Unstructured qualitative data for an agent-based model development.

Authors: Rajiv Paudel*, Michigan State University
Topics: Coupled Human and Natural Systems, Social Geography, Hazards and Vulnerability
Keywords: Agent-based Models, Qualitative data,Text Analysis
Session Type: Paper
Day: 4/5/2019
Start / End Time: 9:55 AM / 11:35 AM
Room: Truman, Marriott, Mezzanine Level
Presentation File: No File Uploaded

Qualitative data is extremely beneficial for model development. Model conceptualization needs system theories, and qualitative information about the systems can help in formulating such theories and establishing necessary assumptions. However, qualitative data processing and information extraction are highly time and resource consuming tasks. Data processing involves manual transcribing and coding and usually requires a number of coders to minimize potential subjective bias. This study aims to develop a methodology for expediting data processing and information extraction (for model conceptualization). The study uses a semi-automated information extraction which, in principle, requires less human involvement (work hours). It also minimizes interpretation bias due to human preconceptions about the system. For this, the study uses Natural Language Processing (NLP), an emerging computational technique for analyzing textual data, to extract specific information necessary to formulate systems theories for model conceptualization. NLP involves loosely-supervised processes for information extraction that are less labor and time intensive. The extracted information is then transposed to a conceptual model. The study utilizes narratives on household food security issues collected through field interviews in rural Southern Mali.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login