Advances in Theories, Methods and Philosophy

Loading...

Media is loading
 

Paper Number

2559

Paper Type

Completed

Description

This paper presents an artifact to automate and quantify qualitative coding of text documents on predefined constructs of interest. The quantified codes, similar to Likert-scaled measures in survey research, can be analyzed using statistical models. Our approach employs sentence transformers, a deep learning model to estimate semantic text similarity between text documents and predefined construct operationalizations. We demonstrate an application of our artifact by coding and scoring a sample of corporate 10-K reports for two types of organizational innovation processes: exploration and exploitation. Our artifact is a significant methodological contribution beyond manual coding of text documents, which cannot scale up to thousands or millions of documents, or word-based automated coding, which cannot capture the semantic meaning of text and cannot score text documents. Our approach offers new possibilities in mixed-mode research by integrating qualitative and quantitative methods using design science research.

Comments

20-Methods

Share

COinS
 
Dec 12th, 12:00 AM

Automated Coding and Scoring of Text: Artifact Design, Application, and Evaluation

This paper presents an artifact to automate and quantify qualitative coding of text documents on predefined constructs of interest. The quantified codes, similar to Likert-scaled measures in survey research, can be analyzed using statistical models. Our approach employs sentence transformers, a deep learning model to estimate semantic text similarity between text documents and predefined construct operationalizations. We demonstrate an application of our artifact by coding and scoring a sample of corporate 10-K reports for two types of organizational innovation processes: exploration and exploitation. Our artifact is a significant methodological contribution beyond manual coding of text documents, which cannot scale up to thousands or millions of documents, or word-based automated coding, which cannot capture the semantic meaning of text and cannot score text documents. Our approach offers new possibilities in mixed-mode research by integrating qualitative and quantitative methods using design science research.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.