested mentions.

This paper introduces a simple yet effective transition-based system for Abstract Meaning Representation (AMR) parsing. We argue that a well-defined search space involved in a transition system is crucial for building an effective parser. We propose to conduct the search in a refined search space based on a new compact AMR graph and an improved oracle. Our end-to-end parser achieves the state-of-the-art performance on various datasets with minimal additional information.

Recent research efforts have shown that neural architectures can be effective in conventional information extraction tasks such as named entity recognition, yielding state-of-the-art results on standard newswire datasets. However, despite significant resources required for training such models, the performance of a model trained on one domain typically degrades dramatically when applied to a different domain, yet extracting entities from new emerging domains such as social media can be of significant interest. In this paper, we empirically investigate effective methods for conveniently adapting an existing, well-trained neural NER model for a new domain. Unlike existing approaches, we propose lightweight yet effective methods for performing domain adaptation for neural models. Specifically, we introduce adaptation layers on top of existing neural architectures, where no re-training using the source domain data is required. We conduct extensive empirical studies and show that our approach significantly outperforms state-of-the-art methods.

We propose a novel dependency-based hybrid tree model for semantic parsing, which converts natural language utterance into machine interpretable meaning representations. Unlike previous state-of-the-art models, the semantic information is interpreted as the latent dependency between the natural language words in our joint representation. Such dependency information can capture the interactions between the semantics and natural language words. We integrate a neural component into our model and propose an efficient dynamic-programming algorithm to perform tractable inference. Through extensive experiments on the standard multilingual GeoQuery dataset with eight languages, we demonstrate that our proposed approach is able to achieve state-of-the-art performance across several languages. Analysis also justifies the effectiveness of using our new dependency-based representation.

This paper describes the SemEval 2018 shared task on semantic extraction from cybersecurity reports, which is introduced for the first time as a shared task on SemEval. This task comprises four SubTasks done incrementally to predict the characteristics of a specific malware using cybersecurity reports. To the best of our knowledge, we introduce the world’s largest publicly available dataset of annotated malware reports in this task. This task received in total 18 submissions from 9 participating teams.

In this paper, we propose a new model that is capable of recognizing overlapping mentions. We introduce a novel notion of mention separators that can be effectively used to capture how mentions overlap with one another. On top of a novel multigraph representation that we introduce, we show that efficient and exact inference can still be performed. We present some theoretical analysis on the differences between our model and a recently proposed model for recognizing overlapping mentions, and discuss the possible implications of the differences. Through extensive empirical analysis on standard datasets, we demonstrate the effectiveness of our approach.

Learning word embeddings has received a significant amount of attention recently. Often, word embeddings are learned in an unsupervised manner from a large collection of text. The genre of the text typically plays an important role in the effectiveness of the resulting embeddings. How to effectively train word embedding models using data from different domains remains a problem that is less explored. In this paper, we present a simple yet effective method for learning word embeddings based on text from different domains. We demonstrate the effectiveness of our approach through extensive experiments on various down-stream NLP tasks.

Structured prediction is one of the most important topics in various fields, including machine learning, computer vision, natural language processing (NLP) and bioinformatics. In this tutorial, we present a novel framework that unifies various structured prediction models.The hidden Markov model (HMM) and the probabilistic context-free grammars (PCFGs) are two classic generative models used for predicting outputs with linear-chain and tree structures, respectively. As HMM’s discriminative counterpart, the linear-chain conditional random fields (CRFs) (Lafferty et al., 2001) model was later proposed. Such a model was shown to yield good performance on standard NLP tasks such as information extraction. Several extensions to such a model were then proposed afterward, including the semi-Markov CRFs (Sarawagi and Cohen, 2004), tree CRFs (Cohn and Blunsom, 2005), as well as discriminative parsing models and their latent variable variants (Petrov and Klein, 2007). On the other hand, utilizing a slightly different loss function, one could arrive at the structured support vector machines (Tsochantaridis et al., 2004) and its latent variable variant (Yu and Joachims, 2009) as well. Furthermore, new models that integrate neural networks and graphical models, such as neural CRFs (Do et al., 2010) were also proposed.In this tutorial, we will be discussing how such a wide spectrum of existing structured prediction models can all be implemented under a unified framework (available at here) that involves some basic building blocks. Based on such a framework, we show how some seemingly complicated structured prediction models such as a semantic parsing model (Lu et al., 2008; Lu, 2014) can be implemented conveniently and quickly. Furthermore, we also show that the framework can be used to solve certain structured prediction problems that otherwise cannot be easily handled by conventional structured prediction models. Specifically, we show how to use such a framework to construct models that are capable of predicting non-conventional structures, such as overlapping structures (Lu and Roth, 2015; Muis and Lu, 2016a). We will also discuss how to make use of the framework to build other related models such as topic models and highlight its potential applications in some recent popular tasks (e.g., AMR parsing (Flanigan et al., 2014)).The framework has been extensively used by our research group for developing various structured prediction models, including models for information extraction (Lu and Roth, 2015; Muis and Lu, 2016a; Jie et al., 2017), noun phrase chunking (Muis and Lu, 2016b), semantic parsing (Lu, 2015; Susanto and Lu, 2017), and sentiment analysis (Li and Lu, 2017). It is our hope that this tutorial will be helpful for many natural language processing researchers who are interested in designing their own structured prediction models rapidly. We also hope this tutorial allows researchers to strengthen their understandings on the connections between various structured prediction models, and that the open release of the framework will bring value to the NLP research community and enhance its overall productivity.The material associated with this tutorial will be available at the tutorial web site: https://web.archive.org/web/20180427113151/http://statnlp.org/tutorials/.

Cybersecurity risks and malware threats are becoming increasingly dangerous and common. Despite the severity of the problem, there has been few NLP efforts focused on tackling cybersecurity. In this paper, we discuss the construction of a new database for annotated malware texts. An annotation framework is introduced based on the MAEC vocabulary for defining malware characteristics, along with a database consisting of 39 annotated APT reports with a total of 6,819 sentences. We also use the database to construct models that can potentially help cybersecurity researchers in their data collection and analytics efforts.

This paper presents an LDA-based model that generates topically coherent segments within documents by jointly segmenting documents and assigning topics to their words. The coherence between topics is ensured through a copula, binding the topics associated to the words of a segment. In addition, this model relies on both document and segment specific topic distributions so as to capture fine grained differences in topic assignments. We show that the proposed model naturally encompasses other state-of-the-art LDA-based models designed for similar tasks. Furthermore, our experiments, conducted on six different publicly available datasets, show the effectiveness of our model in terms of perplexity, Normalized Pointwise Mutual Information, which captures the coherence between the generated topics, and the Micro F1 measure for text classification.

In this paper, we address semantic parsing in a multilingual context. We train one multilingual model that is capable of parsing natural language sentences from multiple different languages into their corresponding formal semantic representations. We extend an existing sequence-to-tree model to a multi-task learning framework which shares the decoder for generating semantic representations. We report evaluation results on the multilingual GeoQuery corpus and introduce a new multilingual version of the ATIS corpus.

We describe the system developed by the team of the National University of Singapore for the Chinese-English BTEC task of the IWSLT 2009 evaluation campaign. We adopted a state-of-the-art phrase-based statistical machine translation approach and focused on experiments with different Chinese word segmentation standards. In our official submission, we trained a separate system for each segmenter and we combined the outputs in a subsequent re-ranking step. Given the small size of the training data, we further re-trained the system on the development data after tuning. The evaluation results show that both strategies yield sizeable and consistent improvements in translation quality.