Some soft measures for performance analysis: The “core” dimensions of expert system quality

The post-implementation evaluation of new technologies such as artificial intelligence, multimedia or parallel computing systems which are in operation is a much neglected problem. There is a lack of methods, techniques and tools for the practitioner to analyze the performance such systems. In earlier work, we have presented a framework and methodology that described a socio-technical approach for evaluating expert systems. In short, we determined the features and characteristics of expert systems that are most critical for their “implementation success”. This was the result of empirical evidence from a first-ever field study of expert systems in production. In this paper, specifically tailored to an audience of generalists, we retrospectively put our work into the framework of performance measurement and analysis. The end-result of our investigations is that we have definitional clarity as to how we may alternately evaluate such decision support systems.     

Scintillation prediction models compared to measurements on a timebase of several days

The prediction models for scintillation variance of ITU-R and Karasawa et al. (1988) are compared to measurements on a time base which is much shorter than originally considered for these models. It is shown that even after averaging over just three days, a good correlation with the measured variance is found. Both models predict higher scintillation variance than is measured at 19.8 GHz on a 12.7° elevation path in Kirkkonummi, Finland     

Pressure release of containments during severe accidents in Switzerland

As required by the Swiss Federal Nuclear Safety Inspectorate (HSK) all Switzerland's five nuclear power plants have to install a containment filtered venting system. The integrity of the containment (the last barrier for radioactive releases to the environment) can be threatened by overpressure due to inadequate heat removal. Design requirements have been provided for a specific class of severe accident scenarios. In general the capacity of the system is considered sufficient if it is able to vent the steam production corresponding to a decay heat level of 1% of the thermal reactor power. The mitigation capacity for the reduction of released radioactive material is specified by a retention factor of 1000 for aerosols to prevent or limit a long term ground contamination and a factor of 100 for elementary iodine for prevention or limiting of thyroid doses and to avoid short term evacuation. Besides existing requirements for design, maintenance and operation, additional claims such as passivity and operability at any pressure conditions inside the containment have to be met. Passivity implies that the system can be initiated after a severe accident without any operator action. The system also has to allow early manual venting. Various filtered venting systems are presently available. The nuclear power plants of Beznau, Gosgen, Leibstadt and Muhleberg have already selected such systems and already implemented them or are going to install them step by step. Beznau selected the Sulzer-EWI system which is using a water pool with nozzles-baffle plates and mixing elements to achieve the required filtration of the aerosols. In both Beznau units, the systems are installed and in standby mode. Gosgen, a pressurized water reactor as well as Beznau, is going to implement a filter system developed by Siemens-KWU, known as sliding pressure venting process, combining a venturi scrubber in a water pool and a mesh filter. The boiling water reactor of Leibstadt also selected the same system as Beznau while Müheberg choose the ABB system but not in the common design. The venturi pipes are thereby integrated in the water pool of the outer torus. The system in all five nuclear power plants is fully operable and in standby mode since December 1993.     

Comparison of physical property-porosity behaviour with minimum solid area models

An extensive survey of the porosity dependence of (room temperature) physical properties shows that mechanical properties and electrical and thermal conductivity, i.e. properties dependent on the local flux or fields in the material, follow minimum solid area models. This is shown extensively for elastic properties and tensile (flexure) strength, but consistency with other properties, e.g. compressive strength, hardness, electrical and thermal conductivity is also shown. Although data for ceramics is most extensive, data for rocks, metals, and carbon are included, since the consistency of these, especially of metals with ceramics, provides important support for the minimum solid area concept. While porosity characterization is generally minimal, expected model trends with pore character are corroborated by correlating processing and resultant expected pore character with porosity-property results. It is argued that properties dependent on mass should be better fit by a linear, i.e. rule of mixture, relationship between such properties and porosity. Support for this is shown in dielectric constant-porosity data.     

Absorbability of Calcium From Common Beans

Absorption of calcium from white, red, and pinto beans, labeled with ⁴⁵Ca, was measured in 24 women in a three-way randomized design using a calcium load of 72 mg and milk as the referent. Fractional absorption for the three beans averaged 0.219 ± 0.047 and did not differ by type. Milk calcium absorption at the same load was more than two times higher, 0.451 ± 0.088 (P < 0.001). Oxalate content averaged 0.34%, and phytate averaged 1.7%, a stoichiometric excess relative to calcium. To evaluate the relation of phytate to reduced absorbability, labeled pinto beans were pre-treated with phytase and fed to 10 subjects. Fractional absorption rose, averaging 0.318 ± 0.071, (P <0.01 vs. untreated beans), but was significantly below that of milk. The difference was partly accounted for by phytate, with the remainder probably due to relatively high oxalate.     

Correlation between clinical course and quantitative analysis of the ischemia related artery in patients with unstable angina pectoris, refractory to medical treatment. Results of two randomized trials. The European Cooperative Study Group

Patients with unstable angina, refractory to intensive medical therapy, are at high risk for developing thrombotic complications, such as recurrent ischemia, myocardial infarction and coronary occlusion during coronary angioplasty. As both platelet aggregation and/or thrombus formation play an important role in this ongoing ischemic process, a monoclonal platelet GPIIb/IIIa receptor antibody (c7E3) or thrombolytic therapy (alteplase) might be able to modify the clinical course and underlying coronary lesion morphology. To evaluate whether alteplase or c7E3 could influence the incidence of complications, we randomized 36 and 60 patients, respectively to alteplase or placebo, or c7E3 or placebo. All patients exhibited dynamic ECG changes and recurrent pain attacks, despite maximal tolerated medical therapy. Patients were randomized in both studies after initial angiography had demonstrated a culprit lesion amenable for angioplasty. After study drug infusion quantitative angiography was repeated and angioplasty performed. Recurrent ischemia during study drug infusion occurred in 5, 6, 9 and 16 patients from the alteplase, placebo, c7E3 and placebo group, respectively. Major events defined as death, myocardial infarction or urgent intervention occurred in 7, 3, 1 and 7 patients, respectively. Two patients died: one in the alteplase group and one in the placebo group from the c7E3 study. The first patient due to retroperitoneal hemorrhage, the second as a result of recurrent infarction. Qualitative angiography showed resolution of clots in the c7E3 group only, while the same group of patients showed in 20% an improvement in TIMI flow grade, without deterioration in any patient from this group.(ABSTRACT TRUNCATED AT 250 WORDS)