The Influence of Geometric Configurations: Exploring Hexagonal and Triangular Cycles in Social Graphs
Presented at the 2nd International Conference on Advances in Data-driven Computing and Intelligent Systems (ADCIS 2023)
One of the favourite shapes of architects and engineers are triangles. Truss Bridges, Pyraminds of Giza, bermuda traingle all symbolizes triangle's strength. It has not been proved that triangles are strongest shape. However, there is another slightly unpopular opinion that says "Hexagon is the strongest shape" They distribute force evenly and they have tessellate structure. Honeycombs, snowflakes and several organic compounds are the examples. Hence, our problem statement is as follows: "Which is the strongest shape TRIANGLES or HEXAGONS?" For further experimental validation, we use the result of the above statement to perform and improve accuracy for community detection and prediction.
The study of triangles and hexagons in community detection also raises questions about the nature of network structure and how it influences the formation and identification of communities. Comparing the strength of two different shapes (triangles and hexagons) in the context of community detection, is a relatively new and complex area of study in network analysis. This problem statement brings attention to the significance of individual shapes and their impact on network behavior. This approach suggests that the strength of specific shapes in a network could inform and improve the effectiveness of community detection and prediction. This combination of theoretical analysis and practical application makes this problem statement unique in the field of network analysis.
We confidently claim, based on our thorough analysis, that hexagonal and triangular cycles have a significant impact on predicting social graph outputs. We observed that hexagonal cycles have a greater impact and consequently, produced a higher accuracy as compared to that of triangular cycles. Four distinct tasks—Strength analysis, Node classification, Graph classification, and Motif analysis—are carried out to support this claim.