Computation and Language (cs.CL)

Abstract: Prompt engineering and calibration make large language models excel at reasoning tasks, including multiple choice commonsense reasoning. From a practical perspective, we investigate and evaluate these strategies on smaller language models. Through experiments on five commonsense reasoning benchmarks, we find that each strategy favors certain models, but their joint effects are mostly negative.

Abstract: Large Language Models (LLMs), such as the LLaMA model, have demonstrated their effectiveness in various general-domain natural language processing (NLP) tasks. Nevertheless, LLMs have not yet performed optimally in biomedical domain tasks due to the need for medical expertise in the responses. In response to this challenge, we propose HuaTuo, a LLaMA-based model that has been supervised-fine-tuned with generated QA (Question-Answer) instances. The experimental results demonstrate that HuaTuo generates responses that possess more reliable medical knowledge. Our proposed HuaTuo model is accessible at https://github.com/SCIR-HI/Huatuo-Llama-Med-Chinese.

Abstract: Text simplification (TS) is the process of generating easy-to-understand sentences from a given sentence or piece of text. The aim of TS is to reduce both the lexical (which refers to vocabulary complexity and meaning) and syntactic (which refers to the sentence structure) complexity of a given text or sentence without the loss of meaning or nuance. In this paper, we present \textsc{SimpLex}, a novel simplification architecture for generating simplified English sentences. To generate a simplified sentence, the proposed architecture uses either word embeddings (i.e., Word2Vec) and perplexity, or sentence transformers (i.e., BERT, RoBERTa, and GPT2) and cosine similarity. The solution is incorporated into a user-friendly and simple-to-use software. We evaluate our system using two metrics, i.e., SARI, and Perplexity Decrease. Experimentally, we observe that the transformer models outperform the other models in terms of the SARI score. However, in terms of Perplexity, the Word-Embeddings-based models achieve the biggest decrease. Thus, the main contributions of this paper are: (1) We propose a new Word Embedding and Transformer based algorithm for text simplification; (2) We design \textsc{SimpLex} -- a modular novel text simplification system -- that can provide a baseline for further research; and (3) We perform an in-depth analysis of our solution and compare our results with two state-of-the-art models, i.e., LightLS [19] and NTS-w2v [44]. We also make the code publicly available online.

Abstract: How does one measure "ability to understand language"? If it is a person's ability that is being measured, this is a question that almost never poses itself in an unqualified manner: Whatever formal test is applied, it takes place on the background of the person's language use in daily social practice, and what is measured is a specialised variety of language understanding (e.g., of a second language; or of written, technical language). Computer programs do not have this background. What does that mean for the applicability of formal tests of language understanding? I argue that such tests need to be complemented with tests of language use embedded in a practice, to arrive at a more comprehensive evaluation of "artificial language understanding". To do such tests systematically, I propose to use "Dialogue Games" -- constructed activities that provide a situational embedding for language use. I describe a taxonomy of Dialogue Game types, linked to a model of underlying capabilites that are tested, and thereby giving an argument for the \emph{construct validity} of the test. I close with showing how the internal structure of the taxonomy suggests an ordering from more specialised to more general situational language understanding, which potentially can provide some strategic guidance for development in this field.

Abstract: Multi-label text classification aims to extract all the related labels from a sentence, which can be viewed as a sequence generation problem. However, the labels in training dataset are unordered. We propose to treat it as a direct set prediction problem and don't need to consider the order of labels. Besides, in order to model the correlation between labels, the adjacency matrix is constructed through the statistical relations between labels and GCN is employed to learn the label information. Based on the learned label information, the set prediction networks can both utilize the sentence information and label information for multi-label text classification simultaneously. Furthermore, the Bhattacharyya distance is imposed on the output probability distributions of the set prediction networks to increase the recall ability. Experimental results on four multi-label datasets show the effectiveness of the proposed method and it outperforms previous method a substantial margin.

Abstract: Entity alignment (EA) aims to find equivalent entities in different knowledge graphs (KGs). State-of-the-art EA approaches generally use Graph Neural Networks (GNNs) to encode entities. However, most of them train the models and evaluate the results in a fullbatch fashion, which prohibits EA from being scalable on largescale datasets. To enhance the usability of GNN-based EA models in real-world applications, we present SEA, a scalable entity alignment system that enables to (i) train large-scale GNNs for EA, (ii) speed up the normalization and the evaluation process, and (iii) report clear results for users to estimate different models and parameter settings. SEA can be run on a computer with merely one graphic card. Moreover, SEA encompasses six state-of-the-art EA models and provides access for users to quickly establish and evaluate their own models. Thus, SEA allows users to perform EA without being involved in tedious implementations, such as negative sampling and GPU-accelerated evaluation. With SEA, users can gain a clear view of the model performance. In the demonstration, we show that SEA is user-friendly and is of high scalability even on computers with limited computational resources.

Abstract: This paper summarizes our contributions to the document-grounded dialog tasks at the 9th and 10th Dialog System Technology Challenges (DSTC9 and DSTC10). In both iterations the task consists of three subtasks: first detect whether the current turn is knowledge seeking, second select a relevant knowledge document, and third generate a response grounded on the selected document. For DSTC9 we proposed different approaches to make the selection task more efficient. The best method, Hierarchical Selection, actually improves the results compared to the original baseline and gives a speedup of 24x. In the DSTC10 iteration of the task, the challenge was to adapt systems trained on written dialogs to perform well on noisy automatic speech recognition transcripts. Therefore, we proposed data augmentation techniques to increase the robustness of the models as well as methods to adapt the style of generated responses to fit well into the proceeding dialog. Additionally, we proposed a noisy channel model that allows for increasing the factuality of the generated responses. In addition to summarizing our previous contributions, in this work, we also report on a few small improvements and reconsider the automatic evaluation metrics for the generation task which have shown a low correlation to human judgments.

Abstract: Having an intelligent dialogue agent that can engage in conversational question answering (ConvQA) is now no longer limited to Sci-Fi movies only and has, in fact, turned into a reality. These intelligent agents are required to understand and correctly interpret the sequential turns provided as the context of the given question. However, these sequential questions are sometimes left implicit and thus require the resolution of some natural language phenomena such as anaphora and ellipsis. The task of question rewriting has the potential to address the challenges of resolving dependencies amongst the contextual turns by transforming them into intent-explicit questions. Nonetheless, the solution of rewriting the implicit questions comes with some potential challenges such as resulting in verbose questions and taking conversational aspect out of the scenario by generating self-contained questions. In this paper, we propose a novel framework, CONVSR (CONVQA using Structured Representations) for capturing and generating intermediate representations as conversational cues to enhance the capability of the QA model to better interpret the incomplete questions. We also deliberate how the strengths of this task could be leveraged in a bid to design more engaging and eloquent conversational agents. We test our model on the QuAC and CANARD datasets and illustrate by experimental results that our proposed framework achieves a better F1 score than the standard question rewriting model.

Abstract: The goal of visual word sense disambiguation is to find the image that best matches the provided description of the word's meaning. It is a challenging problem, requiring approaches that combine language and image understanding. In this paper, we present our submission to SemEval 2023 visual word sense disambiguation shared task. The proposed system integrates multimodal embeddings, learning to rank methods, and knowledge-based approaches. We build a classifier based on the CLIP model, whose results are enriched with additional information retrieved from Wikipedia and lexical databases. Our solution was ranked third in the multilingual task and won in the Persian track, one of the three language subtasks.

Abstract: This paper describes our submission to the SemEval 2023 multilingual tweet intimacy analysis shared task. The goal of the task was to assess the level of intimacy of Twitter posts in ten languages. The proposed approach consists of several steps. First, we perform in-domain pre-training to create a language model adapted to Twitter data. In the next step, we train an ensemble of regression models to expand the training set with pseudo-labeled examples. The extended dataset is used to train the final solution. Our method was ranked first in five out of ten language subtasks, obtaining the highest average score across all languages.

Abstract: Language models pre-trained on large self-supervised corpora, followed by task-specific fine-tuning has become the dominant paradigm in NLP. These pre-training datasets often have a one-to-many structure--e.g. in dialogue there are many valid responses for a given context. However, only some of these responses will be desirable in our downstream task. This raises the question of how we should train the model such that it can emulate the desirable behaviours, but not the undesirable ones. Current approaches train in a one-to-one setup--only a single target response is given for a single dialogue context--leading to models only learning to predict the average response, while ignoring the full range of possible responses. Using text-based games as a testbed, our approach, PASA, uses discrete latent variables to capture the range of different behaviours represented in our larger pre-training dataset. We then use knowledge distillation to distil the posterior probability distribution into a student model. This probability distribution is far richer than learning from only the hard targets of the dataset, and thus allows the student model to benefit from the richer range of actions the teacher model has learned. Results show up to 49% empirical improvement over the previous state-of-the-art model on the Jericho Walkthroughs dataset.