Language Model Pre-training for Hierarchical Document Representations

By Ming-Wei Chang, Kristina Toutanova, Kenton C. T. Lee, Jacob Devlin. 2019

26 citations recorded by the source corpus.

Hierarchical neural architectures are often used to capture long-distance dependencies and have been applied to many document-level tasks such as summarization, document segmentation, and sentiment analysis. However, effective usage of such a large context can be difficult to learn, especially in the case where there is limited labeled data available. Building on the recent success of language model pretraining methods for learning flat representations of text, we propose algorithms for pre-training hierarchical document representations from unlabeled data. Unlike prior work, which has focused on pre-training contextual token representations or context-independent {sentence/paragraph} representations, our hierarchical document representations include fixed-length sentence/paragraph representations which integrate contextual information from the entire documents. Experiments on document segmentation, document-level question answering, and extractive document summarization demonstrate the effectiveness of the proposed pre-training algorithms.

Subfields: Artificial Intelligence, Management Science and Operations Research, Large Language Models, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Management Science and Operations Research

In Large Language Models

In Natural Language Processing

Citations in the corpus

CausaLM: Causal Model Explanation Through Counterfactual Language Models (2020, background): “Since the introduction of pre-trained word-embeddings, there have been an explosion of research on choosing pre-training tasks and understanding their effect (Jernite, Bowman, and Sontag 2017; Logeswaran and Lee 2018; Ziser and Reichart 2018; Dong et al. 2019; Chang et al. 2019; Sun et al. 2019; Rot”
SLM: Learning a Discourse Language Representation with Sentence Unshuffling (2020, background): “HLSTM (Chang et al., 2019) considered contextualization of sentence representations by incorporating previous step’s sentence representation for word prediction.”
ReadTwice: Reading Very Large Documents with Memories (2021, unknown): “This is achieved through hierarchical modeling (Chang et al., 2019; Zhang et al., 2019).”