Multi-View Graph Representation Learning for Answering Hybrid Numerical Reasoning Question
This addresses the challenge of numerical reasoning in hybrid data for financial analysis, representing an incremental improvement over existing methods.
The paper tackled the problem of hybrid question answering over financial reports, which requires selecting evidence from both textual and tabular data for numerical reasoning, by proposing a Multi-View Graph Encoder to preserve granularity relationships and spatial structure, and achieved state-of-the-art performance on the TAT-QA benchmark.
Hybrid question answering (HybridQA) over the financial report contains both textual and tabular data, and requires the model to select the appropriate evidence for the numerical reasoning task. Existing methods based on encoder-decoder framework employ a expression tree-based decoder to solve numerical reasoning problems. However, encoders rely more on Machine Reading Comprehension (MRC) methods, which take table serialization and text splicing as input, damaging the granularity relationship between table and text as well as the spatial structure information of table itself. In order to solve these problems, the paper proposes a Multi-View Graph (MVG) Encoder to take the relations among the granularity into account and capture the relations from multiple view. By utilizing MVGE as a module, we constuct Tabular View, Relation View and Numerical View which aim to retain the original characteristics of the hybrid data. We validate our model on the publicly available table-text hybrid QA benchmark (TAT-QA) and outperform the state-of-the-art model.