Extending a Parser to Distant Domains Using a Few Dozen Partially Annotated Examples

By Vidur Joshi, Matthew E. Peters, Mark Hopkins. 2018

102 citations recorded by the source corpus.

We revisit domain adaptation for parsers in the neural era. First we show that recent advances in word representations greatly diminish the need for domain adaptation when the target domain is syntactically similar to the source domain. As evidence, we train a parser on the Wall Street Journal alone that achieves over 90% F1 on the Brown corpus. For more syntactically distant domains, we provide a simple way to adapt a parser using only dozens of partial annotations. For instance, we increase the percentage of error-free geometry-domain parses in a held-out set from 45% to 73% using approximately five dozen training examples. In the process, we demonstrate a new state-of-the-art single model result on the Wall Street Journal test set of 94.3%. This is an absolute increase of 1.7% over the previous state-of-the-art of 92.6%.

Subfields: Artificial Intelligence, 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 Natural Language Processing

Citations in the corpus

Dissecting Contextual Word Embeddings: Architecture and Representation (2018, background): “All systems use residual connections between the contextual layers (He et al., 2016).”
Generalizing from a Few Examples (2019, background): “Example applications include parsing [64], translation [65], sentence completion (which fills in the blanks using a word chosen from a provided set) [97, 138], sentiment classification from short reviews [150, 157], user intent classification for dialog systems [157], criminal charge prediction [61]”
Multilingual Constituency Parsing with Self-Attention and Pre-Training (2018, background): “This differs slightly from Joshi et al. (2018), where tags are predicted based on span representations instead.”
From LSAT: The Progress and Challenges of Complex Reasoning (2021, methodology): “a) Logic Identification: It employs a constituency parser [89] to extract constituents including noun phrases and gerundial phrases from the context as basic logical symbols.”
Generative artificial intelligence: a systematic review and applications (2024, unknown): “Another model related to Generative Natural language processing is ELMo [55].”
Finding a Balanced Degree of Automation for Summary Evaluation (2021, methodology): “Therefore, we get the Con-stituency Parsing tree (Joshi et al., 2018) of each sentence and deﬁne the following features: (1) sentence length; (2) linearized parsing tree length; (3) parsing tree depth; (4) sentence length / parsing tree depth; (5) the counts for each of the 65 non-terminal tokens…”
Learning Syntax from Naturally-Occurring Bracketings (2021, background): “Our work also points to potential applications in (semi-)supervised settings including active learning and domain adaptation (Joshi et al., 2018).”
Cross-Domain Generalization of Neural Constituency Parsers (2019, background): “Ad-ditionally, Joshi et al. (2018) showed that these representations (in their case, from ELMo, Peters et al. 2018) allow a parser to transfer well across domains.”
ScispaCy: Fast and Robust Models for Biomedical Natural Language Processing (2019, methodology): “(Rimell and Clark, 2008) adapt a CCG parser using only POS and lexical categories, while (Joshi et al., 2018) extend a neural phrase structure parser trained on web text to the biomedical domain with a small number of partially annotated examples.”
Logic-Driven Context Extension and Data Augmentation for Logical Reasoning of Text (2021, methodology): “To ensure the generalizability of our framework without annotated logic forms, we design a fairly simple logical identification approach using an offthe-shelf constituency parser (Joshi et al., 2018)”
Measuring the Feasibility of a Question and Answering System for the Sarawak Gazette Using Chatbot Technology (2025, unknown): “In 2018, the first pre-trained language model (PLM), Embeddings from Language Models (ELMo) was proposed (Joshi et al, 2018).”