Unified Semantic Parsing with Weak Supervision

Agrawal, Priyanka; Dalmia, Ayushi; Jain, Parag; Bansal, Abhishek; Mittal, Anshul; Sankaranarayanan, Karthik

doi:10.18653/v1/p19-1473

Cited by 8 publications

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since annotating a sufficiently large number of SQL programs is a resource heavy task, semantic parsers can be trained with execution output denotations under a weak-supervision setting. These execution denotations can be used to model a reward signal in order to train the underlying semantic parser (Zhong et al, 2017;Liang et al, 2018;Hagopian et al, 2019;Agrawal et al, 2019;Agarwal et al, 2019). They can also be used to train the semantic parser with a log Multiple log Marginal Likelihood (MML) objective by using a limited number of SQL programs as latent logical forms Min et al, 2019;Wang et al, 2021a).…”

Section: Ethical Considerationsmentioning

confidence: 99%

Program Synthesis for Complex QA on Charts via Probabilistic Grammar Based Filtered Iterative Back-Translation

Bhaisaheb,

Paliwal,

Patil

et al. 2023

Findings of the Association for Computational Linguistics: EACL 2023

View full text Add to dashboard Cite

Answering complex reasoning questions from chart images is a challenging problem requiring a combination of natural language understanding, fine-grained perception, and analytical reasoning. Current chart based Question Answering (QA) approaches largely address structural, visual or simple data retrieval type questions with fixed-vocabulary answers and perform poorly on reasoning queries. We focus on answering realistic, complex, reasoningbased questions where the answer needs to be computed and not selected from a fixed set of choices. Our approach employs a neural semantic parser to transform Natural Language (NL) questions into SQL programs and execute them on a standardized schema populated from the extracted chart contents. In the absence of program annotations, i.e., in a weak supervision setting, we obtain initial SQL predictions from a pre-trained CodeT5 semantic parser and employ Filtered Iterative Back-Translation (FIBT) for iteratively augmenting our NL-SQL training set. The forward (neural semantic parser) and backward (language model) models are initially trained with an external NL-SQL bootstrapping data. We iteratively move towards the required NL query distribution by generating NL questions from the synthesized SQL programs using a Probabilistic Context-Free Grammar (PCFG) where the production rule probabilities are induced to be inversely proportional to the probabilities in the training data. We filter out the generated NL queries with mismatched structure and compositions. Our FIBT approach achieves State-of-the-Art (SOTA) results on reasoning-based queries in the PlotQA dataset yielding a test accuracy of 60.44%, superseding the previous baselines by a large margin.

show abstract

Section: Ethical Considerationsmentioning

confidence: 99%

Program Synthesis for Complex QA on Charts via Probabilistic Grammar Based Filtered Iterative Back-Translation

Bhaisaheb,

Paliwal,

Patil

et al. 2023

Findings of the Association for Computational Linguistics: EACL 2023

View full text Add to dashboard Cite

show abstract

“…Witnessed the labeled data bottleneck problem, many techniques have been proposed to reduce the demand for labeled logical forms. Many weakly supervised learning are proposed (Artzi and Zettlemoyer, 2013;Berant et al, 2013;Reddy et al, 2014;Agrawal et al, 2019;, such as denotation-base learning (Pasupat and Liang, 2016;Goldman et al, 2018), iterative searching (Dasigi et al, 2019). Semi-supervised semantic parsing is also proposed (Chen et al, 2018a).…”

Section: Related Workmentioning

confidence: 99%

From Paraphrasing to Semantic Parsing: Unsupervised Semantic Parsing via Synchronous Semantic Decoding

Wu¹,

Chen²,

Xin³

et al. 2021

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Confer

View full text Add to dashboard Cite

Semantic parsing is challenging due to the structure gap and the semantic gap between utterances and logical forms. In this paper, we propose an unsupervised semantic parsing method -Synchronous Semantic Decoding (SSD), which can simultaneously resolve the semantic gap and the structure gap by jointly leveraging paraphrasing and grammarconstrained decoding. Specifically, we reformulate semantic parsing as a constrained paraphrasing problem: given an utterance, our model synchronously generates its canonical utterance 1 and meaning representation. During synchronous decoding: the utterance paraphrasing is constrained by the structure of the logical form, therefore the canonical utterance can be paraphrased controlledly; the semantic decoding is guided by the semantics of the canonical utterance, therefore its logical form can be generated unsupervisedly. Experimental results show that SSD is a promising approach and can achieve competitive unsupervised semantic parsing performance on multiple datasets. * Corresponding Author 1 Canonical utterances are pseudo-language representations of logical forms, which have the synchronous structure of logical forms.

show abstract

“…Supervised approaches often suffer from the lack of training data, because it is expensive to annotate an utterance with its logical form. Many weakly supervised techniques have been investigated, e.g., supervised using denotations for utterances (Liang et al 2017;Guu et al 2017;Cheng and Lapata 2018;Misra et al 2018;Goldman et al 2018;Liang et al 2018;Agrawal et al 2019). One main challenge here is the large search space of potential programs needing to be explored.…”

Section: Related Workmentioning

confidence: 99%

Learning to Map Frequent Phrases to Sub-Structures of Meaning Representation for Neural Semantic Parsing

Chen¹,

Han²,

He³

et al. 2020

AAAI

View full text Add to dashboard Cite

Neural semantic parsers usually generate meaning representation tokens from natural language tokens via an encoder-decoder model. However, there is often a vocabulary-mismatch problem between natural language utterances and logical forms. That is, one word maps to several atomic logical tokens, which need to be handled as a whole, rather than individual logical tokens at multiple steps. In this paper, we propose that the vocabulary-mismatch problem can be effectively resolved by leveraging appropriate logical tokens. Specifically, we exploit macro actions, which are of the same granularity of words/phrases, and allow the model to learn mappings from frequent phrases to corresponding sub-structures of meaning representation. Furthermore, macro actions are compact, and therefore utilizing them can significantly reduce the search space, which brings a great benefit to weakly supervised semantic parsing. Experiments show that our method leads to substantial performance improvement on three benchmarks, in both supervised and weakly supervised settings.

show abstract

Unified Semantic Parsing with Weak Supervision

Cited by 8 publications

References 18 publications

Program Synthesis for Complex QA on Charts via Probabilistic Grammar Based Filtered Iterative Back-Translation

Program Synthesis for Complex QA on Charts via Probabilistic Grammar Based Filtered Iterative Back-Translation

From Paraphrasing to Semantic Parsing: Unsupervised Semantic Parsing via Synchronous Semantic Decoding

Learning to Map Frequent Phrases to Sub-Structures of Meaning Representation for Neural Semantic Parsing

Contact Info

Product

Resources

About