Web Ranking Algorithm Rewired to Weigh Expert Trust in Fuzzy Group Decisions

Every major decision, from which battery powers the next generation of electric vehicles to which medical protocol a hospital adopts, ultimately rests on a group of experts trying to agree. The trouble is that human judgment is messy. Experts are inconsistent, their confidence wavers, and their opinions are shaped by who they trust. A new study published in the International Journal of Machine Learning and Cybernetics tackles this problem head-on, borrowing a page from web search technology to make group decision-making more rigorous, more transparent, and, for the first time, measurably reliable.

The research, led by Yan Chen with Lei Xu and Lin Liu, all of Shenyang University of Technology in China, introduces a group decision-making framework built on picture fuzzy sets, a mathematical structure designed to capture the full spectrum of human opinion. Where classical logic demands that a statement be simply true or false, and even earlier fuzzy extensions allowed only membership and non-membership degrees, picture fuzzy sets add a third dimension: the refusal or neutral degree. An expert evaluating a candidate technology can simultaneously express support, opposition, and abstention, with the three values summing to one. This richer vocabulary matters in real committees, where ‘I am not sure’ is often the most honest answer on the ballot.

Yet the very richness that makes picture fuzzy sets expressive has historically made them unwieldy. When experts record their pairwise comparisons in picture fuzzy preference relations, checking whether those judgments are internally consistent becomes a computational ordeal. Consistency is not a pedantic concern. If an expert prefers option A over B, B over C, but then declares C preferable to A, the entire ranking built on those judgments is suspect. Traditional verification procedures for picture fuzzy matrices have been cumbersome enough to discourage practical adoption, and the new paper identifies this as one of three chronic defects in the existing literature, alongside the absence of an objective basis for weighting experts and the difficulty of quantifying how trustworthy the final decision actually is.

The first breakthrough is mathematical housekeeping of an unusually elegant kind. The authors deploy transformation functions that convert picture fuzzy preference relations into ordinary fuzzy complementary judgment matrices, the well-studied workhorses of classical decision theory. This conversion simplifies the data structure and dramatically reduces computational complexity, but the real innovation lies in what follows. By constructing a derived matrix from the transformed judgments, the team proves an additive consistency determination theorem, a criterion that allows researchers to verify consistency directly rather than iterating through the laborious checks that traditional methods require. In effect, a question that once demanded an elaborate auditing procedure can now be settled by inspecting a single algebraic condition.

The second pillar of the framework addresses a blind spot that has long troubled the field: how much should each expert’s opinion count? Conventional approaches often assign weights arbitrarily or rely on self-reported confidence, ignoring the social fabric in which real experts are embedded. Chen and colleagues turn to BrowseRank, an algorithm originally developed for ranking web pages by modeling user browsing behavior on a network of hyperlinks. In the decision-making context, the experts become nodes in a social network, and trust relationships between them become the links. Because real-world trust data is frequently incomplete, the authors first introduce a picture fuzzy trust propagation operator, which infers missing trust links from the chains of relationships that do exist, much as one might gauge a stranger’s reliability through a friend of a friend.

Once the trust network is complete, BrowseRank computes each decision-maker’s objective weight by integrating multiple network parameters, breaking free of the limitation of ignoring social attributes entirely. An expert who is widely trusted by other trusted experts accrues greater influence, not because of seniority or loudness in the meeting room, but because of their structural position in the web of professional confidence. This is a genuinely different philosophy of weighting. It treats expertise as a network property rather than an individual trait, echoing ideas from citation analysis and social network theory while grounding them in the fuzzy mathematics needed to handle uncertain trust assessments.

Aggregating individual opinions into a collective verdict is where many decision frameworks quietly fail. Even if every expert’s matrix is individually consistent, the merged matrix can drift into inconsistency, contaminating the final ranking. The proposed method guards against this with an aggregation strategy that combines the weighted geometric average operator with the Hadamard product, an element-wise multiplication of matrices. The authors show that this pairing ensures the consistency of the aggregated matrix, filling what they describe as a gap in the research literature. The practical consequence is that the group’s collective judgment inherits the logical soundness of its inputs, rather than requiring a fresh round of correction after aggregation.

Perhaps the most novel contribution is the framework’s answer to a question that decision scientists rarely ask with precision: how reliable is the result? The authors define decision-maker reliability using a picture fuzzy set similarity measure, comparing each expert’s judgments against the group consensus. Experts whose views align closely with the collective position score high reliability; outliers score low. This converts reliability from a vague qualitative impression into a quantitative evaluation, completing an evaluation system in which every component, from individual consistency to expert weight to final confidence, is computed rather than assumed. For organizations making high-stakes choices, such a number could become as standard a deliverable as the recommendation itself.

To demonstrate the framework in action, the team applied it to a problem with obvious contemporary urgency: selecting batteries for new energy vehicles. Battery choice is an ideal stress test for fuzzy decision theory because the criteria, energy density, cost, safety, lifespan, and environmental impact, are riddled with uncertainty, and the experts who assess them come from heterogeneous backgrounds with uneven trust relationships. Through simulation and comparative analysis against existing methods, the authors verified their approach’s advantages in consistency check efficiency, weight rationality, and result reliability. The comparisons showed that the BrowseRank-based weighting produced more defensible expert influence scores than methods lacking a social network foundation, while the additive consistency theorem streamlined a verification step that competing approaches handle far more slowly.

The broader significance of the work extends well beyond battery procurement. Group decision-making under uncertainty underpins technology selection, policy evaluation, medical diagnosis support, disaster risk analysis, and financial investment screening, all areas where picture fuzzy methods have recently gained traction. By fusing a web-ranking algorithm with judgment matrix consistency theory, the Shenyang team has built a pipeline in which fuzzy human opinions are tamed algebraically, social trust is measured rather than guessed, and the final answer arrives with a quantified confidence estimate. As decisions grow more complex and the experts consulted more numerous and dispersed, frameworks of this kind may become the quiet mathematical machinery ensuring that collective judgment is not just a vote, but a verifiable measurement.

Subject of Research: A picture fuzzy group decision-making method integrating the BrowseRank algorithm and additive consistency for expert weighting and reliability evaluation

Article Title: A picture fuzzy group decision-making approach based on BrowseRank algorithm and additive consistency

Article References: Chen, Y., Xu, L., & Liu, L. (2026). A picture fuzzy group decision-making approach based on BrowseRank algorithm and additive consistency. International Journal of Machine Learning and Cybernetics, 17(9), Article 452. https://doi.org/10.1007/s13042-026-03279-y

Image Credits: AI Generated

DOI: 10.1007/s13042-026-03279-y

Keywords: group decision-making, picture fuzzy sets, BrowseRank algorithm, additive consistency, trust propagation, social network analysis, fuzzy preference relations, expert weighting, decision reliability, new energy vehicle battery selection, aggregation operators, similarity measure