Fig 1.
To initiate the workflow, two input taxonomies (T1, T2) are supplied jointly with a set of concept articulations (A) and taxonomic constraints (C). The workflow facilitates an iterative alignment process aimed at rendering the input logically consistent and sufficiently expressive. Negation of either criterion (red arrows) leads to input modulation through either diagnosis/removal of conflicting constraints (no possible worlds; right loop) or exploration of many possible worlds and provision of additional constraints (left loop). The well-specified alignment (green arrows) is output as a set of MIR (maximally informative relations; both immediately deducible and inferred [ded./inf.]), and visualized either as a containment (with overlap [><]) or merge concept graph (see fig. 2).
Fig 2.
Abstract toolkit input and output example.
(A) Input taxonomy T1, with nine concepts named (1.) A-I. (B) Input Taxonomy T2, with eight concepts named (2.) A-I. Concept 2.CD is congruent with (1.C + 2.D). The respective child concepts 1.E/2.E and 1.G/2.G are non-congruently assigned to parent concepts 1.B/2.B and 1.F/2.F. (C) Representation of T1 and T2 and articulations in the toolkit input file (see also S1 and S2 Dataset). (D) Toolkit input visualization, showing both hierarchical (intra-taxonomic; is_a) and lateral (inter-taxonomic; RCC-5) articulations. (E) Single, consistent alignment of the input shown as a containment with overlap graph (legend to the right of [D]). (F) Merge concept analysis of the input, resolving Euler regions that result from overlapping concepts. Annotation convention: 1.B\2.B = the region of the 1.B/2.B overlap which is unique to 1.B; 1.B*2.B = the region which each concept shares; and 2.B\1.B = the region which is unique to 2.B.
Fig 3.
Input taxonomies for the Perelleschus use case.
All 54 concepts are uniquely identified, and labeled with either traditional (ranked) or informal (clade) names. Concept label abbreviations are provided in square brackets and are used throughout this analysis.
Fig 4.
Alignment 1—Günther (1936—T1) and Voss (1954—T2), ostensive reading [OST].
Input articulations are shown on the left; only the articulation 1954.PER > 1936.ELL is represented (as inferred) in the output MIR (maximally informative relations). * = sufficient input articulations—this annotation convention (*) is used in all subsequent figures where sufficiency is obtained with a subset of the input articulations. The containment with overlap and merge concept graphs are identical.
Fig 5.
Alignment 1—Günther (1936—T1) and Voss (1954—T2), intensional/ostensive reading [INT/OST].
Modifications of the input alignment in comparison to that of fig. 4 are shown in bold font. Other conventions as specified in Figs. 2–4. The top-level articulation 1954.PER >< 1936.ELL is provided, and an implied child 1936.ELL_IC is introduced, where 1954.PER | 1936.ELL_IC. (A) Containment with overlap graph. (B) Merge concept graph.
Fig 6.
Alignment 1—Günther (1936—T1) and Voss (1954—T2), intensional reading [INT].
The top-level articulations is asserted as 1954.PER < 1936.ELL and an implied child 1936.ELL_IC is introduced. (A) Containment with overlap graph. (B) Merge concept graph.
Fig 7.
Alignment 1—Günther (1936—T1) and Voss (1954—T2), intensional/ostensive reading [INT/OST]), underspecified input, level 1 (one articulation removed).
Input constraints as in fig. 5, yet without 1954.PER | 1936.ELL_IC, resulting in five disjoint articulations involving concepts 1936.ELL and 1936.ELL_IC in the inferred output MIR. (A-H) Containment with overlap graphs of eight consistent alignments (possible worlds).
Fig 8.
Alignment 1—Günther (1936—T1) and Voss (1954—T2), intensional/ostensive reading [INT/OST]), underspecified input, level 2 (two articulations removed).
Input configuration as in fig. 6, yet without 1954.PER >< 1936.ELL, resulting in six disjoint articulations in the inferred output MIR. (A-I) Containment with overlap graphs of nine additional alignments beyond those shown in fig. 7, resulting in a total of 17 possible worlds.
Fig 9.
Alignment 2—Voss (1954—T1) and Wibmer & O’Brien (1986—T2), intensional/ostensive reading [INT/OST].
Conventions as in fig. 4; the output MIR are provided in the Supporting Information S2.
Fig 10.
Alignment 3—Wibmer & O’Brien (1986—T1) and Franz & O’Brien (2001—T2), intensional/ostensive reading [INT/OST].
The implied child 2001.PHY_IC of parent 2001.PHY is the only input modification in comparison to [1]. (A) Containment with overlap graph. (B) Merge concept graph.
Fig 11.
Alignment 4—Franz & O’Brien (2001—T1) and Franz (2006—T2), ostensive reading [OST].
Modifications of the initial input given in [1] are shown in bold font. Non-coverage (nc) is asserted for parent concept 2006.PHY of child concept 2006.PHYsubcin. (A) Containment with overlap graph. (B) Merge concept graph.
Fig 12.
Alignment 4—Franz & O’Brien (2001—T1) and Franz (2006—T2), intensional reading [INT].
See also fig. 11. The implied child 2006.Pcar_Peve_IC is introduced and asserted to include five species-level concepts sec. Franz & O’Brien (2001). (A) Containment with overlap graph. (B) Merge concept graph.
Fig 13.
Alignment 5—Franz (2006—T1) and Franz & Cardona-Duque (2013—T2), ostensive reading [OST].
Modifications of the initial input given in [1] are shown in bold font. Non-coverage (nc) is stipulated for parent concept 2013.PHY of child concept 2013.PHYsubcin. (A) Containment with overlap graph. (B) Merge concept graph.
Fig 14.
Alignment 5—Franz (2006—T1) and Franz & Cardona-Duque (2013—T2), intensional reading [INT].
See also fig. 13. The implied child 2006.Pcar_Peve_IC (2006:Pis introduced and asserted to include seven species-level concepts sec. Franz & Cardona-Duque (2013). (A) Containment with overlap graph. (B) Merge concept graph.
Fig 15.
Alignment 6—Franz & O’Brien (2001—T1) and Franz & Cardona-Duque (2013—T2), ostensive reading [OST].
Modifications of the initial input given in [1] are shown in bold font. (A) Containment with overlap graph. (B) Merge concept graph.
Fig 16.
Alignment 6—Franz (2006—T1) and Franz & Cardona-Duque (2013—T2), intensional reading [INT].
See also fig. 15. The implied child Ppub_Psul_IC is introduced and asserted to include two species-level concepts sec. Franz & Cardona-Duque (2013). The containment with overlap and merge concept graphs are identical.
Table 1.
Summary of input concepts and input/output articulations (MIR—maximally informative relations) for the Perelleschus use case, corresponding to the 13 readings for alignments 1–6 shown in Figs. 4–16.