Duoquest: A Dual-Specification System for Expressive SQL Queries

Baik, Christopher; Jin, Zhongjun; Cafarella, Michael; Jagadish, H. V.

doi:10.1145/3318464.3389776

Cited by 13 publications

(12 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Zero noise means we sample values from the ground truth columns. In Medium noise we sample 2 3 values from the ground truth columns and 1 3 from a noise column, which is a column with a Jaccard Containment of more than 0.8 with respect to the ground truth column. Finally, in High noise we sample 1 3 values from the ground truth column and 2 3 from the noise columns.…”

Section: Methodsmentioning

confidence: 99%

“…Query-By-Example systems [2,17,39,40,42] and Query-Reverse-Engineering systems [13,31,37,45] are designed to perform on a database with well-defined path information and integrity constraint as indicated in the Supports Pathless column of Table 1. They explore join paths between tables through looking at the primary key/foreign key relationship and all join paths are ensured to be correct.…”

Section: Challenges Of Pathless Table Collectionsmentioning

confidence: 99%

“…Given a column of example values as input (𝜒.𝐴 𝑖 ), this component generates a small set of columns Cand(𝐴 𝑖 ) that contain these values. Many QBE systems [2,17] follow the strict assumption that the columns containing all example values are only considered as candidates for subsequent stages of the pipeline. However, the noisy QBE-style query may not contain all values from the ground truth PJ-view (Challenge 1), thereby affecting its ability to identify adequate PJ-views.…”

Section: Column Retrievalmentioning

confidence: 99%

See 2 more Smart Citations

Niffler: A Reference Architecture and System Implementation for View Discovery over Pathless Table Collections by Example

Gong¹,

Zhu²,

Galhotra³

et al. 2021

Preprint

View full text Add to dashboard Cite

Identifying a project-join view (PJ-view) over collections of tables is the first step of many data management projects, e.g., assembling a dataset to feed into a business intelligence tool, creating a training dataset to fit a machine learning model, and more. When the table collections are large and lack join information-such as when combining databases, or on data lakes-query by example (QBE) systems can help identify relevant data, but they are designed under the assumption that join information is available in the schema, and do not perform well on pathless table collections that do not have join path information.We present a reference architecture that explicitly divides the end-to-end problem of discovering PJ-views over pathless table collections into a human and a technical problem. We then present Niffler, a system built to address the technical problem. We introduce algorithms for the main components of Niffler, including a signal generation component that helps reduce the size of the candidate views that may be large due to errors and ambiguity in both the data and input queries. We evaluate Niffler on real datasets to demonstrate the effectiveness of the new engine in discovering PJ-views over pathless table collections.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Challenges Of Pathless Table Collectionsmentioning

confidence: 99%

Section: Column Retrievalmentioning

confidence: 99%

See 1 more Smart Citation

Niffler: A Reference Architecture and System Implementation for View Discovery over Pathless Table Collections by Example

Gong¹,

Zhu²,

Galhotra³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We note that our system's input is comprised entirely of weakly supervised data that can be procured without the use of expert annotators. As described in §2, question-answer annotations can be provided by non-experts, unfamiliar with SQL [6,36,61,68]. As for QDMR instances, they can also be crowdsourced to non-experts [63] or automatically generated using a trained ML model [22,49].…”

Section: System Overviewmentioning

confidence: 99%

“…The NL-to-SQL model is then trained on these synthesized examples. 6 5.3.1 Training Data. We experiment with two large NL-to-SQL datasets: Spider [66] and Geo880 [67].…”

Section: Automatically Predicted Decompositionsmentioning

confidence: 99%

Weakly Supervised Text-to-SQL Parsing through Question Decomposition

Wolfson¹,

Deutch²,

Berant³

2021

Preprint

View full text Add to dashboard Cite

Natural Language Interfaces to Databases (NLIDBs), where users pose queries in Natural Language (NL), are crucial for enabling non-experts to gain insights from data. Developing such interfaces, by contrast, is dependent on experts who often code heuristics for mapping NL to SQL. Alternatively, NLIDBs based on machine learning models rely on supervised examples of NL to SQL mappings (NL-SQL pairs) used as training data. Such examples are again procured using experts, which typically involves more than a oneoff interaction. Namely, each data domain in which the NLIDB is deployed may have different characteristics and therefore require either dedicated heuristics or domain-specific training examples.To this end, we propose an alternative approach for training machine learning-based NLIDBs, using weak supervision. We use the recently proposed question decomposition representation called QDMR, an intermediate between NL and formal query languages. Recent work has shown that non-experts are generally successful in translating NL to QDMR. We consequently use NL-QDMR pairs, along with the question answers, as supervision for automatically synthesizing SQL queries. The NL questions and synthesized SQL are then used to train NL-to-SQL models, which we test on five benchmark datasets. Extensive experiments show that our solution, requiring zero expert annotations, performs competitively with models trained on expert annotated data.

show abstract

mRAT-SQL+GAP: A Portuguese Text-to-SQL Transformer

José

Cozman

2021

Intelligent Systems

View full text Add to dashboard Cite

Duoquest: A Dual-Specification System for Expressive SQL Queries

Cited by 13 publications

References 19 publications

Niffler: A Reference Architecture and System Implementation for View Discovery over Pathless Table Collections by Example

Niffler: A Reference Architecture and System Implementation for View Discovery over Pathless Table Collections by Example

Weakly Supervised Text-to-SQL Parsing through Question Decomposition

mRAT-SQL+GAP: A Portuguese Text-to-SQL Transformer

Contact Info

Product

Resources

About