R2D2: Robust Data-to-Text with Replacement Detection

Nan, Linyong; Flores, Lorenzo Jaime; Zhao, Yilun; Liu, Yixin; Benson, Luke; Weijin, Zou,; Radev, Dragomir

doi:10.18653/v1/2022.emnlp-main.464

Cited by 4 publications

(14 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similar to pretraining in other domains, table-to-text pre-training endeavours to create a generalizable model and jointly learn an enhanced representation of both tabular and textual data. This process usually hinges on the aggregation or synthesis of substantial amounts of data, coupled with the formulation of appropriate pretext tasks and objectives (Nan et al, 2022;Zhao et al, 2022). Nonetheless, table pre-training confronts discernible constraints, such as high consumption of computing resources and relatively poor generalizability.…”

Section: Logic Table-to-text Generationmentioning

confidence: 99%

“…LoFT (Zhao et al, 2023), based on BART-large (Lewis et al, 2020), utilizes logic forms both as fact validators and content planners to control the generative process. R2D2 (Nan et al, 2022) trains T5-base (Raffel et al, 2020) to function both as a generator and a faithfulness discriminator, incorporating supplementary replacement detection and unlikelihood learning tasks.…”

Section: Baselinesmentioning

confidence: 99%

“…In contrast to models such as DCVED (Chen, Tian, et al, 2021) and R2D2 (Nan et al, 2022), which employ a selector following the generation process to select sentences based on specific evaluation criteria, our approach does not incorporate any selector for catering purposes. As a result, our method exhibits certain weaknesses when compared to these models in some assessments.…”

Section: Limitationsmentioning

confidence: 99%

“…Previous models (Chen, Chen, Su, et al, 2020;Chen, Tian, et al, 2021;Nan et al, 2022) mainly focused on addressing the issue of unfaithfulness in logic table-to-text generation. However, these models confront a more underlying challenge: a noticeable deficiency in diversity.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Enhancing diversity for logical table‐to‐text generation with mixture of experts

Wu,

Hou

2024

Expert Systems

View full text Add to dashboard Cite

Logical table‐to‐text generation is a task within the realm of natural language generation (NLG) that aims to generate coherent and logically faithful sentences based on tables. Unlike conventional NLG tasks, this task demands not only surface‐level fluency but also a high degree of logic‐level fidelity in the generated outputs. Current table‐to‐text systems grapple with various quality issues, such as repetitive generation, insufficient reasoning and limited complexity. Therefore, we introduce LogicMoE, a dedicated Mixture‐of‐Experts (MoE) model tailored for logical table‐to‐text generation. The primary objective of LogicMoE is to enrich the diversity of generated sentences from both semantic and logical perspectives. In particular, each expert within the model serves as a specialized generator responsible for generating sentences of a specific logical type. Additionally, we propose and employ novel evaluation metrics to comprehensively assess the diversity of generated outputs. Our experimental results showcase LogicMoE's superiority with absolute improvements of 0.8 and 2.2 in BLEU‐3 over the strong baselines on LogicNLG and Logic2Text datasets, respectively, driving the state‐of‐the‐art performance to a new level. Furthermore, we highlight its inherent advantages in terms of diversity and controllability, signifying its potential to spearhead advancements in logical table‐to‐text generation applications.

show abstract

Section: Logic Table-to-text Generationmentioning

confidence: 99%

Section: Baselinesmentioning

confidence: 99%

Section: Limitationsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Enhancing diversity for logical table‐to‐text generation with mixture of experts

Wu,

Hou

2024

Expert Systems

View full text Add to dashboard Cite

show abstract

“…These techniques can be incorporated into a broad range of applications, including but not limited to game strategy development, financial analysis, and human resources management. However, existing fine-tuned table-to-text generation models (Nan et al, 2022a;Liu et al, 2022b,a;Zhao et al, 2023b) are typically task-specific, limiting their adaptability to real-world applications.…”

Section: Introductionmentioning

confidence: 99%

Investigating Table-to-Text Generation Capabilities of Large Language Models in Real-World Information Seeking Scenarios

Zhao,

Zhang,

et al. 2023

Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: Industry Track

View full text Add to dashboard Cite

Tabular data is prevalent across various industries, necessitating significant time and effort for users to understand and manipulate for their information-seeking purposes. The advancements in large language models (LLMs) have shown enormous potential to improve user efficiency. However, the adoption of LLMs in real-world applications for table information seeking remains underexplored. In this paper, we investigate the table-to-text capabilities of different LLMs using four datasets within two real-world information seeking scenarios. These include the LOGICNLG and our newlyconstructed LOTNLG datasets for data insight generation, along with the FeTaQA and our newly-constructed F2WTQ datasets for querybased generation. We structure our investigation around three research questions, evaluating the performance of LLMs in table-to-text generation, automated evaluation, and feedback generation, respectively. Experimental results indicate that the current high-performing LLM, specifically GPT-4, can effectively serve as a table-to-text generator, evaluator, and feedback generator, facilitating users' information seeking purposes in real-world scenarios. However, a significant performance gap still exists between other open-sourced LLMs (e.g., TÜLU and LLaMA-2) and GPT-4 models. Our data and code are publicly available at https: //github.com/yale-nlp/LLM-T2T.

show abstract