Towards Natural Language Understanding of Procedural Text Using Recipes

With the notable success of pretrained language models, the pretraining-fine-tuning paradigm has become a dominant solution for natural language understanding (NLU) tasks. Typically, the training instances of a target NLU task are introduced in a completely random order and treated equally at the fine-tuning stage. However, these instances can vary greatly in difficulty, and similar to human learning procedures, language models can benefit from an easy-to-difficult curriculum. Based on this concept, we propose a curriculum learning (CL) framework. Our framework consists of two stages, Review and Arrange, targeting the two main challenges in curriculum learning, i.e., how to define the difficulty of instances and how to arrange a curriculum based on the difficulty, respectively. In the first stage, we devise a cross-review (CR) method to train several teacher models first and then review the training set in a crossed manner to distinguish easy instances from difficult instances. In the second stage, two sampling algorithms, a coarse-grained arrangement (CGA) and a fine-grained arrangement (FGA), are proposed to arrange a curriculum for language models in which the learning materials start from the easiest instances, and more difficult instances are gradually added into the training procedure. Compared to previous heuristic CL methods, our framework can avoid the errors caused by a gap in difficulty between humans and machines and has strong generalization ability. We conduct comprehensive experiments, and the results show that our curriculum learning framework, without any manual model architecture design or use of external data, obtains significant and universal performance improvements on a wide range of NLU tasks in different languages.

Detecting biases in natural language understanding (NLU) for African American Vernacular English (AAVE) is crucial to developing inclusive natural language processing (NLP) systems.To address dialectinduced performance discrepancies, we introduce AAVENUE (AAVE Natural Language Understanding Evaluation), a benchmark for evaluating large language model (LLM) performance on NLU tasks in AAVE and Standard American English (SAE).AAVENUE builds upon and extends existing benchmarks like VALUE, replacing deterministic syntactic and morphological transformations with a more flexible methodology leveraging LLMbased translation with few-shot prompting, improving performance across several evaluation metrics when translating key tasks from the GLUE and SuperGLUE benchmarks.We compare AAVENUE and VALUE translations using five popular LLMs and a comprehensive set of metrics including fluency, BARTScore, quality, coherence, and understandability.Additionally, the fluency of AAVENUE is validated by annotations from AAVE speakers.Our evaluations reveal that LLMs consistently perform better on SAE tasks than AAVEtranslated versions, underscoring inherent biases and highlighting the need for more inclusive NLP models.We have open-sourced our source code on GitHub and created a website to showcase our work at https://aavenue.live.

An essential part of AI is NLU, or natural language understanding, which allows computers to understand and use human language. Training distinct models for tasks like sentiment analysis, question answering, and text categorization is a common practice in traditional natural language understanding (NLU) techniques; nevertheless, this can lead to high computing costs and poor generalizability. a method for improving performance on several natural language understanding (NLU) problems through multi-task learning with Deep Neural Networks (DNNs). Models may take advantage of shared linguistic structures through multi-task learning, which improves efficiency and generalizability by allowing representation sharing between tasks. We showcase an integrated design that acquires knowledge of many natural language understanding tasks all at once and evaluate its efficacy on benchmark datasets such as GLUE and SuperGLUE. We found that by training the model to recognize common semantic and syntactic patterns across tasks, multi-task learning improved task performance while also reducing computing burden. Improved and more reliable natural language systems are possible thanks to the findings of this work, which shed light on the practical application of multi-task learning to NLU.

A method for creating a vision-and-language (V&L) model is to extend a language model through structural modifications and V&L pre-training. Such an extension aims to make a V&L model inherit the capability of natural language understanding (NLU) from the original language model. To see how well this is achieved, we propose to evaluate V&L models using an NLU benchmark (GLUE). We compare five V&L models, including single-stream and dual-stream models, trained with the same pre-training. Dual-stream models, with their higher modality independence achieved by approximately doubling the number of parameters, are expected to preserve the NLU capability better. Our main finding is that the dual-stream scores are not much different than the single-stream scores, contrary to expectation. Further analysis shows that pre-training causes the performance drop in NLU tasks with few exceptions. These results suggest that adopting a single-stream structure and devising the pre-training could be an effective method for improving the maintenance of language knowledge in V&L extensions.

It is by now widely accepted that a number of tasks in natural language understanding (NLU) require the storage of and reasoning with a vast amount of background (commonsense) knowledge. While several efforts have been made to build such ontologies, a consensus on a scientific methodology for ontological design is yet to emerge. In this paper we suggest an approach to building a commonsense ontology for language understanding using language itself as a design guide. The idea is rooted in Frege's conception of compositional semantics and is related to the idea of type inferences in strongly-typed, polymorphic programming languages. The method proposed seems to (i) resolve the problem of multiple inheritance; (ii) suggest an explanation for polysemy and metaphor; and (iii) provide a step towards establishing a systematic approach to ontological design.

Document-level relation extraction with multi-layer heterogeneous graph attention network

Coreference Resolution (CR) is a crucial yet challenging task in natural language understanding, often constrained by task-specific architectures and encoder-based language models that demand extensive training and lack adaptability. This study introduces the first multilingual CR methodology which leverages decoder-only LLMs to handle both overt and zero mentions. The article explores how to model the CR task for LLMs via five different instruction sets using a controlled inference method. The approach is evaluated across three LLMs: Llama 3.1, Gemma 2, and Mistral 0.3. The results indicate that LLMs, when instruction-tuned with a suitable instruction set, can surpass state-of-the-art task-specific architectures. Specifically, our best model, a fully fine-tuned Llama 3.1 for multilingual CR, outperforms the leading multilingual CR model (i.e., Corpipe 24 single stage variant) by 2 percentage points on average across all languages in the CorefUD v1.2 dataset collection.

Paraphrase identification is a crucial task in natural language understanding, especially in cross-language information retrieval. Nowadays, Multi-Task Deep Neural Network (MT-DNN) has become a state-of-the-art method that brings outstanding results in paraphrase identification [1]. In this paper, our proposed method based on MT-DNN [2] to detect similarities between English and Vietnamese sentences, is proposed. We changed the shared layers of the original MT-DNN from original the BERT [3] to other pre-trained multi-language models such as M-BERT [3] or XLM-R [4] so that our model could work on cross-language (in our case, English and Vietnamese) information retrieval. We also added some tasks as improvements to gain better results. As a result, we gained 2.3% and 2.5% increase in evaluated accuracy and F1. The proposed method was also implemented on other language pairs such as English – German and English – French. With those implementations, we got a 1.0%/0.7% improvement for English – German and a 0.7%/0.5% increase for English – French.

Text matching is a basic and important task in natural language understanding, this paper proposes a new model BBMC for the problem of insufficient feature extraction ability of existing text matching models, which integrates BiLSTM and multi-scale CNN on the basis of BERT. First, the word embedding representation of the text is obtained by the BERT, and then the semantic features of the text are further extracted by the double-layer BiLSTM, followed by the multi-scale CNN model, the key local features are extracted, and finally the linear and SoftMax function are used to classify. Experimental results on the LCQMC dataset show that the BBMC has been improved to a certain extent compared with other methods, and the accuracy on the test set can be best achieved 88.01%.

Detecting events and their evolution through time is a crucial task in natural language understanding. Recent neural approaches to event temporal relation extraction typically map events to embeddings in the Euclidean space and train a classifier to detect temporal relations between event pairs. However, embeddings in the Euclidean space cannot capture richer asymmetric relations such as event temporal relations. We thus propose to embed events into hyperbolic spaces, which are intrinsically oriented at modeling hierarchical structures. We introduce two approaches to encode events and their temporal relations in hyperbolic spaces. One approach leverages hyperbolic embeddings to directly infer event relations through simple geometrical operations. In the second one, we devise an end-to-end architecture composed of hyperbolic neural units tailored for the temporal relation extraction task. Thorough experimental assessments on widely used datasets have shown the benefits of revisiting the tasks on a different geometrical space, resulting in state-of-the-art performance on several standard metrics. Finally, the ablation study and several qualitative analyses highlighted the rich event semantics implicitly encoded into hyperbolic spaces.

ELICE: Embedding Language through Informative Contrastive-Encoder

Language Model (LM) which is commonly trained on a large corpora has been proven the robustness and effectiveness for tasks of Natural Language Understanding (NLU) in many applications such as virtual assistant or recommendation system. These applications normally receive outputs of automatic speech recognition (ASR) module as spoken form inputs which generally lack both lexical and syntactic information. Pre-trained language models, for example BERT [1] or XLM-RoBERTa [2], which are often pre-trained on written form corpora perform decreased performance on NLU tasks with spoken form inputs. In this paper, we propose a novel model to train a language model namely CapuBERT that is able to deal with spoken form input from ASR module. The experimental results show that the proposed model achieves state-of-the-art results on several NLU tasks included Part-of-speech tagging, Named-entity recognition and Chunking in English, German, and Vietnamese languages.

Natural Language Understanding (NLU) and Natural Language Generation (NLG) are the general methods that support machine understanding of text content. They play a very important role in the text information processing system including recommendation and question and answer systems. There are many researches in the field of NLU such as Bag of words, N-Gram, and neural network language model. These models have achieved a good performance in NLU and NLG tasks. However, since they require lots of training data, it is difficult to obtain rich data in practical applications. Thus, pretraining becomes important. This paper proposes a semisupervised way to deal with math word problem (MWP) tasks using unsupervised pretraining and supervised tuning methods, which are based on the Unified pretrained Language Model (UniLM). The proposed model requires fewer training data than traditional models since it uses model parameters of tasks that have been learned before to initialize the model parameters of new tasks. In this way, old knowledge helps new models successfully perform new tasks from old experiences instead of from scratch. Moreover, in order to help the decoder make accurate predictions, we combine the advantages of AR and AE language models to support one-way, sequence-to-sequence, and two-way predictions. Experiments, carried out on MWP tasks with 20,000+ mathematical questions, show that the improved model outperforms the traditional models with a maximum accuracy of 79.57%. The impact of different experiment parameters is also studied in the paper and we found that a wrong arithmetic order leads to incorrect solution expression generation.

Unit testing is a fundamental aspect of software development, ensuring the correctness and robustness of code implementations. Traditionally, unit tests are manually crafted by developers based on their understanding of the code and its requirements. However, this process can be time-consuming, errorprone, and may overlook certain edge cases. In recent years, there has been growing interest in leveraging large language models (LLMs) for automating the generation of unit tests. LLMs, such as GPT (Generative Pre-trained Transformer), CodeT5, StarCoder, LLaMA, have demonstrated remarkable capabilities in natural language understanding and code generation tasks. By using LLMs, researchers aim to develop techniques that automatically generate unit tests from code snippets or specifications, thus optimizing the software testing process. This paper presents a literature review of articles that use LLMs for unit test generation tasks. It also discusses the history of the most commonly used large language models and their parameters, including the first time they have been used for code generation tasks. The result of this study presents the large language models for code and unit test generation tasks and their increasing popularity in code generation domain, indicating a great promise for the future of unit test generation using LLMs.

Early Exiting is one of the most popular methods to achieve efficient inference. Current early exiting methods adopt the (weighted) sum of the cross entropy loss of all internal classifiers as the objective function during training, imposing all these classifiers to predict all instances correctly. However, during inference, as long as one internal classifier predicts an instance correctly, it can accelerate without losing accuracy. Thus, there is a notable gap between training and inference. We propose ConsistentEE, an early exiting method that is consistent in training and inference. ConsistentEE formulates the early exiting process as a reinforcement learning problem. A policy network is added to decide whether an instance should exit or continue. The training objective of ConsistentEE only requires each instance to be predicted correctly by one internal classifier. Additionally, we introduce the concept "Memorized Layer" to measure the hardness of an instance. We incorporate the memorized layer into reward function design, which allows "easy'' instances to focus more on acceleration while ``hard'' instances to focus more on accuracy. Experimental results show that our method outperforms other baselines on various natural language understanding and generation tasks using PLMs and LLMs as backbones respectively.