arXiv subjects

Ilias Fountalis

Publications and source records attributed to Ilias Fountalis.

6 recordsLinked to original sources

arXiv2026

Database Views as Explanations for Relational Deep Learning

In recent years, there has been significant progress in the development of deep learning models over relational databases, including architectures based on heterogeneous graph neural networks (hetero-GNNs) and heterogeneous graph transformers. In effect, such architectures state how the database records and links (e.g., foreign-key references) translate into a large, complex numerical expression, involving numerous learnable parameters. This complexity makes it hard to explain, in human-understandable terms, how a model uses the available data to arrive at a given prediction. We present a novel framework for explaining machine-learning models over relational databases, where explanations are view definitions that highlight focused parts of the database that mostly contribute to the model's prediction. We establish such global abductive explanations by adapting the classic notion of determinacy by Nash, Segoufin, and Vianu (2010). In addition to tuning the tradeoff between determinacy and conciseness, the framework allows controlling the level of granularity by adopting different fragments of view definitions, such as ones highlighting whole columns, foreign keys between tables, relevant groups of tuples, and so on. We investigate the realization of the framework in the case of hetero-GNNs, and develop a model-specific approach via the notion of learnable masks. For comparison, we propose model-agnostic heuristic baselines and show that our approach is both more efficient and achieves better explanation quality in most cases. Our extensive empirical evaluation on the RelBench collection across diverse domains and record-level tasks demonstrates both the usefulness of our explanations and the efficiency of their generation.

cs.DB↗

arXiv2024

Towards Agentic Schema Refinement

Large enterprise databases can be complex and messy, obscuring the data semantics needed for analytical tasks. We propose a semantic layer in-between the database and the user as a set of small and easy-to-interpret database views, effectively acting as a refined version of the schema. To discover these views, we introduce a multi-agent Large Language Model (LLM) simulation where LLM agents collaborate to iteratively define and refine views with minimal input. Our approach paves the way for LLM-powered exploration of unwieldy databases.

cs.DB↗

arXiv2024

QirK: Question Answering via Intermediate Representation on Knowledge Graphs

We demonstrate QirK, a system for answering natural language questions on Knowledge Graphs (KG). QirK can answer structurally complex questions that are still beyond the reach of emerging Large Language Models (LLMs). It does so using a unique combination of database technology, LLMs, and semantic search over vector embeddings. The glue for these components is an intermediate representation (IR). The input question is mapped to IR using LLMs, which is then repaired into a valid relational database query with the aid of a semantic search on vector embeddings. This allows a practical synthesis of LLM capabilities and KG reliability. A short video demonstrating QirK is available at https://youtu.be/6c81BLmOZ0U.

cs.DB↗

arXiv2024

AnnotatedTables: A Large Tabular Dataset with Language Model Annotations

Tabular data is ubiquitous in real-world applications and abundant on the web, yet its annotation has traditionally required human labor, posing a significant scalability bottleneck for tabular machine learning. Our methodology can successfully annotate a large amount of tabular data and can be flexibly steered to generate various types of annotations based on specific research objectives, as we demonstrate with SQL annotation and input-target column annotation as examples. As a result, we release AnnotatedTables, a collection of 32,119 databases with LLM-generated annotations. The dataset includes 405,616 valid SQL programs, making it the largest SQL dataset with associated tabular data that supports query execution. To further demonstrate the value of our methodology and dataset, we perform two follow-up research studies. 1) We investigate whether LLMs can translate SQL programs to Rel programs, a database language previously unknown to LLMs, while obtaining the same execution results. Using our Incremental Prompt Engineering methods based on execution feedback, we show that LLMs can produce adequate translations with few-shot learning. 2) We evaluate the performance of TabPFN, a recent neural tabular classifier trained on Bayesian priors, on 2,720 tables with input-target columns identified and annotated by LLMs. On average, TabPFN performs on par with the baseline AutoML method, though the relative performance can vary significantly from one data table to another, making both models viable for practical applications depending on the situation. Our findings underscore the potential of LLMs in automating the annotation of large volumes of diverse tabular data.

cs.LG↗

arXiv2024

CHORUS: Foundation Models for Unified Data Discovery and Exploration

We apply foundation models to data discovery and exploration tasks. Foundation models include large language models (LLMs) that show promising performance on a range of diverse tasks unrelated to their training. We show that these models are highly applicable to the data discovery and data exploration domain. When carefully used, they have superior capability on three representative tasks: table-class detection, column-type annotation and join-column prediction. On all three tasks, we show that a foundation-model-based approach outperforms the task-specific models and so the state of the art. Further, our approach often surpasses human-expert task performance. We investigate the fundamental characteristics of this approach including generalizability to several foundation models and the impact of non-determinism on the outputs. All in all, this suggests a future direction in which disparate data management tasks can be unified under foundation models.

cs.DB↗

arXiv2016

δ-MAPS: From spatio-temporal data to a weighted and lagged network between functional domains

We propose δ-MAPS, a method that analyzes spatio-temporal data to first identify the distinct spatial components of the underlying system, referred to as "domains", and second to infer the connections between them. A domain is a spatially contiguous region of highly correlated temporal activity. The core of a domain is a point or subregion at which a metric of local homogeneity is maximum across the entire domain. We compute a domain as the maximum-sized set of spatially contiguous cells that include the detected core and satisfy a homogeneity constraint, expressed in terms of the average pairwise cross-correlation across all cells in the domain. Domains may be spatially overlapping. Different domains may have correlated activity, potentially at a lag, because of direct or indirect interactions. The proposed edge inference method examines the statistical significance of each lagged cross-correlation between two domains, infers a range of lag values for each edge, and assigns a weight to each edge based on the covariance of the two domains. We illustrate the application of δ-MAPS on data from two domains: climate science and neuroscience.

cs.OH↗