Fig 1.
The backbone of the Constitutional legal network (CLN).
Each node indicates a Constitutional article, and links are the co-cited relations regarding same decisions. Node sizes are scaled according to strength (Eq 7) and the thickness of each link denotes the number of shared citations between two articles. All articles (nodes) are included in one of 12 categories of the Constitution. Thus, we present the information as colors of different nodes. Since the original CLN is too dense for direct presentation, we use the backbone extraction algorithm [41] for better visualization. One could find the central core articles in the CLN such as “Equal rights”, “Dignity and pursuit of happiness” and “Protection of freedom and rights.”
Fig 2.
Topological properties of the CLN.
(A) The cumulative strength distribution Pc(s) of the CLN (solid red) compared with the null model counterparts (dotted line). (B) The scaling relation between degrees (k) and strengths (s) for each node (red dots) compared with the null model (gray stars). The dashed (slope of 1.5) and dotted (slope of 1.0) lines are presented as guides. All data is binned for each degree k. (C) A box plot of strengths of articles in each category in the CLN. Significant outliers of high strength (“Equal rights”, “Protection of freedom and rights”, and “Dignity and pursuit of happiness”) are identified as texts. (D) The z-score observation of strengths for each category with 104 samples of null model type I (see Materials and methods). (E) The chord diagram, which presents the net flow between categories of the CLN. (F) The weighted rich-club coefficient ϕw(s) of the CLN and (inset) the normalized one ρw(s) with 104 samples of null model type II (see Materials and methods).
Fig 3.
Analysis of the CLN for 6 different regimes and the core structure.
(A) Shared link weights of the “Rights and Duties of Citizens” category for 6 regimes. The solid red area denotes the ratio of intra-relations within the category and the checked area is inter-relations with the “Right and Duties of Citizens” and the other categories. (B) Z-score observation, and (C) the normalized rich-club coefficient show a similar statistical pattern for all regimes. (D) The identified core structure of all regimes. Shaded areas denote the results of the s-core decomposition for different regimes.
Fig 4.
Representative articles for each regime with visualized text groups.
Font sizes are scaled according to their nodes’ prevalence measures in the given regime, and the text position is adjusted by the algorithm to produce word clouds. The text is colored according to the categories in which it is included. One would find the keywords representing each regime’s identity.
Fig 5.
The distinctive properties of each regime.
(A) The core structure of each regime’s network identified by link prevalence. Here we present two adjacent regimes (regimes 4 and 5) with markedly different structures within their networks. This indicates that the CLN of each regime would be influenced by the social circumstnaces and identity of the regime. (B) The correlation matrix among the 6 different regimes is depicted by a heat map. We found that some adjacent regimes formed strong clusters, especially regimes 5 and 6, which originiated from the same political parties and share similar government principles. (C) The flow of rank-ordering according to changes in regime from 1 to 6. Each rank is based on each article’s strength in the given regime. We found that the top-rank articles were maintained, but there was a large amount of fluctuation in the middle-ranking articles.