Table 1.
Basic notations and definitions.
Table 2.
Bibliographic profiles of datasets retrieved using CCE method for the selected Nobel Prize cases.
Fig 1.
The boundary-spanning mechanism exhibited by Gurdon’s landmark publication in 1962.
Table 3.
The SVA scores of top-6 most cited papers in the network (1957-1962). SVA score ranges: ΔM:0.0–64.69, CL:-15.00–145.71, CKL:0.0–0.58, H:0.0–1.72, α:0.0–0.5, β:0.0–1.0, E: 0.0–1.01. The subscript next to each SVA score of the seed paper indicates the paper’s relative rank according to that metric, sorted in a descending order.
Fig 2.
Case 1, Seed 3 (S1,3) and Seed 4 (S1,4).
The co-citation network surrounding the publication of S1,3 and S1,4 (2001–2007). The purple and red dash lines are co-citation links made by one of the Yamanaka’s competitors [65]. The numbered cluster labels were generated using CiteSpace’s implementation of Latent Semantic Indexing.
Fig 3.
Novel co-citation links introduced by S1,3 and S1,4.
(A) depicts novel links introduced by S1,3 [51], and (B) shows novel links added by S1,4 [52]. Note that the underlying network is identical to that is shown in Fig 2.
Table 4.
The SVA scores of top-10 most cited papers in the network (2001-2007). SVA score ranges: The ranges of the obtained SVA metric scores are: ΔM:0.0–89.36, CL:-76.24–494.69, CKL:0.0–0.87, H:0.0–2.54, α:0.0–1.0, β:0.0–1.0, E: 0.0–2.10. The subscript next to an SVA score of each seed paper indicates the paper’s relative rank according to that metric, sorted in a descending order.
Fig 4.
The boundary-spanning mechanism exhibited by O’Keefe’s landmark publication in 1971.
Table 5.
The SVA scores of top-10 most cited papers in the network (1966-1971). SVA score ranges: ΔM:0.0–75.41, CL:-48.05–0.00, CKL:0.0–0.54, H:0.0–1.58, α:0.0–1.0, β:0.0–1.0, E: 0.0–1.3. The subscript next to an SVA score of each seed paper indicates the paper’s relative rank according to that metric, sorted in a descending order.
Table 6.
The SVA scores of top-10 citing papers in the network (2000-2005). SVA score ranges: ΔM:0.0–62.54, CL:-93.62–54.84, CKL:0.0–0.25, H:0.0–0.67, α:0.0–1.0, β:0.0–1.0, E: 0.0–1.43. The subscript next to an SVA score of each seed paper indicates the paper’s relative rank according to that metric, sorted in a descending order.
Fig 5.
The co-citation network surrounding S2,2 (2000–2005). The clusters were labeled with cited publications’ keywords. (A) shows the underlying network. (B) illustrates the novel links introduced by S2,2.
Table 7.
The SVA scores of the top-10 citing papers in the network (1987-1993). SVA score ranges: ΔM:0.0–80.62, CL:0.0–49.42, CKL:0.0–0.27, H:0.0–0.70, α:0.0–1.0, β:0.0–1.0, E: 0.0–1.55. The subscript next to an SVA score of each seed paper indicates the paper’s relative rank according to that metric, sorted in a descending order.
Fig 6.
The co-citation network surrounding Ohsumi’s breakthrough papers (1987–1993), showing novel links added by S3,1 [58]. Cluster labels were generated with log-likelihood ratio and the node sizes correspond to the degree of betweenness centrality of a cited reference.
Fig 7.
Comparing the structural variations induced by Ohsumi’s papers against the most cited paper in our SVA result set [67] (see Table 7). (A) outlines the novel links induced by S3,2 [59]. (B) compares the novel links induced by [67]. The underlying network is identical to that in Fig 6.
Table 8.
Summary performance of SVA metrics for Cases 1–3.
The pseudopaper strategy amplifies SVA signals that are otherwise hard to detect from the original seed papers. Ps(s1 ⊕ s2) denotes a pseudopaper generated from seed papers s1 and s2. Where applied, ρs represents a collection of non-seed paper(s) published by the Nobel laureates of s in the same year as seed paper s’s publication.
Table 9.
SVA metric scores of the top-10 most cited papers in the network containing Ps(S3,1 ⊕ S3,2)) (denoted by ⋆) with scaling factor 25 (k = 25; 1988-1993). A total of 208 citing papers made novel co-citation links. SVA score ranges: ΔM:0.0–80.09 | CL:0.0–66.06 | CKL:0.0–0.08 | H:0.0–0.23 | α:0.0–1.0 | β:0.0–1.0 | E: 0.0–1.68.
Table 10.
The effects of varying scaling factor k on detecting structural variation signals from pseudopaper Ps(S3,1 ⊕ S3,2) are demonstrated for Case 3. The numbers in parentheses indicate to the pseudopaper’s ranks by the corresponding metrics.
Fig 8.
Visualizing the effects of increased scaling factors.
Increasing the scaling factor of co-citation network in Case 3 (1988–1993) enriches the intellectual representation of the network and broadens the coverage of a pseudopaper’s novel links. (A) depicts the novel links added by Ps(S3,1 ⊕ S3,2) (k = 5). (B) shows the novel links added by the same pseudopaper for k = 25.