Learning Context-Dependent Mappings from Sentences to Logical Form

By Luke S. Zettlemoyer, Michael Collins. 2009

208 citations recorded by the source corpus.

We consider the problem of learning context-dependent mappings from sentences to logical form. The training examples are sequences of sentences annotated with lambda-calculus meaning representations. We develop an algorithm that maintains explicit, lambda-calculus representations of salient discourse entities and uses a context-dependent analysis pipeline to recover logical forms. The method uses a hidden-variable variant of the perception algorithm to learn a linear model used to select the best analysis. Experiments on context-dependent utterances from the ATIS corpus show that the method recovers fully correct logical forms with 83.7% accuracy.

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

MCTest: A Challenge Dataset for the Open-Domain Machine Comprehension of Text (2013, methodology): “On the other hand, the goal of semantic parsing is the machine comprehension of text (MCT), yet its evaluation requires adherence to a specific knowledge representation, and it is currently unclear what the best representation is, for open-domain text.”
STaR: Bootstrapping Reasoning With Reasoning (2022, background): “The idea of alternating between filtering steps and training steps more broadly is also well-grounded in prior work in NLP such as [23] and in other weak supervision contexts [24, 25].”
Grammar-Based Grounded Lexicon Learning (2022, background): “For this reason, previous approaches to lexicon learning have either assumed an expert-annotated set of lexical entries [48] or only attempted to learn at very small scales [16].”
Learning to Automatically Solve Algebra Word Problems (2014, background): “Techniques that consider multiple sentences typically do so in a serial fashion, processing each in turn with limited cross-sentence reasoning (Branavan et al., 2009; Zettlemoyer and Collins, 2009; Chen and Mooney, 2011; Artzi and Zettlemoyer, 2013).”
Scaling Semantic Parsers with On-the-Fly Ontology Matching (2013, methodology): “, 2010, 2011), building derivations to transform the output of the CCG parser based on context (Zettlemoyer and Collins, 2009), and using weakly supervised margin-sensitive parameter updates (Artzi and Zettlemoyer, 2011, 2013).”
Search-based Neural Structured Learning for Sequential Question Answering (2017, background): “Perhaps most related to our task is the context-dependent sentence analysis described in (Zettlemoyer and Collins, 2009), where conversations between customers and travel agents are mapped to logical forms after resolving referential expressions.”
Learning to Win by Reading Manuals in a Monte-Carlo Framework (2011, neutral): “…non-linguistic situated context (Oates, 2001; Barnard & Forsyth, 2001; Siskind, 2001; Roy & Pentland, 2002; Yu & Ballard, 2004; Chen & Mooney, 2008; Zettlemoyer & Collins, 2009; Liang, Jordan, & Klein, 2009; Branavan, Chen, Zettlemoyer, & Barzilay, 2009; Branavan, Zettlemoyer, & Barzilay, 2010;…”
Information Extraction over Structured Data: Question Answering with Freebase (2014, background): “Most work approaches this via the bridge of various intermediate representations, including combinatory categorial grammar (Zettlemoyer and Collins, 2005, 2007, 2009; Kwiatkowski et al., 2010, 2011, 2013), synchronous context-free grammars (Wong and Mooney, 2007), dependency trees (Liang et al.,…”
JuICe: A Large Scale Distantly Supervised Dataset for Open Domain Context-based Code Generation (2019, methodology): “, 2018) ATIS (Zettlemoyer and Collins, 2009) SequentialQA (Iyyer et al.”
Inducing Probabilistic CCG Grammars from Logical Form with Higher-Order Unification (2010, background): “We are also interested in developing similar grammar induction techniques for context-dependent understanding problems, such as the one considered by Zettlemoyer & Collins (2009).”
A survey on question answering technology from an information retrieval perspective (2011, methodology): “Along these lines is the work of Zettlemoyer and Collins [151], who map sentences to lambda calculus encodings of their meaning λx.”
Answering Complicated Question Intents Expressed in Decomposed Question Sequences (2016, methodology): “questions, there have been many studies linking increased sentence complexity to longer reading times (Hale, 2006; Levy, 2008; Frank, 2013). 4To be released at http://aka.ms/sqa analysis described in Zettlemoyer and Collins (2009), where conversations between customers and travel agents are mapped t”
Task-Oriented Dialogue as Dataflow Synthesis (2020, methodology): “Most existing methods interpret context-dependent user questions (What is the next flight to Atlanta? When does it land?) by learning to copy subtrees (Zettlemoyer and Collins, 2009; Iyyer et al., 2017; Suhr et al., 2018) or tokens (Zhang et al.”
Straight to the Facts: Learning Knowledge Base Retrieval for Factual Visual Question Answering (2018, background): “There has been interest in the natural language processing community in answering questions based on knowledge bases (KBs) using either semantic parsing [37,38,39,40,41,42,43,44,45,46,47] or information retrieval [48,49,50,51,52,53,54] methods.”
Context Dependent Semantic Parsing: A Survey (2020, background): “Meaning Representations In (Zettlemoyer and Collins, 2009), there is a feature indicating if the predicates exist in the history LFs.”
Bootstrapping Semantic Parsers from Conversations (2011, background): “…see Steedman (2000) for details. lated, natural language semantic analysis tasks from context-dependent database queries (Miller et al., 1996; Zettlemoyer and Collins, 2009), grounded event streams (Chen et al., 2010; Liang et al., 2009), environment interactions (Branavan et al., 2009;…”