Position: Foundation Models for Tabular Data within Systemic Contexts Need Grounding

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Klein, Tassilo, Hoffart, Johannes
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918293048131584
author Klein, Tassilo
Hoffart, Johannes
author_facet Klein, Tassilo
Hoffart, Johannes
contents This position paper argues that foundation models for tabular data face inherent limitations when isolated from operational context - the procedural logic, declarative rules, and domain knowledge that define how data is created and governed. Current approaches focus on single-table generalization or schema-level relationships, fundamentally missing the operational knowledge that gives data meaning. We introduce Semantically Linked Tables (SLT) and Foundation Models for SLT (FMSLT) as a new model class that grounds tabular data in its operational context. We propose dual-phase training: pre-training on open-source code-data pairs and synthetic systems to learn business logic mechanics, followed by zero-shot inference on proprietary data. We introduce the ``Operational Turing Test'' benchmark and argue that operational grounding is essential for autonomous agents in complex data environments.
format Preprint
id arxiv_https___arxiv_org_abs_2505_19825
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Position: Foundation Models for Tabular Data within Systemic Contexts Need Grounding
Klein, Tassilo
Hoffart, Johannes
Machine Learning
Artificial Intelligence
Databases
This position paper argues that foundation models for tabular data face inherent limitations when isolated from operational context - the procedural logic, declarative rules, and domain knowledge that define how data is created and governed. Current approaches focus on single-table generalization or schema-level relationships, fundamentally missing the operational knowledge that gives data meaning. We introduce Semantically Linked Tables (SLT) and Foundation Models for SLT (FMSLT) as a new model class that grounds tabular data in its operational context. We propose dual-phase training: pre-training on open-source code-data pairs and synthetic systems to learn business logic mechanics, followed by zero-shot inference on proprietary data. We introduce the ``Operational Turing Test'' benchmark and argue that operational grounding is essential for autonomous agents in complex data environments.
title Position: Foundation Models for Tabular Data within Systemic Contexts Need Grounding
topic Machine Learning
Artificial Intelligence
Databases
url https://arxiv.org/abs/2505.19825