Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Tytan: Interactive Neurosymbolic Construction of Analytic Semantic Schemas from Relational Data

Source

arXiv

arxiv.org/abs/2608.06331v1

Read the source

Source

Plain-Language Brief

The paper presents TYTAN, a system that automatically generates an analytic semantic schema from a relational database and user-provided descriptions. It combines symbolic analysis with LLM-based semantic inference for entity proposal, role assignment, and naming. The system was evaluated on eight databases across various domains and showed high coverage, retrieval correctness, and characterization accuracy.

What the Document Says

  • TYTAN is a system that constructs an analytic semantic schema from relational databases with user-provided descriptions.
  • The system uses symbolic analysis and LLM-based semantic inference for entity proposal, role assignment, and naming.
  • When evidence is ambiguous, TYTAN asks the user targeted natural-language questions.
  • Evaluation was done on eight databases across real-world and benchmark domains.
  • Across seven reference domains, TYTAN reached 100% coverage of entities, attributes, and aggregable features in the expert-corrected reference schemas.
  • 100% of its retrieval instructions executed correctly (1,678 claims).
  • Semantic roles agreed with the reference on 92-100% of matched attributes.
  • On a held-out blind test, TYTAN recovered the full entity structure and satisfied 100% of the satisfiable expectations of five independent blind annotators.

Key Concepts Mentioned

  • TYTAN
  • analytic semantic schema
  • relational database
  • LLM-based semantic inference
  • natural-language query interfaces
  • automated report generation

Claims Made by the Document

  • TYTAN can construct an analytic semantic schema from a relational database.
  • TYTAN uses symbolic analysis and LLM-based semantic inference for entity proposal, role assignment, and naming.
  • When evidence is ambiguous, TYTAN asks the user targeted natural-language questions.
  • The system achieved 100% coverage on seven reference domains.
  • 100% of its retrieval instructions executed correctly.
  • Semantic roles agreed with the reference on 92-100% of matched attributes.
  • On a held-out blind test, TYTAN recovered the full entity structure and satisfied 100% of the satisfiable expectations.

Evidence Used

  • From natural-language query interfaces to automated report generation, data analysis tools need a description of the data... Today, this semantic layer is usually written by hand... We present TYTAN... TYTAN combines symbolic analysis with LLM-based semantic inference for entity proposal... When evidence leaves a decision ambiguous... Additionally, 100% of its retrieval instructions execute correctly... Semantic roles agree with the reference on 92-100% of matched attributes... On a held-out blind test...

Methods / Approach

  • Symbolic analysis
  • LLM-based semantic inference
  • natural-language questioning

Limitations Stated by the Authors

The small disagreement between TYTAN and the reference on semantic roles is stated to be in the reference, not in TYTAN.

Limitations Noted by the Agent

Agent interpretation — not from the source.

Not stated in source

Open Questions

  • What is LLM used for in this context?
  • How does TYTAN determine when evidence leaves a decision ambiguous?

Notes for Future Q&A

May include interpretation. Labeled where it does.

  • LLM likely stands for Language Model, but the specific model used is not specified in the text.
  • The text doesn't explicitly explain how TYTAN determines when evidence leaves a decision ambiguous.

Raw Abstract / Excerpt

From natural-language query interfaces to automated report generation, data analysis tools need a description of the data: the real-world entities it contains, which columns function as measures or identifiers, and how tables connect into units of analysis. Today, this semantic layer is usually written by hand. This is a knowledge-acquisition bottleneck that limits the scalability of analytic systems, keeps non-technical users dependent on experts, and is itself error-prone. We present TYTAN, a system for automatically constructing an analytic semantic schema from a relational database and, when available, a short user-provided description. TYTAN combines symbolic analysis of the database with LLM-based semantic inference for entity proposal, role assignment, and naming. When the evidence leaves a decision ambiguous, TYTAN asks the user a targeted natural-language question. We evaluate TYTAN on eight databases spanning real-world and benchmark domains along the three axes that define a schema's functional utility: (i) coverage, are all important entities and features captured?; (ii) retrieval correctness, do the schema's instructions actually reach the data; and (iii) characterization accuracy, are semantic types correct? Across the seven reference domains, TYTAN reaches every entity, attribute, and aggregable feature of the expert-corrected reference schemas (100% coverage). Additionally, 100% of its retrieval instructions execute correctly (1,678 of 1,678 self-generated claims), and semantic roles agree with the reference on 92-100% of matched attributes. Checking the underlying data showed the small disagreement is in the reference, not in TYTAN. On a held-out blind test (a live, ten-table database with no declared keys), TYTAN recovers the full entity structure with verified keys and satisfies 100% of the satisfiable expectations of five independent blind annotators.