SPARTA: Scalable and Principled Benchmark of Tree-Structured Multi-hop QA over Text and Tables

TL;DR

SPARTA is a scalable framework for automatically generating large, high-quality multi-hop table-text question answering benchmarks that reveal weaknesses in current models' reasoning abilities.

Contribution

The paper introduces SPARTA, a novel automated framework for creating extensive, realistic multi-hop QA datasets over tables and text, with techniques ensuring question fluency and logical correctness.

Findings

State-of-the-art models drop over 30 F1 points on SPARTA, indicating current weaknesses.

SPARTA significantly reduces annotation time compared to previous benchmarks.

Generated questions cover complex reasoning operations like aggregation and multi-hop inference.

Abstract

Real-world Table-Text question answering (QA) tasks require models that can reason across long text and source tables, traversing multiple hops and executing complex operations such as aggregation. Yet existing benchmarks are small, manually curated - and therefore error-prone - and contain shallow questions that seldom demand more than two hops or invoke aggregations, grouping, or other advanced analytical operations expressible in natural-language queries. We present SPARTA, an end-to-end construction framework that automatically generates large-scale Table-Text QA benchmarks with lightweight human validation, requiring only one quarter of the annotation time of HybridQA. The framework first constructs a reference fact database by enriching each source table with grounding tables whose tuples are atomic facts automatically extracted from the accompanying unstructured passages, then synthesizes nested queries whose number of nested predicates matches the desired hop count. To ensure that every SQL statement is executable and that its verbalization yields a fluent, human-sounding question, we propose two novel techniques: provenance-based refinement, which rewrites any syntactically valid query that returns a non-empty result, and realistic-structure enforcement, which confines generation to post-order traversals of the query graph. The resulting pipeline produces thousands of high-fidelity question-answer pairs covering aggregations, grouping, and deep multi-hop reasoning across text and tables. On SPARTA, state-of-the-art models that reach over 70 F1 on HybridQA or over 50 F1 on OTT-QA drop by more than 30 F1 points, exposing fundamental weaknesses in current cross-modal reasoning. Our benchmark, construction code, and baseline models are available at https://github.com/pshlego/SPARTA/tree/main.