FDOSM (Decision by Opinion Score Method)

FDOSM stands for "Fuzzy Decision by Opinion Score Method." Salih, Zaidan and Zaidan (2020) proposed it, and since then it has built a wide family of applications, particularly in technology and system benchmarking studies, health-monitoring systems, network-security protocols and biometric-recognition systems (Muhsen et al., 2023). The method's core idea is that an alternative's performance on the criteria is expressed not through a direct physical measurement but through an "opinion score" (an assessment an expert or decision-maker gives relative to the ideal). The manifest DecisionMind implements runs the method's single-decision-maker, crisp (non-fuzzy) base version; its output is a score for every alternative and a rank produced by that score.

In some decisions a direct physical measurement is not possible. Questions such as "how good is this software's usability" or "how high is this system's reliability" cannot be measured with a ruler; an expert or user expresses this with a score. FDOSM takes this score as an answer already given to the question "how does this compare with the ideal," and does not build a new ideal point on top of it; it merely brings these scores onto a common scale and combines them with weights. In the method's fuzzy version, when several experts give different opinions, these opinions are expressed as triangular fuzzy numbers (a lower, middle and upper value), averaged, and reduced to a single crisp number (defuzzification); in the crisp version DecisionMind runs in this manifest, the input is already a single expert's crisp score, so these steps act as a passthrough.

This idea has one consequence. FDOSM is compensatory and sits close to the weighted-sum family of methods, but its emphasis differs. Where TOPSIS assesses an alternative's raw measurement by its distance to ideal and anti-ideal points, FDOSM assumes the input is already an assessment (an opinion) and ratios this assessment against the best value.

The method consists of six steps; in DecisionMind's crisp-data version, the first three steps act as a passthrough because the input is already a single crisp number.

First, converting linguistic expressions to triangular fuzzy numbers. If several experts have given linguistic assessments such as "good," "medium" or "poor," these are converted into triangular fuzzy numbers made of a lower, middle and upper value. This step is skipped in the crisp-data version.

Second, combining decision-makers' opinions. If more than one expert exists, their opinions are averaged. This step is skipped in the single-decision-maker crisp version.

Third, defuzzification. If a fuzzy value exists, it is reduced to a single crisp number. This step is skipped in the crisp version.

Fourth, linear-maximum normalisation. Every cell is ratioed against the best value in its own column: for benefit criteria the cell value is divided by the column's largest value, for cost criteria the column's smallest value is divided by the cell value.

Fifth, the weighted opinion score. The normalised values are multiplied by the criterion weights and summed.

Sixth, ranking. Alternatives are ranked by this score from highest to lowest.

The formulas behind each step, the intermediate tables and citation formats are given on the DecisionMind method page; this card carries no formulas.

The opinion score is the weighted average of how close an alternative sits to the best value on every criterion. Unlike TOPSIS's closeness score, this score never looks at the worst value (the anti-ideal); it only ratios against the best. A score close to 1 means "reaches almost the best value on every criterion"; a low score shows the alternative falls short of the best value on one or more criteria. This score assumes the input is already an expert assessment (an opinion); if the input is a raw physical measurement, the method normalises it as given and does not itself question how reliable an "opinion" the measurement carries.

Thus instead of writing:

"System 2's score is the highest, so it is objectively the best system"

the report should read:

"Given the opinion scores and weights supplied, System 2 sits at the highest ratio to the best value; who gave these scores and how must be stated in the report"

Crisp data. In DecisionMind, this manifest runs the method's single-decision-maker base version, which requires no defuzzification, and currently has no extension. The original FDOSM in the literature carries the name "fuzzy" because it combines several experts' triangular fuzzy opinions; this extended multi-expert version is not covered by DecisionMind's manifest. You need an opinion score (or a direct measurement) for every alternative on every criterion, criterion directions, and weights summing to 1. FDOSM does not produce weights, it takes them from outside.

If the alternatives being compared are already scored against an ideal reference, FDOSM is a natural choice, particularly in technology or system benchmarking studies; the literature has used it frequently in health-monitoring systems, network-security protocols and similar technical comparisons (Muhsen et al., 2023).

There are two situations where it should not be used. If your data is a raw physical measurement and jointly assessing distance to the ideal and the anti-ideal (as TOPSIS does) would be more informative, FDOSM's structure of ratioing only against the best falls short. If you need to combine several experts' disagreement in a fuzzy way, DecisionMind's version, being limited to a single decision-maker with crisp data, is not sufficient.

Input is already an expert opinion score, single decision-maker, crisp data → FDOSM

Several experts disagree significantly, fuzzy combination is needed → FDOSM's multi-expert fuzzy version (not covered by this manifest)

Input is a raw measurement, both the ideal and the anti-ideal must be considered together → TOPSIS

Weights, not a ranking, are needed → AHP, BWM, SWARA (subjective); Entropy, CRITIC (objective)

FDOSM's greatest strength is its simplicity; its computational burden is small, and because it sits close to the weighted-sum family, the result is easy to interpret. The opinion-score philosophy offers a natural framework for comparisons that rest on expert assessment where pure physical measurement is not possible. Since Salih, Zaidan and Zaidan's (2020) proposal, the method has gained a wide family of applications, showing that it offers a skeleton adaptable to different fuzzy environments (Pythagorean fuzzy, neutrosophic, dual-hesitant fuzzy, and so on) (Alamoodi et al., 2022).

Its limitations stem from its structure. First, the method ratios only against the best value and never uses the worst value (the anti-ideal); it does not offer the "both close to the good and far from the bad" balance that TOPSIS provides. Second, DecisionMind's version is limited to a single decision-maker with crisp data; the method's actual "fuzzy" contribution, the multi-expert opinion-combining and defuzzification steps, acts here as a passthrough and is not active. Third, Muhsen and colleagues' (2023) systematic review shows that consistency problems exist across the literature in how opinion scores are gathered within the FDOSM family of studies. Fourth, the quality of the opinion scores lies outside the method; a flawless calculation built on poor or inconsistent scores still gives a weak ranking.

The most common mistake is presenting the opinion score as a raw measurement without reporting who gave it and how. A second mistake is assuming that because the method's name contains "fuzzy," it automatically performs a multi-expert calculation; DecisionMind's manifest works with crisp, single-expert data. A third mistake is confusing the FDOSM score with TOPSIS's closeness score and assuming it also accounted for distance to the anti-ideal; FDOSM only ratios against the best. A fourth mistake is placing opinion scores given on different scales (one on a 1–5 scale, another on a 1–10 scale) into the same table without correction.

The governing principle is this:

An FDOSM result is the weighted ratio of the opinion scores supplied as input against the best value; if the report does not state how and by whom these scores were given, what the result is measuring remains unclear.