Expert Knowledge Challenge Problem (EKCP) Scenario for the Kickoff Meeting (produced by SRI/ISI, and PSR) We anticipate our EKCP knowledge base to contain at least three kinds of knowledge. (1) Domain-independent background knowledge (2) Domain-specific knowledge entered by our team's SMEs (3) Domain-specific knowledge entered by the test taking SMEs The goal of the EKCP end-to-end (E2E) system is to enable an analyst to create and maintain a generic modeling capability. This will involve extending the default BW process models that are in the system, and specializing them to new situations. It will also involve creating and maintaining problem-solving methods for reasoning with those models. We first give a description of (3), that is, a sample BW case report to be entered into a KB by an SME. Instead of presenting the full case report, we include only snippets from it. Then we describe (2) - the knowledge that we expect to be already present in the system. Finally, we consider brief examples of (1), that is, the domain-independent knowledge needed for entering (3). DOMAIN-SPECIFIC KNOWLEDGE TO BE ENTERED BY AN SME BW CASE DESCRIPTION FOR IRAQ (adapted from IET/PSR's RKF proposal) (Adapted from Raymond A. Zilinska "Iraq's Biological Weapons: The Past as Future" JAMA (Vol. 278, No. 5: 6 August 1997), pp. 418-424.) Iraq's anthrax program began in 1985 at the Muthanna State Establishment, but moved to Salman Pak in 1987. Utilizing samples from both the US and France along with local isolates, Iraqi scientists produced and animal-tested the virulence of four strains of anthrax. The findings from this research led to mass production of anthrax at the Al Hakam Single Cell Protein Production Plant in 1989. Iraq's current anthrax production capabilities are unknown because the government has made an effort to conceal its BW activities. The United Nations Commission (UNSCOM) that carried out inspection activities in Iraq has determined that Iraq was capable of maufacturing some dual-use equipment that could be used for agent production. If this includes fermenters, for example, then Iraq might have the capability to produce wet anthrax. Also, given the very nature of dual-use equipment, Iraq may not have too much difficulty in replacing equipment destroyed or removed during the war. The key ingredient in their BW program - the trained personnel - are still largely in place. Since the Iraqi defeat in the 1991 Persian Gulf War, however, the international community has tried to ensure that all vestiges of the Iraqi BW program are eliminated. Iraq retained 200 scientists that worked within its BW probram and rebuilt some of the research facilities bombed during the 1991 Persian Gulf War. -----End of Case Report---------- SAMPLE QUESTIONS AND ANSWERS BASED ON THE REPORT Question: If Iraq acquired expertise on dry anthrax production, how might this affect a future Iraqi anthrax weaponization program? Answer: Iraq had the equipment to produce anthrax before 1991. Since dry anthrax is a much more potent weaponization of the agent, it is likely that Iraq would try to pursue this route as opposed to the wet anthrax weaponization. ------- DOMAIN-SPECIFIC KNOWLEDGE BUILT BY TEAM'S SMEs prior to evaluation We expect our team's SMEs to build at least the following domain-specific knowledge in the system. - An ontology of BW related terms representing terms such as: transnational actors, produce agent, prepare wet agent etc. A more extensive listing of such an ontology is attached in the file ekcp-kickoff-supporting-material. - A Model of BW proliferation Developing a BW agent has the following steps: o Acquire Seed stock o Acquire production material o Establish production facilities o Acquire equipment o Establish worker safety o Produce BW agent For each activity, the model will specify the necessary materials, personnel, resources, etc. as well as duration, necessary previous activities or conditions, etc. A graphical representation of such a model is attached in the file ekcp-kickoff-supporting-material. Domain-specific knowledge about Anthrax production and delivery (Note: we could envision a test taker SME entering the knowledge such as the one shown here. For the kickoff meeting, we will assume the knowledge about anthrax production is entered by our team's SME.) Wet anthrax takes the form of a liquid slurry. Dry anthrax is a non-liquid, powder-like substance. The production methods used to develop wet and dry anthrax share some common elements, but differ in several key final process steps. Fermentation is used to grow the anthrax spores, regardless of the wet or dry end state. Following fermenation, however, dry agents then go through several process steps - including milling and drying (e.g., spray dry, vacuum drum dry, or azeotropic distillation) - that the wet agent does not. Once produced, dry agents have far less stringent storage requirements than wet agent, which require environmentally-controlled storage. Also, dry agent can be dispensed more easily than wet agent, which requires sophisticated aerosol dispersion mechanisms to acheive the most effective spore size. (Note: The knowledge in the previous paragraph is an interesting exercise for core theory development. For example, we will need core theory suppport to model production, delivery, and liquid/dry/powdered objects.) Anthrax in dry form, is far more difficult to produce - in terms of product stabilization and quality (i.e., spore size), and in terms of worker safety. On the other hand, dry agents are very easy to dispense and are hardier in the environment - hence their battlefield utility. Liquid agent dispensers require very precise aerosol mechanisms to yield the desired particle size without too much product destruction in the process. The principal tradeoff between production difficulty and delivery difficulty is one which occurs with many BW agents. This takes us then to the particular actor involved, and what they're expected objective and delivery mode is likely to be. An Iraqi program designed to produce an agent for battlefield employment, with the resources of a state actor, is more likely to pursue the dry agent form. A terrorist actor, with more limited resources and different employment objectives, might instead pursue the more simple production avenue of the liquid agent, even sacrificing dispensing efficiency. (Note: The knowledge in the previous paragraph may be best represented as a problem solving method. Here is an example sketch of a PSM to capture this knowledge. If objective = Maximum Damage Explore X=dry anthrax If dont use dry anthrax Explore X=wet anthrax If objective = Cause scare Explore X= wet anthrax Explore X Can produce X Can deliver X Use X Cannnot deliver X Dont use X Cannot produce X Dont use X) DOMAIN-INDEPENDENT KNOWLEDGE An agent can perform a production process if it can meet all the preconditions necessary to enable that process. Distinction between dry and wet objects Dual-use equipment can be acquired and/or repaired more easily than military equipment. Fermenters are a dual use equipment. Spores can be grown through a fermentation process. ;;------------------------------------------ REASONING objective-of for producing anthrax --> Maximum Damage( battlefield utility) --> then look for something easy to dispense and hardier in the environment --> dry anthrax dry anthrax production needs more expertise (far more difficult to produce - in terms of product stabilization and quality (i.e., spore size), and in terms of worker safety) --> Iraq *might* have the capability to produce wet anthrax. --> Iraq was capable of maufacturing some dual-use equipment that could be used for agent production objective-of for producing anthrax --> terror with limited resources --> then find simple production --> wet anthrax