Citation Context

...ngthened GSKB in first-order logic is also the basis for a couple of AURA knowledge base exports in SILK [17], answer-set programming syntax [18], and TPTP FOF syntax [19]. We also have a set of OWL2 =-=[20, 21]-=- exports [22], but the OWL2 KBs are underspecified in the sense that we cannot use Skolem functions here and hence, no equality atoms can be used to establish co-references, so those KBs are approxima...

Citation Context

...the full-scale AURA GSKB, so scalability of the approach is still open. Considering our graph-based approach, we note that graph morphism-based approaches were employed since the early days of KL-ONE =-=[11]-=-, to decide subsumption between concepts in description logics. Those approaches are called structural subsumption algorithms nowadays [1]. Note that we do not decide subsumption between concepts, as ...

Citation Context

...hora resolution; for example, [5, 6] use default rules to hypothesize equality assertions between referents in order to guess and establish co-references. The reasoning system Knowledge Machine (KM), =-=[9]-=-, uses a so-called unification mechanism, Umap, to strengthen the GSKB by establishing co-references. Unfortunately, the lack of a formal semantics makes it very difficult to understand. A major probl...

Citation Context

... is talking about actually the same Chromosome that is mentioned in S1? These kinds of question are studied to some extent in the field of computational linguistics under the term anaphora resolution =-=[5]-=-, [6]. We will use the term co-reference resolution in the following. Obviously, natural language is also underspecified in that sense. From a logical point of view, the sentences S1 to S5 (and their ...

Citation Context

...cept subsumption in EL++, which are similar to morphisms. But note that, unlike EL, we are supporting graph-based descriptions, and this is a key requirement in AURA. Morphisms are also considered in =-=[14]-=- for simple conceptual graphs for deciding “projection” (a form of subsumption). However, no implementation is described there, and we are using different morphisms, as we do not require that all rela...

Citation Context

...y result in decidability problems [15] which can often only be regained by imposing severe artificial modeling restrictions. Although some progress has been made on modeling graph structures with DLs =-=[16]-=-, those extensions are still too restricted to be useful here. Our experience is that graph structures are central to biology, and approximating them by trees results in coarse models. Our framework a...

Citation Context

...m a success story. It is well-known that the modeling of graph structures is challenging in description logic (DL), as derivations from the tree-model property usually result in decidability problems =-=[15]-=- which can often only be regained by imposing severe artificial modeling restrictions. Although some progress has been made on modeling graph structures with DLs [16], those extensions are still too r...

Citation Context

...ion. Nevertheless, we suspect that our algorithm could be turned into a structural subsumption checker for certain description logics, similar as in [12] for EL. So-called simulations are employed in =-=[13]-=- to decide concept subsumption in EL++, which are similar to morphisms. But note that, unlike EL, we are supporting graph-based descriptions, and this is a key requirement in AURA. Morphisms are also ...

Citation Context

...valent clusters in ∆A to generate equality atoms. 5. From Σ′ it is possible to decide the provenance and the co-reference problem, on a syntactic basis. Since steps 1 and 2 are standard and well-know =-=[8]-=-, let us define the algorithm for step 3. We need two more utility notions before we can proceed: Definition 9. Relations E and U , and equivalence classes. Let IH = (∆H, ·IH) be the minimal unique He...

Citation Context

...an atom is inherited was called provenance computation. Nevertheless, we suspect that our algorithm could be turned into a structural subsumption checker for certain description logics, similar as in =-=[12]-=- for EL. So-called simulations are employed in [13] to decide concept subsumption in EL++, which are similar to morphisms. But note that, unlike EL, we are supporting graph-based descriptions, and thi...

Citation Context

...first-order logic is also the basis for a couple of AURA knowledge base exports in SILK [17], answer-set programming syntax [18], and TPTP FOF syntax [19]. We also have a set of OWL2 [20, 21] exports =-=[22]-=-, but the OWL2 KBs are underspecified in the sense that we cannot use Skolem functions here and hence, no equality atoms can be used to establish co-references, so those KBs are approximations. This s...

Citation Context

...e subject matter experts in our knowledge factory. The strengthened GSKB in first-order logic is also the basis for a couple of AURA knowledge base exports in SILK [17], answer-set programming syntax =-=[18]-=-, and TPTP FOF syntax [19]. We also have a set of OWL2 [20, 21] exports [22], but the OWL2 KBs are underspecified in the sense that we cannot use Skolem functions here and hence, no equality atoms can...

Citation Context

...lemented this novel GSKB covering in a scalable way, and have successfully applied it to the AURA GSKB [2], [4]. The AURA GSKB is the basis of the intelligent question answering textbook Inquire, see =-=[7]-=-2, and currently contains around 1 There is a possibility that two inherited nodes get equated / merged into one node, which means that those two (or more) nodes will be mapped to one node by the morp...

Citation Context

...ain motivation for the work described in this paper. It turned out that the recomputed provenance information was quite accurate, as confirmed by the experts in our knowledge factory [4]. The work of =-=[10]-=- uses answer set programming (ASP) to formalize KM’s Umap operator. The GSKB is specified as an ASP program, together with an axiomatic system of ASP inference rules. These rules capture the semantics...

Citation Context

...alking about actually the same Chromosome that is mentioned in S1? These kinds of question are studied to some extent in the field of computational linguistics under the term anaphora resolution [5], =-=[6]-=-. We will use the term co-reference resolution in the following. Obviously, natural language is also underspecified in that sense. From a logical point of view, the sentences S1 to S5 (and their corre...

Citation Context

... of the mappings got testified by the subject matter experts in our knowledge factory. The strengthened GSKB in first-order logic is also the basis for a couple of AURA knowledge base exports in SILK =-=[17]-=-, answer-set programming syntax [18], and TPTP FOF syntax [19]. We also have a set of OWL2 [20, 21] exports [22], but the OWL2 KBs are underspecified in the sense that we cannot use Skolem functions h...

Citation Context

...n our knowledge factory. The strengthened GSKB in first-order logic is also the basis for a couple of AURA knowledge base exports in SILK [17], answer-set programming syntax [18], and TPTP FOF syntax =-=[19]-=-. We also have a set of OWL2 [20, 21] exports [22], but the OWL2 KBs are underspecified in the sense that we cannot use Skolem functions here and hence, no equality atoms can be used to establish co-r...

Citation Context

...cCell in AURA. In the AURA project, such CGraphs were modeled by subject-matter experts (SMEs) from the pages of a biology college textbook [3], following a detailed “text to CGraph” encoding process =-=[4]-=-: S1 Every Cell has part a Ribosome and a Chromosome . S2 Every EukaryoticCell is a Cell . S3 EveryEukaryoticCell has part aEukaryoticRibosome , aEukaryoticChromosome , a Nucleus, such that the Eukary...

Citation Context

...s, so those KBs are approximations. This suite of exported KBs is called the Bio KB 101 suite, and is made available to the public under a Creative Commons License, and can be found here for download =-=[23]-=-, together with a detailed description of the various variants of the exports. In future work, we need to establish a closer correspondence between the logical and the graph-based formalizations. It s...