China’s Scenic Attractions See a Tourism Ranking Breakthrough With Dynamic Clustering and Borda Count Technology

China’s scenic attractions undergo a new phase of dynamic clustering and Borda Count in their tourism rankings as a way to improve accuracy.

China’s scenic attractions are entering a new phase of tourism ranking technology as researchers develop a data-driven framework to deal with the increasingly complex information that traveler evaluations generate. The method, published on 21 September 2026, was tested on China’s prestigious 5A scenic attractions and was found to produce rankings that showed a high degree of consistency with existing tourism assessments. The breakthrough is important because modern tourism platforms must process large numbers of overlapping and sometimes conflicting evaluation factors before being able to recommend destinations reliably.

China’s Scenic Attractions Face a Growing Data Ranking Challenge

Choosing a scenic destination may seem straightforward to travellers, but the technology behind tourism recommendation systems faces a far more complicated problem. Visitors can evaluate an attraction according to scenery, cleanliness, accessibility, facilities, service standards, convenience, value and many other factors. As thousands of individual assessments accumulate, the resulting dataset becomes extremely large and difficult to interpret.

The challenge becomes greater because many evaluation factors overlap. Two different criteria may measure almost the same part of the visitor experience, while other criteria may produce conflicting results. If all these factors are placed directly into a single ranking system, repeated information can receive too much importance and contradictory data can weaken the overall result. The new framework addresses this problem by organising the information before producing the final destination ranking. The research specifically targets high-dimensional and heterogeneous tourism evaluation data, which conventional Multi-Attribute Decision-Making methods can struggle to process effectively.

Multi-Attribute Decision-Making Sits at the Heart of the Research

The ranking challenge centres on Multi-Attribute Decision-Making, commonly known as MADM. This type of analysis is used when several alternatives must be compared against multiple criteria at the same time.

In tourism, the alternatives are scenic attractions. The attributes may include visitor service, environmental conditions, accessibility, infrastructure, perceived value and overall experience. Traditional MADM systems can perform well when relatively few attributes are involved and when the data behave consistently. Large-scale traveller data create different conditions.

Hundreds of attributes can appear simultaneously. Some can be highly correlated. Others may be repetitive, noisy or inconsistent. The research therefore avoids treating every factor as an isolated indicator. Instead, it introduces a three-stage structure: cluster related attributes, rank destinations within each cluster and then combine those rankings into one overall result. This structure is designed to preserve useful information while reducing the distortions created by complex datasets.

Dynamic Clustering Organised Similar Tourism Evaluation Factors

The first major step is consistency-driven dynamic attribute clustering.

Rather than clustering the tourist destinations themselves, the method groups the evaluation attributes. Criteria that repeatedly produce similar ranking patterns across attractions are placed together.

For example, two visitor-evaluation factors may consistently place the same scenic destinations near the top and the same destinations near the bottom. The framework interprets that similarity as evidence that the two attributes may represent related dimensions of the tourism experience.

This process reduces the effective dimensionality of the problem without simply removing large quantities of data.

The word dynamic is particularly important. The structure is not based only on permanently predetermined groups. The clustering process responds to the relationships found within the tourism data. Attributes are organised according to the consistency of the ranking patterns they produce.

That means the system first identifies coherent groups of tourism information before asking which attraction performs best. As a result, overlapping signals are less likely to dominate the final result merely because they appear repeatedly in the dataset.

TOPSIS Creates Separate Rankings Inside Every Cluster

After the tourism attributes have been organised, the second stage uses TOPSIS, short for the Technique for Order Preference by Similarity to Ideal Solution.

TOPSIS evaluates alternatives by considering their distance from an ideal positive solution and an ideal negative solution. In simple terms, a stronger scenic attraction should be closer to the best possible combination of conditions and further away from the weakest possible combination.

The important difference in the new framework is that TOPSIS is not immediately applied to every tourism attribute together.

Instead, it is used separately within each cluster.

Each cluster therefore creates its own local ranking of scenic attractions. One group of evaluation criteria may favour certain destinations, while another group may produce a different order.

This allows different dimensions of the visitor experience to retain their own influence.

Rather than forcing environmental quality, accessibility, visitor services, facilities and other potentially conflicting factors into a single calculation at the beginning, each coherent family of indicators can produce an independent ranking first.

Borda Count Turns Multiple Rankings Into One Tourism Verdict

The third stage introduces the Borda Count, which is used to combine the different local rankings into one global tourism ranking.

The concept is relatively easy to understand.

A scenic attraction receives points according to where it appears in each local ranking. Destinations consistently appearing near the top across several clusters accumulate stronger scores, while attractions performing poorly across several groups receive fewer points.

The scores are then combined.

The attraction with the strongest accumulated performance can take the highest position in the overall ranking.

This approach provides an important advantage. One unusual cluster is less likely to determine the entire outcome because the final result reflects performance across several independent groups of evaluation criteria.

The Borda Count therefore acts as a ranking-fusion mechanism, bringing different tourism perspectives together only after each has been assessed separately.

The complete process can be viewed as three connected stages:

Dynamic clustering organised related tourism attributes.

TOPSIS produces a destination ranking within each group.

Borda Count combines the individual rankings into a final result.

This structure forms the central breakthrough of the research.

China’s Prestigious 5A Scenic Attractions Provide the Real-World Test

The framework was tested using evaluation information linked to China’s 5A scenic attractions.

The choice is significant because 5A represents the highest level within China’s five-level scenic-attraction classification structure. Official tourism information identifies the levels as AAAAA, AAAA, AAA, AA and A, with 5A or AAAAA representing the highest classification.

These attractions are therefore an important testing ground for a new tourism ranking system. The destinations have already reached the highest formal classification, meaning the research was not simply attempting to separate weak attractions from strong ones. It was examining whether complex traveller-generated information could create meaningful distinctions among destinations already recognised at the highest level.

The empirical results showed a high degree of consistency between the new framework and existing tourism ratings. That finding suggests the model is capable of detecting important destination-quality signals despite working with complex, high-dimensional visitor evaluation information.

The importance of 5A attractions to China’s visitor economy can also be seen in recent official tourism figures. During the 2026 Spring Festival holiday, 12 monitored 5A scenic attractions in Hunan recorded about 4.279 million visits, representing a 60.82 per cent year-on-year increase, according to provincial government tourism data.

Why the Model Groups Attributes Instead of Destinations

One of the most distinctive elements of the framework is its decision to cluster evaluation attributes rather than scenic attractions.

This matters because the relationship between evaluation factors contains valuable information.

Travellers can experience the same attraction in very different ways. One visitor may care heavily about accessibility. Another may focus on scenery. Others may respond to cleanliness, facilities or service quality.

Some of these factors frequently move together. Others behave independently.

By grouping criteria with similar ranking patterns, the system can identify broader families of visitor experience without assuming that every tourism factor is unrelated.

Each cluster effectively becomes a separate perspective from which the destination can be judged.

TOPSIS then measures destination performance within that perspective, while the Borda Count brings those different judgments together.

The system therefore accommodates disagreement instead of attempting to eliminate it immediately.

New Framework Could Strengthen Tourism Recommendation Technology

The implications go beyond academic ranking exercises.

Travel recommendation systems increasingly depend on massive volumes of traveller-generated data. More information can potentially improve recommendations, but only when that information is processed correctly.

Duplicated indicators can exaggerate particular aspects of a destination. Conflicting ratings can weaken confidence in results. Large numbers of attributes can also make traditional decision models harder to manage.

The cluster-rank-fuse framework offers one possible solution.

Tourism platforms could potentially apply similar systems to create more dependable recommendations from large volumes of visitor information. Destination authorities could also use the approach to identify which groups of factors are driving changes in the perceived quality of an attraction.

A destination could remain strong overall while weakening in one specific group of visitor-experience indicators. Another could climb because accessibility, facilities or service performance has improved.

Analysing clusters individually could reveal those movements more clearly than relying only on one overall score.

High-Dimensional Tourism Data Could Become More Useful

The wider importance of the research lies in the rapid growth of digital tourism information.

Online reviews and ratings now create an enormous pool of traveller judgement. Yet collecting more information does not automatically produce better destination decisions.

The challenge is increasingly about how tourism data are organised, weighted and combined.

The research demonstrates that complex evaluation information does not necessarily need to be compressed immediately into one giant calculation. Related criteria can first be separated into meaningful groups. Attractions can then be ranked independently within those groups before a final ranking is generated.

This allows disagreement within tourism data to remain visible while still producing a usable overall result.

The study also acknowledges restrictions surrounding its original dataset. Publicly accessible traveller-evaluation information was used, but the complete raw dataset cannot be freely redistributed because third-party terms and intellectual-property restrictions apply. The published research instead documents the data structure, prepossessing, clustering process, parameter settings and experimental design to support understanding and reproducibility.

China’s Tourism Ranking Technology Moves Towards a Smarter Data Era

The research ultimately highlights a larger transformation taking place across tourism technology.

Destination ranking is moving beyond simple star scores or single averages. Modern systems must understand relationships between hundreds of factors, identify repeated signals, manage conflicting opinions and still produce recommendations that travellers can understand.

By combining dynamic clustering, TOPSIS and Borda Count, the new framework provides a structured way of tackling that challenge.

For China’s scenic attractions, the approach demonstrates how visitor-generated data can be organised into more meaningful ranking signals. For tourism authorities, it could offer a clearer view of destination performance. For digital travel systems, it points towards recommendation technology capable of processing increasingly complex traveller information without allowing duplicated or conflicting attributes to overwhelm the result.

China’s scenic attractions are seeing a tourism ranking breakthrough as dynamic clustering organised complex visitor data and Borda Count combines multiple results into a clearer final ranking. The approach improves reliability by reducing data noise, overlap and conflicting evaluation signals.

As online tourism data continue to expand, the breakthrough suggests the next generation of destination rankings may depend not simply on collecting more traveller opinions, but on becoming significantly smarter about how those opinions are counted.