SILK is an expressive Semantic Web rule language and system equipped with scalable reactive higher-order defaults. We present one of its latest novel features: a graphical user interface (GUI) for knowledge entry, query answering, and justification browsing that supports user specification and understanding of advanced courteous prioritized defeasible reasoning. We illustrate the use of the GUI in an example from college-level biology of modeling and reasoning about hierarchicallystructured causal processes with interfering multiple causes.
SILK has a new fundamental KR: hyper logic programs, which extends
normal declarative logic programs (LP). Hyper LP is the first to tightly combine
several key advanced expressive features: defaults , with strong negation and
priorities, cf. courteous LP [
Outline and Contributions: A previous version of SILK was presented in
[
SILK Graphical User Interface & Defeat Justifications We have developed a graphical user interface (GUI) to the SILK system for knowledge entry, query answering, and justification browsing. The GUI is currently used by KEs and is being extended to support use by subject matter experts (SMEs). The GUI supports user specification and understanding of advanced courteous prioritized defeasible reasoning. It is implemented as a plug-in to the Eclipse Integrated Development Environment (IDE).
The GUI, pictured in Figure 1, offers users a number of capabilities. Entered SILK statements are syntactically validated and statement components (e.g., annotations) are color-coded for clarity. User debugging of rule bases is facilitated by automatic tracking of target queries’ results against user changes to the rules. This also allows what-if explorations. The GUI also offers query result justification trees (technically, graphs) that can be explored incrementally, by expanding each tree node to display its children. At each node, the user can specify particular bindings to filter the portion of the justification tree that is displayed. Trees of this sort are also available for negative results (i.e., when a literal cannot be inferred), allowing developers to drill down and identify flaws in a desired chain of logical reasoning. This display mechanism also supports the reasoning chains found in courteous defaults by showing defeated ground rule instances — rules whose heads are not true, despite their bodies being true, due to conflict with other rules. Figure 1 shows an example of refutation-flavor defeat of a rule instance (and thus of its head atom A). The rule instance has a candidate argument — i.e., the rule’s body is satisfied. But there also is a candidate counter argument (whose head is neg A) that has a higher-priority rule tag.
In (ASCII) SILK syntax, a # c means that a is an instance of class c. Skolems are prefixed by the underscore character (“ ”). A courteous rule label term, used for prioritization, is called a tag.
Flexible control of selection and layout of items to display is achieved via running rules that are specified (e.g., by power users) in SILK itself.
To our knowledge, the SILK GUI is only the second justification exploration
GUI for (prioritized) defeasible rules that have (declarative, model-theoretic)
semantics. The first such system was DR-DEVICE [
We have developed a novel approach to modeling and reasoning about hierarchicallystructured causal processes, that smoothly handles interference/exceptions between multiple causes and elegantly treats the “frame problem” (inertia / persistence of causal fluents). It leverages hyper LP’s prioritized defaults. To fully describe the approach is beyond the scope of this paper, however. Instead, we illustrate the approach with an example of college-level biology that shows the SILK GUI’s novel capability to explore justifications in the presence of prioritized defeat.
In biology and medicine, a key process is the cell cycle in which a cell grows and then divides. (Control failure in this process causes cancer.) The cell cycle is a complex hierarchically-structured process. It consists of two phases (subprocesses): interphase and mitosis, in that temporal order. Interphase, in turn, consists of three subphases G1, S, and G2, in that order. Mitosis too has several subphases. Many of the above subphases in turn have sub-subphases, etc. DNA synthesis occurs during S phase, and indeed begins when S phase begins. This is the knowledge required to answer the following first-year college exam question6: A researcher treats cells with a chemical that prevents DNA synthesis from starting. This treatment traps the cells in which part of the cell cycle?
Correct answer: G1.
That is, if DNA synthesis does not occur, then S Phase does not occur, and the cell cycle stops in the preceding phase which is G1.
Figure 1 shows a key intermediate step in SILK’s inferencing — defeat of a candidate argument that: DNA Synthesis will occur for the question’s focal cell cycle ( Cell Cycle79), as is normal (i.e., “intended”) in the cell cycle’s process’s cascade of causal phase steps. That argument is refuted because there is a higherpriority counter argument (itself undefeated) based on the preventive/inhibitory causal effect of the hypothetical scenario’s chemical treatment. 4
A key direction in current and future SILK work is to increase SME friendliness of the UI in collaborative KA and querying, using in part controlled natural language. SILK is now being integrated with other portions of Project Halo, particularly AURA. We are refining a translation of Cyc biology etc. knowledge to SILK. See the SILK website for an extended version of this paper.
Acknowledgements: Thanks to the SILK team, esp. Paul V. Haley, Terrance Swift, Michael Kifer, David Gunning, Vinay Chaudhri, Michael Gelfond.