Filtering Risks using Bipolar Fuzzy Nominal Classification

Managing risks for large-scale complex systems requires identifying, assessing, and prioritizing different risk scenarios or undesirable events that systems may face; coping with risks when engaged in some activities or decision making processes such as deciding to invest in a company or a country, to offer loans to applicants, to recruit a candidate to fill vacant position, to fund projects or infrastructures, etc. necessitate to categorize the targeted entities. Prioritizing or filtering risks (undesirable events or events scenarios) or categorizing risky entities for some activities return to assigning them to a predefined classes or categories for their appropriate treatment. On other hand a scenario or an entity as well as a class or category will be characterized by some attributes that represent basically its impact or constraints on different objectives, issues, stakes, or consequences that do represent some interests for decision makers and/or stakeholders. This is a nominal classification problem that is a subfield of multi-criteria decision making problems in general. As the attributes and/or features characterizing classes may not be defined precisely, a fuzzy representation can be used to treat this uncertainty leading to what is known as fuzzy nominal classification. If one was to classify using only one feature there will be some values of that feature that leads to certain choice or certain rejection of a particulate class and those for which some hesitation or doubt will exist leading to some bipolarity. In this communication, a bipolar fuzzy nominal classification framework will be built to address risks filtering and/or categorization issues; a real world application in the domain of countries' risk classification show the potentiality of the derived approach.