Answering questions over financial reports containing both tabular and textual data (hybrid data) is challenging as it requires models to select information from financial reports and perform complex quantitative analyses. Although current models have demonstrated a solid capability to solve simple questions, they struggle with complex questions that require a multiple-step numerical reasoning process. This paper proposes a new framework named FinMath, which improves the model’s numerical reasoning capacity by injecting a tree-structured neural model to perform multi-step numerical reasoning. Specifically, FinMath extracts supporting evidence from the financial reports given the question in the first phase. In the second phase, a tree-structured neural model is applied to generate a tree expression in a top-down recursive way. Experiments on the TAT-QA dataset show that our proposed approach improves the previous best result by 8.5% absolute for Exact Match (EM) score (50.1% to 58.6%) and 6.1% absolute for numeracy-focused F1 score (58.0% to 64.1%).

Numerical reasoning over hybrid data containing both textual and tabular content (e.g., financial reports) has recently attracted much attention in the NLP community. However, existing question answering (QA) benchmarks over hybrid data only include a single flat table in each document and thus lack examples of multi-step numerical reasoning across multiple hierarchical tables. To facilitate data analytical progress, we construct a new large-scale benchmark, MultiHiertt, with QA pairs over Multi Hierarchical Tabular and Textual data. MultiHiertt is built from a wealth of financial reports and has the following unique characteristics: 1) each document contain multiple tables and longer unstructured texts; 2) most of tables contained are hierarchical; 3) the reasoning process required for each question is more complex and challenging than existing benchmarks; and 4) fine-grained annotations of reasoning processes and supporting facts are provided to reveal complex numerical reasoning. We further introduce a novel QA model termed MT2Net, which first applies facts retrieving to extract relevant supporting facts from both tables and text and then uses a reasoning module to perform symbolic reasoning over retrieved facts. We conduct comprehensive experiments on various baselines. The experimental results show that MultiHiertt presents a strong challenge for existing baselines whose results lag far behind the performance of human experts. The dataset and code are publicly available at https://github.com/psunlpgroup/MultiHiertt.