Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferen 2019
DOI: 10.18653/v1/d19-1263
|View full text |Cite
|
Sign up to set email alerts
|

Leveraging Frequent Query Substructures to Generate Formal Queries for Complex Question Answering

Abstract: Formal query generation aims to generate correct executable queries for question answering over knowledge bases (KBs), given entity and relation linking results. Current approaches build universal paraphrasing or ranking models for the whole questions, which are likely to fail in generating queries for complex, long-tail questions. In this paper, we propose SubQG, a new query generation approach based on frequent query substructures, which helps rank the existing (but nonsignificant) query structures or build … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

1
32
0
3

Year Published

2021
2021
2024
2024

Publication Types

Select...
3
2
2

Relationship

0
7

Authors

Journals

citations
Cited by 45 publications
(36 citation statements)
references
References 13 publications
1
32
0
3
Order By: Relevance
“…The follow-up work tried to improve the formulation of query graphs. To generalize to unseen and long-tail question types, Ding et al [2019] proposed to leverage frequent query substructure for formal query generation. Abujabal et al [2017] utilized syntactic annotation to enhance the structural complexity of the query graph.…”
Section: Semantic Parsing-based Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…The follow-up work tried to improve the formulation of query graphs. To generalize to unseen and long-tail question types, Ding et al [2019] proposed to leverage frequent query substructure for formal query generation. Abujabal et al [2017] utilized syntactic annotation to enhance the structural complexity of the query graph.…”
Section: Semantic Parsing-based Methodsmentioning
confidence: 99%
“…To derive a more interpretable reasoning process, multi-hop reasoning is introduced. Specifically, and Xu et al [2019] proposed to make the relation or entity predicted at each hop traceable and observable. They output intermediate predictions (i.e., matched relations or entities) from predefined memory as the interpretable reasoning path.…”
Section: Information Retrieval-based Methodsmentioning
confidence: 99%
“…Um dos principais problemas da RNN é a queda no desempenho para sequências mais longas e complexas. Para resolvê-lo, trabalhos recentes utilizam o mecanismo de atenc ¸ão para enfatizar as partes mais relevantes de uma NLQ e preservar o contexto das sentenc ¸as [Bhutani et al 2019, Ding et al 2019, Tong et al 2019, Bhutani et al 2020. Embora RNNs sejam amplamente utilizadas, Redes de Memória [Miller et al 2016, Hao et al 2019, Saha et al 2018, Hua et al 2020b, Hua et al 2020a] e Redes Neurais Convolucionais [Hu et al 2018, Bao et al 2016] podem ser usadas nesta etapa.…”
Section: Representac ¸ãO Das Perguntas E Gerac ¸ãO De Candidatosunclassified
“…As abordagens de análise semântica baseada em redes neurais tentam resolver perguntas complexas usando uma combinac ¸ão de análise semântica e arquiteturas de redes neurais e estão se tornando o estado da arte [Luo et al 2018, Ding et al 2019. Essa abordagem consiste em treinar uma rede neural para corresponder a um conjunto de regras de análise semântica, em vez de apenas a resposta final.…”
Section: Representac ¸ãO Das Perguntas E Gerac ¸ãO De Candidatosunclassified
See 1 more Smart Citation