This paper describes our submission for SemEval-2024 Task 6: SHROOM, a Shared-task on Hallucinations and Related Observable Overgeneration Mistakes. We propose four groups of methods for hallucination detection: 1) Entailment Recognition; 2) Similarity Search; 3) Factuality Verification; 4) Confidence Estimation. The four methods rely on either the semantic relationship between the hypothesis and its source (target) or on the model-aware features during decoding. We participated in both the model-agnostic and model-aware tracks. Our method’s effectiveness is validated by our high rankings 3rd in the model-agnostic track and 5th in the model-aware track. We have released our code on GitHub.

We consider an unanswered question in the discourse processing community: why do relation classifiers trained on explicit examples (with connectives removed) perform poorly in real implicit scenarios? Prior work claimed this is due to linguistic dissimilarity between explicit and implicit examples but provided no empirical evidence. In this study, we show that one cause for such failure is a label shift after connectives are eliminated. Specifically, we find that the discourse relations expressed by some explicit instances will change when connectives disappear. Unlike previous work manually analyzing a few examples, we present empirical evidence at the corpus level to prove the existence of such shift. Then, we analyze why label shift occurs by considering factors such as the syntactic role played by connectives, ambiguity of connectives, and more. Finally, we investigate two strategies to mitigate the label shift: filtering out noisy data and joint learning with connectives. Experiments on PDTB 2.0, PDTB 3.0, and the GUM dataset demonstrate that classifiers trained with our strategies outperform strong baselines.

In the era of code large language models (code LLMs), data engineering plays a pivotal role during the instruction fine-tuning phase. To train a versatile model, previous efforts devote tremendous efforts into crafting instruction data covering all the downstream scenarios. Nonetheless, this will incur significant expenses in constructing data and training model. Therefore, this paper introduces CodeM, a novel data construction strategy, which can efficiently train a versatile model using less data via our newly proposed ability matrix. CodeM uses ability matrix to decouple code LLMs’ abilities into two dimensions, constructing a lightweight training corpus that only covers a subset of target scenarios. Extensive experiments on HumanEvalPack and MultiPL-E imply that code LLMs can combine the single-dimensional abilities to master composed abilities, validating the effectiveness of CodeM.

Structured pruning fundamentally reduces computational and memory overheads of large language models (LLMs) and offers a feasible solution for end-side LLM deployment. Structurally pruned models remain dense and high-precision, highly compatible with further tuning and compression. However, as the coarse-grained structured pruning poses large damage to the highly interconnected model, achieving a high compression ratio for scaled-up LLMs remains a challenge. In this paper, we introduce a task-agnostic structured pruning approach coupled with a compact Transformer architecture design. The proposed approach, named TransAct, reduces transitional activations inside multi-head attention (MHA) and multi-layer perceptron (MLP) modules, while preserving the inter-module activations that are sensitive to perturbations. Hence, the LLM is pruned into an intra-module low-rank architecture, significantly reducing weights, KV Cache and attention computation. TransAct is implemented on the LLaMA model and evaluated on downstream benchmarks. Results verify the optimality of our approach at high compression with respect to both efficiency and performance. Further, ablation studies reveal the strength of activation-guided iterative pruning and provide experimental analysis on the redundancy of MHA and MLP modules.

Increasing the number of parameters in large language models (LLMs) usually improves performance in downstream tasks but raises compute and memory costs, making deployment difficult in resource-limited settings. Quantization techniques, which reduce the bits needed for model weights or activations with minimal performance loss, have become popular due to the rise of LLMs. However, most quantization studies use pre-trained LLMs, and the impact of quantization on instruction-tuned LLMs and the relationship between perplexity and benchmark performance of quantized LLMs are not well understood. Evaluation of quantized LLMs is often limited to language modeling and a few classification tasks, leaving their performance on other benchmarks unclear. To address these gaps, we propose a structured evaluation framework consisting of three critical dimensions: (1) knowledge & capacity, (2) alignment, and (3) efficiency, and conduct extensive experiments across ten diverse benchmarks. Our experimental results indicate that LLMs with 4-bit quantization can retain performance comparable to their non-quantized counterparts, and perplexity can serve as a proxy metric for quantized LLMs on most benchmarks. Furthermore, quantized LLMs with larger parameter scales can outperform smaller LLMs. Despite the memory savings achieved through quantization, it can also slow down the inference speed of LLMs. Consequently, substantial engineering efforts and hardware support are imperative to achieve a balanced optimization of decoding speed and memory consumption in the context of quantized LLMs.

Maintenance short texts (MST), derived from maintenance work order records, encapsulate crucial information in a concise yet information-rich format. These user-generated technical texts provide critical insights into the state and maintenance activities of machines, infrastructure, and other engineered assets–pillars of the modern economy. Despite their importance for asset management decision-making, extracting and leveraging this information at scale remains a significant challenge. This paper presents MaintIE, a multi-level fine-grained annotation scheme for entity recognition and relation extraction, consisting of 5 top-level classes: PhysicalObject, State, Process, Activity and Property and 224 leaf entities, along with 6 relations tailored to MSTs. Using MaintIE, we have curated a multi-annotator, high-quality, fine-grained corpus of 1,076 annotated texts. Additionally, we present a coarse-grained corpus of 7,000 texts and consider its performance for bootstrapping and enhancing fine-grained information extraction. Using these corpora, we provide model performance measures for benchmarking automated entity recognition and relation extraction. The MaintIE scheme, corpus, and model are publicly available at https://github.com/nlp-tlp/maintie under the MIT license, encouraging further community exploration and innovation in extracting valuable insights from MSTs.

Tool learning aims to extend the capabilities of large language models (LLMs) with external tools. A major challenge in tool learning is how to support a large number of tools, including unseen tools. To address this challenge, previous studies have proposed retrieving suitable tools for the LLM based on the user query. However, previously proposed methods do not consider the differences between seen and unseen tools, nor do they take the hierarchy of the tool library into account, which may lead to suboptimal performance for tool retrieval. Therefore, to address the aforementioned issues, we propose ToolRerank, an adaptive and hierarchy-aware reranking method for tool retrieval to further refine the retrieval results. Specifically, our proposed ToolRerank includes Adaptive Truncation, which truncates the retrieval results related to seen and unseen tools at different positions, and Hierarchy-Aware Reranking, which makes retrieval results more concentrated for single-tool queries and more diverse for multi-tool queries. Experimental results show that ToolRerank can improve the quality of the retrieval results, leading to better execution results generated by the LLM.

Recent advancements in large language models (LLMs) have brought significant changes to various domains, especially through LLM-driven autonomous agents. A representative scenario is in software development, where LLM agents demonstrate efficient collaboration, task division, and assurance of software quality, markedly reducing the need for manual involvement. However, these agents frequently perform a variety of tasks independently, without benefiting from past experiences, which leads to repeated mistakes and inefficient attempts in multi-step task execution. To this end, we introduce Experiential Co-Learning, a novel LLM-agent learning framework in which instructor and assistant agents gather shortcut-oriented experiences from their historical trajectories and use these past experiences for future task execution. The extensive experiments demonstrate that the framework enables agents to tackle unseen software-developing tasks more effectively. We anticipate that our insights will guide LLM agents towards enhanced autonomy and contribute to their evolutionary growth in cooperative learning. The code and data are available at https://github.com/OpenBMB/ChatDev.

Most language model pre-training frameworks concatenate multiple documents into fixed-length sequences and use causal masking to compute the likelihood of each token given its context; this strategy is widely adopted due to its simplicity and efficiency. However, to this day, the influence of the pre-training sequence composition strategy on the generalisation properties of the model remains under-explored.In this work, we find that applying causal masking can lead to the inclusion of distracting information from previous documents during pre-training, which negatively impacts the performance of the models on language modelling and downstream tasks. In intra-document causal masking, the likelihood of each token is only conditioned on the previous tokens in the same document, eliminating potential distracting information from previous documents and significantly improving performance. Furthermore, we find that concatenating related documents can reduce some potential distractions during pre-training, and our proposed efficient retrieval-based sequence construction method, Bm25Chunk, can improve in-context learning (+11.6%), knowledge memorisation (+9.8%), and context utilisation (+7.2%) abilities of language models without sacrificing efficiency.

With the remarkable advancements of large language models (LLMs), LLM-based agents have become a research hotspot in human-computer interaction.However, there is a scarcity of benchmarks available for LLM-based mobile agents.Benchmarking these agents generally faces three main challenges:(1) The inefficiency of UI-only operations imposes limitations to task evaluation.(2) Specific instructions within a singular application lack adequacy for assessing the multi-dimensional reasoning and decision-making capacities of LLM mobile agents.(3) Current evaluation metrics are insufficient to accurately assess the process of sequential actions. To this end, we propose Mobile-Bench, a novel benchmark for evaluating the capabilities of LLM-based mobile agents.First, we expand conventional UI operations by incorporating 103 collected APIs to accelerate the efficiency of task completion.Subsequently, we collect evaluation data by combining real user queries with augmentation from LLMs.To better evaluate different levels of planning capabilities for mobile agents, our data is categorized into three distinct groups: SAST, SAMT, and MAMT, reflecting varying levels of task complexity. Mobile-Bench comprises 832 data entries, with more than 200 tasks specifically designed to evaluate multi-APP collaboration scenarios.Furthermore, we introduce a more accurate evaluation metric, named CheckPoint, to assess whether LLM-based mobile agents reach essential points during their planning and reasoning steps. Dataset and platform will be released in the future.

Recent advances in large language models (LLMs) have revolutionized the landscape of reasoning tasks. To enhance the capabilities of LLMs to emulate human reasoning, prior studies have focused on modeling reasoning steps using various thought structures like chains, trees, or graphs. However, LLM-based reasoning still encounters the following challenges: (1) Limited adaptability of preset structures to diverse tasks; (2) Insufficient precision in exploiting known conditions to derive new ones; and (3) Inadequate consideration of historical reasoning experiences for subsequent reasoning steps. To this end, we propose DetermLR, a novel perspective that rethinks the reasoning process as an evolution from indeterminacy to determinacy. First, we categorize known conditions into two types: determinate and indeterminate premises, facilitating the transformation process. Subsequently, we leverage quantitative measurements to prioritize more relevant premises to explore new insights. Furthermore, we automate the storage and extraction of available premises and reasoning paths with reasoning memory, preserving historical reasoning details for subsequent reasoning steps. Comprehensive experimental results demonstrate that DetermLR surpasses all baselines on various logical reasoning benchmarks: LogiQA, ProofWriter, FOLIO, PrOntoQA, and LogicalDeduction. Compared to previous multi-step reasoning methods, DetermLR achieves higher accuracy with fewer reasoning steps, highlighting its superior efficiency and effectiveness in solving logical reasoning tasks.

Software development is a complex task that necessitates cooperation among multiple members with diverse skills. Numerous studies used deep learning to improve specific phases in a waterfall model, such as design, coding, and testing. However, the deep learning model in each phase requires unique designs, leading to technical inconsistencies across various phases, which results in a fragmented and ineffective development process. In this paper, we introduce ChatDev, a chat-powered software development framework in which specialized agents driven by large language models (LLMs) are guided in what to communicate (via chat chain) and how to communicate (via communicative dehallucination). These agents actively contribute to the design, coding, and testing phases through unified language-based communication, with solutions derived from their multi-turn dialogues. We found their utilization of natural language is advantageous for system design, and communicating in programming language proves helpful in debugging. This paradigm demonstrates how linguistic communication facilitates multi-agent collaboration, establishing language as a unifying bridge for autonomous task-solving among LLM agents. The code and data are available at https://github.com/OpenBMB/ChatDev.

Maintenance short texts are invaluable unstructured data sources, serving as a diagnostic and prognostic window into the operational health and status of physical assets. These user-generated texts, created during routine or ad-hoc maintenance activities, offer insights into equipment performance, potential failure points, and maintenance needs. However, the use of information captured in these texts is hindered by inherent challenges: the prevalence of engineering jargon, domain-specific vernacular, random spelling errors without identifiable patterns, and the absence of standard grammatical structures. To transform these texts into accessible and analysable data, we introduce the MaintNorm dataset, the first resource specifically tailored for the lexical normalisation task of maintenance short texts. Comprising 12,000 examples, this dataset enables the efficient processing and interpretation of these texts. We demonstrate the utility of MaintNorm by training a lexical normalisation model as a sequence-to-sequence learning task with two learning objectives, namely, enhancing the quality of the texts and masking segments to obscure sensitive information to anonymise data. Our benchmark model demonstrates a universal error reduction rate of 95.8%. The dataset and benchmark outcomes are available to the public.

This paper proposes a new task of commonsense question generation, which aims to yield deep-level and to-the-point questions from the text. Their answers need to reason over disjoint relevant contexts and external commonsense knowledge, such as encyclopedic facts and causality. The knowledge may not be explicitly mentioned in the text but is used by most humans for problem-shooting. Such complex reasoning with hidden contexts involves deep semantic understanding. Thus, this task has great application value, such as making high-quality quizzes in advanced exams. Due to the lack of modeling complexity, existing methods may produce shallow questions that can be answered by simple word matching. To address these challenges, we propose a new QG model by simultaneously considering asking contents, expressive ways, and answering complexity. We first retrieve text-related commonsense context. Then we disentangle the key factors that control questions in terms of reasoning content and verbalized way. Independence priors and constraints are imposed to facilitate disentanglement. We further develop a discriminator to promote the deep results by considering their answering complexity. Through adversarial inference, we learn the latent factors from data. By sampling the expressive factor from the data distributions, diverse questions can be yielded. Evaluations of two typical data sets show the effectiveness of our approach.

In the context-dependent Text-to-SQL task, the generated SQL statements are refined iteratively based on the user input utterance from each interaction. The input text from each interaction can be viewed as component modifications to the previous SQL statements, which could be further extracted as the modification patterns. Since these modification patterns could also be combined with other SQL statements, the models are supposed to have the compositional generalization to these novel combinations. This work is the first exploration of compositional generalization in context-dependent Text-to-SQL scenarios. To facilitate related studies, we constructed two challenging benchmarks named CoSQL-CG and SParC-CG by recombining the modification patterns and existing SQL statements. The following experiments show that almost all current models struggle on our proposed benchmarks. Furthermore, we found that better aligning the previous SQL statements with the input utterance could give models better combinatorial generalization ability. Based on these observations, we propose a method name p-align to improve the combinatorial generalization of Text-to-SQL models. Further experiments validate the effectiveness of our model.

In this work, we enhance higher-order graph-based approaches for span-based semantic role labeling (SRL) by means of structured modeling. To decrease the complexity of higher-order modeling, we decompose the edge from predicate word to argument span into three different edges, predicate-to-head (P2H), predicate-to-tail (P2T), and head-to-tail (H2T), where head/tail means the first/last word of the semantic argument span. As such, we use a CRF-based higher-order dependency parser and leverage Mean-Field Variational Inference (MFVI) for higher-order inference. Moreover, since semantic arguments of predicates are often constituents within a constituency parse tree, we can leverage such nice structural property by defining a TreeCRF distribution over all H2T edges, using the idea of partial marginalization to define structural training loss. We further leverage structured MFVI to enhance inference. We experiment on span-based SRL benchmarks, showing the effectiveness of both higher-order and structured modeling and the combination thereof. In addition, we show superior performance of structured MFVI against vanilla MFVI.

Geometric representation of query embeddings (using points, particles, rectangles and cones) can effectively achieve the task of answering complex logical queries expressed in first-order logic (FOL) form over knowledge graphs, allowing intuitive encodings. However, current geometric-based methods depend on the neural approach to model FOL operators (conjunction, disjunction and negation), which are not easily explainable with considerable computation cost. We overcome this challenge by introducing a symbolic modeling approach for the FOL operators, emphasizing the direct calculation of the intersection between geometric shapes, particularly sector-cones in the embedding space, to model the conjunction operator. This approach reduces the computation cost as a non-neural approach is involved in the core logic operators. Moreover, we propose to accelerate the learning in the relation projection operator using the neural approach to emphasize the essential role of this operator in all query structures. Although empirical evidence for explainability is challenging, our approach demonstrates a significant improvement in answering complex logical queries (both non-negative and negative FOL forms) over previous geometric-based models.

Scaling dense PCFGs to thousands of nonterminals via low-rank parameterizations of the rule probability tensor has been shown to be beneficial for unsupervised parsing. However, PCFGs scaled this way still perform poorly as a language model, and even underperform similarly-sized HMMs. This work introduces SimplePCFG, a simple PCFG formalism with independent left and right productions. Despite imposing a stronger independence assumption than the low-rank approach, we find that this formalism scales more effectively both as a language model and as an unsupervised parser. We further introduce FlashInside, a hardware IO-aware implementation of the inside algorithm for efficiently scaling simple PCFGs. Through extensive experiments on multiple grammar induction benchmarks, we validate the effectiveness of simple PCFGs over low-rank baselines.

Recent geometric-based approaches have been shown to efficiently model complex logical queries (including the intersection operation) over Knowledge Graphs based on the natural representation of Venn diagram. Existing geometric-based models (using points, boxes embeddings), however, cannot handle the logical negation operation. Further, those using cones embeddings are limited to representing queries by two-dimensional shapes, which reduced their effectiveness in capturing entities query relations for correct answers. To overcome this challenge, we propose unbounded cylinder embeddings (namely CylE), which is a novel geometric-based model based on three-dimensional shapes. Our approach can handle a complete set of basic first-order logic operations (conjunctions, disjunctions and negations). CylE considers queries as Cartesian products of unbounded sector-cylinders and consider a set of nearest boxes corresponds to the set of answer entities. Precisely, the conjunctions can be represented via the intersections of unbounded sector-cylinders. Transforming queries to Disjunctive Normal Form can handle queries with disjunctions. The negations can be represented by considering the closure of complement for an arbitrary unbounded sector-cylinder. Empirical results show that the performance of multi-hop reasoning task using CylE significantly increases over state-of-the-art geometric-based query embedding models for queries without negation. For queries with negation operations, though the performance is on a par with the best performing geometric-based model, CylE significantly outperforms a recent distribution-based model.

Semantic parsing maps natural language questions into logical forms, which can be executed against a knowledge base for answers. In real-world applications, the performance of a parser is often limited by the lack of training data. To facilitate zero-shot learning, data synthesis has been widely studied to automatically generate paired questions and logical forms. However, data synthesis methods can hardly cover the diverse structures in natural languages, leading to a large gap in sentence structure between synthetic and natural questions. In this paper, we propose a decomposition-based method to unify the sentence structures of questions, which benefits the generalization to natural questions. Experiments demonstrate that our method significantly improves the semantic parser trained on synthetic data (+7.9% on KQA and +8.9% on ComplexWebQuestions in terms of exact match accuracy). Extensive analysis demonstrates that our method can better generalize to natural questions with novel text expressions compared with baselines. Besides semantic parsing, our idea potentially benefits other semantic understanding tasks by mitigating the distracting structure features. To illustrate this, we extend our method to the task of sentence embedding learning, and observe substantial improvements on sentence retrieval (+13.1% for Hit@1).

Stock market prediction is considered a complex task due to the non-stationary and volatile nature of the stock markets. With the increasing amount of online data, various information sources have been analyzed to understand the underlying patterns of the price movements. However, most existing works in the literature mostly focus on either the intra-modality information within each input data type, or the inter-modal relationships among the input modalities. Different from these, in this research, we propose a novel Multi-Modal Cross Attention Network for Stock Market Prediction (MCASP) by capturing both modality-specific features and the joint influence of each modality in a unified framework. We utilize financial news, historical market data and technical indicators to predict the movement direction of the market prices. After processing the input modalities with three separate deep networks, we first construct a self-attention network that utilizes multiple Transformer models to capture the intra-modal information. Then we design a novel cross-attention network that processes the inputs in pairs to exploit the cross-modal and joint information of the modalities. Experiments with real world datasets for S&P500 index forecast and the prediction of five individual stocks, demonstrate the effectiveness of the proposed multi-modal design over several state-of-the-art baseline models.

Entity and Relation Extraction (ERE) is an important task in information extraction. Recent marker-based pipeline models achieve state-of-the-art performance, but still suffer from the error propagation issue. Also, most of current ERE models do not take into account higher-order interactions between multiple entities and relations, while higher-order modeling could be beneficial.In this work, we propose HyperGraph neural network for ERE (HGERE), which is built upon the PL-marker (a state-of-the-art marker-based pipleline model). To alleviate error propagation, we use a high-recall pruner mechanism to transfer the burden of entity identification and labeling from the NER module to the joint module of our model. For higher-order modeling, we build a hypergraph, where nodes are entities (provided by the span pruner) and relations thereof, and hyperedges encode interactions between two different relations or between a relation and its associated subject and object entities. We then run a hypergraph neural network for higher-order inference by applying message passing over the built hypergraph. Experiments on three widely used benchmarks (ACE2004, ACE2005 and SciERC) for ERE task show significant improvements over the previous state-of-the-art PL-marker.

HITS participated in the Discourse Segmentation (DS, Task 1) and Connective Detection (CD, Task 2) tasks at the DISRPT 2023. Task 1 focuses on segmenting the text into discourse units, while Task 2 aims to detect the discourse connectives. We deployed a framework based on different pre-trained models according to the target language for these two tasks.HITS also participated in the Relation Classification track (Task 3). The main task was recognizing the discourse relation between text spans from different languages. We designed a joint model for languages with a small corpus while separate models for large corpora. The adversarial training strategy is applied to enhance the robustness of relation classifiers.

Morality in dialogue systems has raised great attention in research recently. A moral dialogue system aligned with users’ values could enhance conversation engagement and user connections. In this paper, we propose a framework, MoralDial to train and evaluate moral dialogue systems. In our framework, we first explore the communication mechanisms of morality and resolve expressed morality into three parts, which indicate the roadmap for building a moral dialogue system. Based on that, we design a simple yet effective method: constructing moral discussions between simulated specific users and the dialogue system. The constructed discussions consist of expressing, explaining, revising, and inferring moral views in dialogue exchanges, which makes conversational models learn morality well in a natural manner. Furthermore, we propose a novel evaluation method under the framework. We evaluate the multiple aspects of morality by judging the relation between dialogue responses and human values in discussions, where the multifaceted nature of morality is particularly considered. Automatic and manual experiments demonstrate that our framework is promising to train and evaluate moral dialogue systems.

Coherence is an important aspect of text quality, and various approaches have been applied to coherence modeling. However, existing methods solely focus on a single document’s coherence patterns, ignoring the underlying correlation between documents. We investigate a GCN-based coherence model that is capable of capturing structural similarities between documents. Our model first creates a graph structure for each document, from where we mine different subgraph patterns. We then construct a heterogeneous graph for the training corpus, connecting documents based on their shared subgraphs. Finally, a GCN is applied to the heterogeneous graph to model the connectivity relationships. We evaluate our method on two tasks, assessing discourse coherence and automated essay scoring. Results show that our GCN-based model outperforms all baselines, achieving a new state-of-the-art on both tasks.

Rationalization is to employ a generator and a predictor to construct a self-explaining NLP model in which the generator selects a subset of human-intelligible pieces of the input text to the following predictor. However, rationalization suffers from two key challenges, i.e., spurious correlation and degeneration, where the predictor overfits the spurious or meaningless pieces solely selected by the not-yet well-trained generator and in turn deteriorates the generator. Although many studies have been proposed to address the two challenges, they are usually designed separately and do not take both of them into account. In this paper, we propose a simple yet effective method named MGR to simultaneously solve the two problems. The key idea of MGR is to employ multiple generators such that the occurrence stability of real pieces is improved and more meaningful pieces are delivered to the predictor. Empirically, we show that MGR improves the F1 score by up to 20.9% as compared to state-of-the-art methods.

Implicit discourse relation classification is a challenging task due to the absence of discourse connectives. To overcome this issue, we design an end-to-end neural model to explicitly generate discourse connectives for the task, inspired by the annotation process of PDTB. Specifically, our model jointly learns to generate discourse connectives between arguments and predict discourse relations based on the arguments and the generated connectives. To prevent our relation classifier from being misled by poor connectives generated at the early stage of training while alleviating the discrepancy between training and inference, we adopt Scheduled Sampling to the joint learning. We evaluate our method on three benchmarks, PDTB 2.0, PDTB 3.0, and PCC. Results show that our joint model significantly outperforms various baselines on three datasets, demonstrating its superiority for the task.

Natural language inference (NLI) is a task to infer the relationship between a premise and a hypothesis (e.g., entailment, neutral, or contradiction), and transformer-based models perform well on current NLI datasets such as MNLI and SNLI. Nevertheless, given the linguistic complexity of the large-scale datasets, it remains controversial whether these models can truly infer the relationship between sentences or they simply guess the answer via shallow heuristics. Here, we introduce a controlled evaluation set called Simple Pair to test the basic sentence inference ability of NLI models using sentences with syntactically simple structures. Three popular transformer-based models, i.e., BERT, RoBERTa, and DeBERTa, are employed. We find that these models fine-tuned on MNLI or SNLI perform very poorly on Simple Pair (< 35.4% accuracy). Further analyses reveal event coreference and compositional binding problems in these models. To improve the model performance, we augment the training set, i.e., MNLI or SNLI, with a few examples constructed based on Simple Pair ( 1% of the size of the original SNLI/MNLI training sets). Models fine-tuned on the augmented training set maintain high performance on MNLI/SNLI and perform very well on Simple Pair (~100% accuracy). Furthermore, the positive performance of the augmented training models can transfer to more complex examples constructed based on sentences from MNLI and SNLI. Taken together, the current work shows that (1) models achieving high accuracy on mainstream large-scale datasets still lack the capacity to draw accurate inferences on simple sentences, and (2) augmenting mainstream datasets with a small number of target simple sentences can effectively improve model performance.

Acquiring high-quality annotated corpora for complex multi-task information extraction (MT-IE) is an arduous and costly process for human-annotators. Adoption of unsupervised techniques for automated annotation have thus become popular. However, these techniques rely heavily on dictionaries, gazetteers, and knowledge bases. While such resources are abundant for general domains, they are scarce for specialised technical domains. To tackle this challenge, we present QuickGraph, the first collaborative MT-IE annotation tool built with indirect weak supervision and clustering to maximise annotator productivity. QuickGraph’s main contribution is a set of novel features that enable knowledge graph extraction through rapid and consistent complex multi-task entity and relation annotation. In this paper, we discuss these key features and qualitatively compare QuickGraph to existing annotation tools.

Hidden Markov Models (HMMs) and Probabilistic Context-Free Grammars (PCFGs) are widely used structured models, both of which can be represented as factor graph grammars (FGGs), a powerful formalism capable of describing a wide range of models. Recent research found it beneficial to use large state spaces for HMMs and PCFGs. However, inference with large state spaces is computationally demanding, especially for PCFGs. To tackle this challenge, we leverage tensor rank decomposition (aka. CPD) to decrease inference computational complexities for a subset of FGGs subsuming HMMs and PCFGs. We apply CPD on the factors of an FGG and then construct a new FGG defined in the rank space. Inference with the new FGG produces the same result but has a lower time complexity when the rank size is smaller than the state size. We conduct experiments on HMM language modeling and unsupervised PCFG parsing, showing better performance than previous work. Our code is publicly available at https://github.com/VPeterV/RankSpace-Models.

NLP systems are often challenged by difficulties arising from noisy, non-standard, and domain specific corpora. The task of lexical normalisation aims to standardise such corpora, but currently lacks suitable tools to acquire high-quality annotated data to support deep learning based approaches. In this paper, we present LexiClean, the first open-source web-based annotation tool for multi-task lexical normalisation. LexiClean’s main contribution is support for simultaneous in situ token-level modification and annotation that can be rapidly applied corpus wide. We demonstrate the usefulness of our tool through a case study on two sets of noisy corpora derived from the specialised-domain of industrial mining. We show that LexiClean allows for the rapid and efficient development of high-quality parallel corpora. A demo of our system is available at: https://youtu.be/P7_ooKrQPDU.

Precise information of word boundary can alleviate the problem of lexical ambiguity to improve the performance of natural language processing (NLP) tasks. Thus, Chinese word segmentation (CWS) is a fundamental task in NLP. Due to the development of pre-trained language models (PLM), pre-trained knowledge can help neural methods solve the main problems of the CWS in significant measure. Existing methods have already achieved high performance on several benchmarks (e.g., Bakeoff-2005). However, recent outstanding studies are limited by the small-scale annotated corpus. To further improve the performance of CWS methods based on fine-tuning the PLMs, we propose a novel neural framework, LBGCN, which incorporates a lexicon-based graph convolutional network into the Transformer encoder. Experimental results on five benchmarks and four cross-domain datasets show the lexicon-based graph convolutional network successfully captures the information of candidate words and helps to improve performance on the benchmarks (Bakeoff-2005 and CTB6) and the cross-domain datasets (SIGHAN-2010). Further experiments and analyses demonstrate that our proposed framework effectively models the lexicon to enhance the ability of basic neural frameworks and strengthens the robustness in the cross-domain scenario.

Lexicon information and pre-trained models, such as BERT, have been combined to explore Chinese sequence labeling tasks due to their respective strengths. However, existing methods solely fuse lexicon features via a shallow and random initialized sequence layer and do not integrate them into the bottom layers of BERT. In this paper, we propose Lexicon Enhanced BERT (LEBERT) for Chinese sequence labeling, which integrates external lexicon knowledge into BERT layers directly by a Lexicon Adapter layer. Compared with existing methods, our model facilitates deep lexicon knowledge fusion at the lower layers of BERT. Experiments on ten Chinese datasets of three tasks including Named Entity Recognition, Word Segmentation, and Part-of-Speech Tagging, show that LEBERT achieves state-of-the-art results.

Our system participates in two shared tasks, CL-SciSumm 2020 and LongSumm 2020. In the CL-SciSumm shared task, based on our previous work, we apply more machine learning methods on position features and content features for facet classification in Task1B. And GCN is introduced in Task2 to perform extractive summarization. In the LongSumm shared task, we integrate both the extractive and abstractive summarization ways. Three methods were tested which are T5 Fine-tuning, DPPs Sampling, and GRU-GCN/GAT.

This paper focuses on generating multi-hop reasoning questions from the raw text in a low resource circumstance. Such questions have to be syntactically valid and need to logically correlate with the answers by deducing over multiple relations on several sentences in the text. Specifically, we first build a multi-hop generation model and guide it to satisfy the logical rationality by the reasoning chain extracted from a given text. Since the labeled data is limited and insufficient for training, we propose to learn the model with the help of a large scale of unlabeled data that is much easier to obtain. Such data contains rich expressive forms of the questions with structural patterns on syntax and semantics. These patterns can be estimated by the neural hidden semi-Markov model using latent variables. With latent patterns as a prior, we can regularize the generation model and produce the optimal results. Experimental results on the HotpotQA data set demonstrate the effectiveness of our model. Moreover, we apply the generated results to the task of machine reading comprehension and achieve significant performance improvements.

A recently proposed lattice model has demonstrated that words in character sequence can provide rich word boundary information for character-based Chinese NER model. In this model, word information is integrated into a shortcut path between the start and the end characters of the word. However, the existence of shortcut path may cause the model to degenerate into a partial word-based model, which will suffer from word segmentation errors. Furthermore, the lattice model can not be trained in batches due to its DAG structure. In this paper, we propose a novel word-character LSTM(WC-LSTM) model to add word information into the start or the end character of the word, alleviating the influence of word segmentation errors while obtaining the word boundary information. Four different strategies are explored in our model to encode word information into a fixed-sized representation for efficient batch training. Experiments on benchmark datasets show that our proposed model outperforms other state-of-the-arts models.

Various Seq2Seq learning models designed for machine translation were applied for abstractive summarization task recently. Despite these models provide high ROUGE scores, they are limited to generate comprehensive summaries with a high level of abstraction due to its degenerated attention distribution. We introduce Diverse Convolutional Seq2Seq Model(DivCNN Seq2Seq) using Determinantal Point Processes methods(Micro DPPs and Macro DPPs) to produce attention distribution considering both quality and diversity. Without breaking the end to end architecture, DivCNN Seq2Seq achieves a higher level of comprehensiveness compared to vanilla models and strong baselines. All the reproducible codes and datasets are available online.

This paper describes our end-to-end speech translation system for the speech translation task of lectures and TED talks from English to German for IWSLT Evaluation 2019. We propose layer-tied self-attention for end-to-end speech translation. Our method takes advantage of sharing weights of speech encoder and text decoder. The representation of source speech and the representation of target text are coordinated layer by layer, so that the speech and text can learn a better alignment during the training procedure. We also adopt data augmentation to enhance the parallel speech-text corpus. The En-De experimental results show that our best model achieves 17.68 on tst2015. Our ASR achieves WER of 6.6% on TED-LIUM test set. The En-Pt model can achieve about 11.83 on the MuST-C dev set.

We introduce Redcoat, a web-based annotation tool that supports collaborative hierarchical entity typing. As an annotation tool, Redcoat also facilitates knowledge elicitation by allowing the creation and continuous refinement of concept hierarchies during annotation. It aims to minimise not only annotation time but the time it takes for project creators to set up and distribute projects to annotators. Projects created using the web-based interface can be rapidly distributed to a list of email addresses. Redcoat handles the propagation of documents amongst annotators and automatically scales the annotation workload depending on the number of active annotators. In this paper we discuss these key features and outline Redcoat’s system architecture. We also highlight Redcoat’s unique benefits over existing annotation tools via a qualitative comparison.

MultiLing 2019 Headline Generation Task on Wikipedia Corpus raised a critical and practical problem: multilingual task on low resource corpus. In this paper we proposed QDAS extractive summarization model enhanced by sentence2vec and try to apply transfer learning based on large multilingual pre-trained language model for Wikipedia Headline Generation task. We treat it as sequence labeling task and develop two schemes to handle with it. Experimental results have shown that large pre-trained model can effectively utilize learned knowledge to extract certain phrase using low resource supervised data.

We present NovelPerspective: a tool to allow consumers to subset their digital literature, based on point of view (POV) character. Many novels have multiple main characters each with their own storyline running in parallel. A well-known example is George R. R. Martin’s novel: “A Game of Thrones”, and others from that series. Our tool detects the main character that each section is from the POV of, and allows the user to generate a new ebook with only those sections. This gives consumers new options in how they consume their media; allowing them to pursue the storylines sequentially, or skip chapters about characters they find boring. We present two heuristic-based baselines, and two machine learning based methods for the detection of the main character.

The availability of large collections of text have made it possible to build language models that incorporate counts of billions of n-grams. This paper proposes two new methods of efficiently storing large language models that allow O(1) random access and use significantly less space than all known approaches. We introduce two novel data structures that take advantage of the distribution of n-grams in corpora and make use of various numbers of minimal perfect hashes to compactly store language models containing full frequency counts of billions of n-grams using 2.5 Bytes per n-gram and language models of quantized probabilities using 2.26 Bytes per n-gram. These methods allow language processing applications to take advantage of much larger language models than previously was possible using the same hardware and we additionally describe how they can be used in a distributed environment to store even larger models. We show that our approaches are simple to implement and can easily be combined with pruning and quantization to achieve additional reductions in the size of the language model.

The Internet has become the most popular platform for communication. However because most of the modern computer keyboard is Latin-based, Asian languages such as Chinese cannot input its characters (Hanzi) directly with these keyboards. As a result, methods for representing Chinese characters using Latin alphabets were introduced. The most popular method among these is the Pinyin input system. Pinyin is also called “Romanised” Chinese in that it phonetically resembles a Chinese character. Due to the highly ambiguous mapping from Pinyin to Chinese characters, word misuses can occur using standard computer keyboard, and more commonly so in internet chat-rooms or instant messengers where the language used is less formal. In this paper we aim to develop a system that can automatically identify such anomalies, whether they are simple typos or whether they are intentional. After identifying them, the system should suggest the correct word to be used.

Data sparsity is a large problem in natural language processing that refers to the fact that language is a system of rare events, so varied and complex, that even using an extremely large corpus, we can never accurately model all possible strings of words. This paper examines the use of skip-grams (a technique where by n-grams are still stored to model language, but they allow for tokens to be skipped) to overcome the data sparsity problem. We analyze this by computing all possible skip-grams in a training corpus and measure how many adjacent (standard) n-grams these cover in test documents. We examine skip-gram modelling using one to four skips with various amount of training data and test against similar documents as well as documents generated from a machine translation system. In this paper we also determine the amount of extra training data required to achieve skip-gram coverage using standard adjacent tri-grams.