TabularNet: A Neural Network Architecture for Understanding Semantic Structures of Tabular Data

TL;DR

TabularNet is a novel neural network architecture designed to automatically understand the semantic structures of tabular data by capturing both spatial and relational information, improving performance on classification tasks.

Contribution

The paper introduces TabularNet, combining spatial encoding with hierarchical relational graph encoding, a novel approach for semantic understanding of tabular data.

Findings

Outperforms state-of-the-art baselines on classification tasks

Effectively captures hierarchical and paratactic relationships between cells

Demonstrates versatility in multitask scenarios

Abstract

Tabular data are ubiquitous for the widespread applications of tables and hence have attracted the attention of researchers to extract underlying information. One of the critical problems in mining tabular data is how to understand their inherent semantic structures automatically. Existing studies typically adopt Convolutional Neural Network (CNN) to model the spatial information of tabular structures yet ignore more diverse relational information between cells, such as the hierarchical and paratactic relationships. To simultaneously extract spatial and relational information from tables, we propose a novel neural network architecture, TabularNet. The spatial encoder of TabularNet utilizes the row/column-level Pooling and the Bidirectional Gated Recurrent Unit (Bi-GRU) to capture statistical information and local positional correlation, respectively. For relational information, we design a new graph construction method based on the WordNet tree and adopt a Graph Convolutional Network (GCN) based encoder that focuses on the hierarchical and paratactic relationships between cells. Our neural network architecture can be a unified neural backbone for different understanding tasks and utilized in a multitask scenario. We conduct extensive experiments on three classification tasks with two real-world spreadsheet data sets, and the results demonstrate the effectiveness of our proposed TabularNet over state-of-the-art baselines.