-
Ultrasolid Homotopical Algebra
Authors:
Sofía Marlasca Aparicio
Abstract:
Solid modules over $\mathbb{Q}$ or $\mathbb{F}_p$, introduced by Clausen and Scholze, are a well-behaved variant of complete topological vector spaces that forms a symmetric monoidal Grothendieck abelian category. For a discrete field $k$, we construct the category of ultrasolid $k$-modules, which specialises to solid modules over $\mathbb{Q}$ or $\mathbb{F}_p$. In this setting, we show some commu…
▽ More
Solid modules over $\mathbb{Q}$ or $\mathbb{F}_p$, introduced by Clausen and Scholze, are a well-behaved variant of complete topological vector spaces that forms a symmetric monoidal Grothendieck abelian category. For a discrete field $k$, we construct the category of ultrasolid $k$-modules, which specialises to solid modules over $\mathbb{Q}$ or $\mathbb{F}_p$. In this setting, we show some commutative algebra results like an ultrasolid variant of Nakayama's lemma. We also explore higher algebra in the form of animated and $\mathbb{E}_\infty$ ultrasolid $k$-algebras, and their deformation theory. We focus on the subcategory of complete profinite $k$-algebras, which we prove is contravariantly equivalent to equal characteristic formal moduli problems with coconnective tangent complex.
△ Less
Submitted 6 June, 2024;
originally announced June 2024.
-
On Repairing Natural Language to SQL Queries
Authors:
Aidan Z. H. Yang,
Ricardo Brancas,
Pedro Esteves,
Sofia Aparicio,
Joao Pedro Nadkarni,
Miguel Terra-Neves,
Vasco Manquinho,
Ruben Martins
Abstract:
Data analysts use SQL queries to access and manipulate data on their databases. However, these queries are often challenging to write, and small mistakes can lead to unexpected data output. Recent work has explored several ways to automatically synthesize queries based on a user-provided specification. One promising technique called text-to-SQL consists of the user providing a natural language des…
▽ More
Data analysts use SQL queries to access and manipulate data on their databases. However, these queries are often challenging to write, and small mistakes can lead to unexpected data output. Recent work has explored several ways to automatically synthesize queries based on a user-provided specification. One promising technique called text-to-SQL consists of the user providing a natural language description of the intended behavior and the database's schema. Even though text-to-SQL tools are becoming more accurate, there are still many instances where they fail to produce the correct query.
In this paper, we analyze when text-to-SQL tools fail to return the correct query and show that it is often the case that the returned query is close to a correct query. We propose to repair these failing queries using a mutation-based approach that is agnostic to the text-to-SQL tool being used. We evaluate our approach on two recent text-to-SQL tools, RAT-SQL and SmBoP, and show that our approach can repair a significant number of failing queries.
△ Less
Submitted 5 October, 2023;
originally announced October 2023.
-
Natural language to SQL in low-code platforms
Authors:
Sofia Aparicio,
Samuel Arcadinho,
João Nadkarni,
David Aparício,
João Lages,
Mariana Lourenço,
Bartłomiej Matejczyk,
Filipe Assunção
Abstract:
One of the developers' biggest challenges in low-code platforms is retrieving data from a database using SQL queries. Here, we propose a pipeline allowing developers to write natural language (NL) to retrieve data. In this study, we collect, label, and validate data covering the SQL queries most often performed by OutSystems users. We use that data to train a NL model that generates SQL. Alongside…
▽ More
One of the developers' biggest challenges in low-code platforms is retrieving data from a database using SQL queries. Here, we propose a pipeline allowing developers to write natural language (NL) to retrieve data. In this study, we collect, label, and validate data covering the SQL queries most often performed by OutSystems users. We use that data to train a NL model that generates SQL. Alongside this, we describe the entire pipeline, which comprises a feedback loop that allows us to quickly collect production data and use it to retrain our SQL generation model. Using crowd-sourcing, we collect 26k NL and SQL pairs and obtain an additional 1k pairs from production data. Finally, we develop a UI that allows developers to input a NL query in a prompt and receive a user-friendly representation of the resulting SQL query. We use A/B testing to compare four different models in production and observe a 240% improvement in terms of adoption of the feature, 220% in terms of engagement rate, and a 90% decrease in failure rate when compared against the first model that we put into production, showcasing the effectiveness of our pipeline in continuously improving our feature.
△ Less
Submitted 29 August, 2023;
originally announced August 2023.
-
Accretion and photodesorption of CO ice as a function of the incident angle of deposition
Authors:
C González Díaz,
H Carrascosa de Lucas,
S Aparicio,
G M Muñoz Caro,
N-E Sie,
L-C Hsiao,
Y-J Chen
Abstract:
Non-thermal desorption of inter- and circum-stellar ice mantles on dust grains, in particular ultraviolet photon-induced desorption, has gained importance in recent years. These processes may account for the observed gas phase abundances of molecules like CO toward cold interstellar clouds. Ice mantle growth results from gas molecules im**ing on the dust from all directions and incidence angles.…
▽ More
Non-thermal desorption of inter- and circum-stellar ice mantles on dust grains, in particular ultraviolet photon-induced desorption, has gained importance in recent years. These processes may account for the observed gas phase abundances of molecules like CO toward cold interstellar clouds. Ice mantle growth results from gas molecules im**ing on the dust from all directions and incidence angles. Nevertheless, the effect of the incident angle for deposition on ice photo-desorption rate has not been studied. This work explores the impact on the accretion and photodesorption rates of the incidence angle of CO gas molecules with the cold surface during deposition of a CO ice layer. Infrared spectroscopy monitored CO ice upon deposition at different angles, ultraviolet-irradiation, and subsequent warm-up. Vacuum-ultraviolet spectroscopy and a Ni-mesh measured the emission of the ultraviolet lamp. Molecules ejected from the ice to the gas during irradiation or warm-up were characterized by a quadrupole mass spectrometer. The photodesorption rate of CO ice deposited at 11 K and different incident angles was rather stable between 0 and 45$^{\circ}$. A maximum in the CO photodesorption rate appeared around 70$^{\circ}$-incidence deposition angle. The same deposition angle leads to the maximum surface area of water ice. Although this study of the surface area could not be performed for CO ice, the similar angle dependence in the photodesorption and the ice surface area suggests that they are closely related. Further evidence for a dependence of CO ice morphology on deposition angle is provided by thermal desorption of CO ice experiments.
△ Less
Submitted 7 May, 2019;
originally announced May 2019.
