Building Textual Fuzzy Interpretive Structural Modeling to Analyze Complex Problems

Author/Creator ORCID




Information Systems


Information Systems

Citation of Original Publication


Distribution Rights granted to UMBC by the author.
This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see or contact Special Collections at speccoll(at)


Organizations regularly make decisions on complex issues with multiple variables or parameters. Hence, there is a strong motivation for the organization's Management to formalize and adopt a scientific approach to their decision-making based on their ground truth data. From a scientific point of view, each variable or parameter that influences a complex issue must be analyzed independently and within the networks of other parameters. This methodology enables us to understand each parameter's role and the potential for affecting or changing the whole system. However, with the complexity of issues and the deluge of Big Data to be analyzed, it is not feasible to apply several different approaches before making a timely and informed decision. We have developed a novel methodology, called Textual Fuzzy Interpretive Structural Modeling (TFISM), that helps us understand the variables related to complex issues and the connection between these variables. TFISM identifies vital terms or influential factors in a textual dataset, prioritizes the power of each factor, and then determines the associations and hierarchies between these factors. This computational social science methodology enhances Interpretive Structural Modeling (ISM) approaches to allow the input to be textual data. TFISM is multi-disciplinary and integrates ISM with Artificial Intelligence, text extraction, and information retrieval techniques. It is a domain-free methodology that can assist in complex decision-making, and we have applied this methodology to different datasets from social media and academic articles. Three separate domains were analyzed and validated with the technique during this research study.