Revealing the Semantics of Data Wrangling Scripts With COMANTICS

TL;DR

COMANTICS is a novel pipeline that automatically infers the semantics of data wrangling scripts by analyzing table differences and employing neural networks, aiding understanding and reuse of such scripts.

Contribution

The paper introduces COMANTICS, a three-step method combining table difference analysis and neural networks to detect data transformation semantics automatically.

Findings

High accuracy in detecting transformation types

Effective across multiple domains

Improves understanding of data wrangling scripts

Abstract

Data workers usually seek to understand the semantics of data wrangling scripts in various scenarios, such as code debugging, reusing, and maintaining. However, the understanding is challenging for novice data workers due to the variety of programming languages, functions, and parameters. Based on the observation that differences between input and output tables highly relate to the type of data transformation, we outline a design space including 103 characteristics to describe table differences. Then, we develop COMANTICS, a three-step pipeline that automatically detects the semantics of data transformation scripts. The first step focuses on the detection of table differences for each line of wrangling code. Second, we incorporate a characteristic-based component and a Siamese convolutional neural network-based component for the detection of transformation types. Third, we derive the parameters of each data transformation by employing a "slot filling" strategy. We design experiments to evaluate the performance of COMANTICS. Further, we assess its flexibility using three example applications in different domains.