While governments have long discussed the promise of delegating important decisions to machines, actual use often lags. Consequently, we know little about the variation in the deployment of such delegations in large numbers of similar governmental organizations. Using data from crime laboratories in the United States, we examine the uneven distribution over time of a specific, well-known expert system for ballistics imaging for a large sample of local and regional public agencies; an expert system is an inference engine joined with a knowledge base. Our statistical model is informed by the push-pull-capability theory of innovation in the public sector. We test hypotheses about the probability of deployment and provide evidence that the use of this expert system varies with the pull of agency task environments and the enabling support of organizational resources—and that the impacts of those factors have changed over time. Within this context, we also present evidence that general knowledge of the use of expert systems has supported the use of this specific expert system in many agencies. This empirical case and this theory of innovation provide broad evidence about the historical utilization of expert systems as algorithms in public sector applications.

Tools for analysing additive manufacturability often employ complex models that lack transparency; this impedes user understanding and has detrimental effects on the implementation of results. An expert system tool that transparently learns features for successful printing has been created. The tool uses accessible data from STL models and printer configurations to create explainable parameters and identify risks. Testing has shown good agreement to print behaviour and easy adaptability. The tool reduces the learning curves designers face in understanding design for additive manufacturing.

Due to the complexity of financial products, consumers with low financial literacy are left behind. However, few practical studies investigated the impact of digital advisory systems to improve customer experience. Configuration systems, as the most popular expert systems, have never been applied as a financial tool. In this study, we explore the application of configuration systems to educate consumers in achieving their financial goals. The results highlight multiple benefits from using the configurator, including greater financial literacy and improved individual financial decision-making.

The WEEDSIM bioeconomic model was developed in Minnesota and was designed to support weed management decisions for both soil-applied and postemergence weed control programs in Zea mays and Glycine max. In this research, we adapted the WEEDSIM weed management model to Michigan by modifying the crop yield loss functions and herbicide efficacy ratings. We then validated the components of the model and determined whether WEEDSIM led to more profitable weed management than recommendations from SOYHERB or CORNHERB, computer decision aids based solely on herbicide efficacy and cost. The crop year significantly influenced the weed-free yield in Z. mays and G. max, but the weed—crop interference function did not change each year. Total weed seed increased in the untreated compared with the weed-free control over the 3-yr period. Weed seed did not increase significantly in WEEDSIM preemergence/postemergence (PRE/POST), WEEDSIM postemergence, or CORNHERB or SOYHERB treatments compared with the weed-free control, although annual grass seedling density at the time of postemergence herbicide application had increased by 1995 in the WEEDSIM postemergence treatment in G. max because of a 2,4-D amine application only in Z. mays in 1994. WEEDSIM PRE/POST and CORNHERB provided excellent weed control in all three years, and WEEDSIM PRE/POST resulted in gross margins over weed control costs equal to or greater than CORNHERB recommendations. In G. max, Chenopodium album and annual grass control was excellent in all three years for WEEDSIM PRE/POST, WEEDSIM postemergence, and SOYHERB treatments. The highest average gross margin for the 3-yr study was from mechanical weed control in 76-cm-wide rows of G. max ($806 ha−1) and from SOYHERB in 38- and 19-cm-wide rows of G. max ($776 and $808 ha−1, respectively). WEEDSIM recommendations controlled weeds and maintained crop yield in both Z. mays and G. max.

A bioeconomic weed management model was tested as a decision aid for weed control in corn at Rosemount, MN, from 1991 to 1994. The model makes recommendations for preemergence control tactics based on the weed seed content of the soil and postemergence decisions based on weed seedling densities. Weed control, corn yield, herbicide active ingredient applied, and economic return with model-generated treatments were compared to standard herbicide and mechanical control treatments. Effects of these treatments on weed populations and soybean yield the following year were also determined. In most cases, the model-generated treatments controlled weeds as well as the standard herbicide treatment. The quantity of herbicide active ingredient applied decreased 27% with the seed bank model and 68% with the seedling model relative to the standard herbicide treatment. However, the frequency of herbicide application was not reduced. In 1 yr, seed bank model treatments did not control weeds as well as the standard herbicide or seedling model treatments. Corn yields reflected differences in weed control. Net economic return to weed control was not increased by using model-generated control recommendations. Weed control treatments the previous year affected weed density in the following soybean crop. In 2 of 3 yr, these differences did not after weed control or soybean yield. Although tactics differed, the bioeconomic model generally resulted in weed control and corn yield similar to the standard herbicide. The model was responsive to differing weed populations, but did not greatly after economic returns under the weed species and densities in this research.

A bioeconomic model was tested as a decision aid for weed control in soybean at Rosemount, MN, from 1991 to 1994. The model makes recommendations for preplant incorporated and preemergence control tactics based on the weed seed content of the soil and postemergence decisions based on weed seedling densities. Weed control, soybean yield, herbicide use, and economic return with model-generated treatments were compared to standard herbicide and mechanical control systems. Effects of these treatments on weed populations and corn yield the following year were also determined. In most cases, the model-generated treatments controlled weeds as well as a standard herbicide treatment. Averaged over the 3 yr, the quantity of herbicide active ingredient applied was decreased by 47% with the seedbank model and 93% with the seedling model compared with a standard soil-applied herbicide treatment. However, the frequency of herbicide application was not reduced. Soybean yields reflected differences in weed control and crop injury. Net economic return to weed control was increased 50% of the time using model-recommended treatments compared with a standard herbicide treatment. Weed control treatments the previous year affected weed density in the following corn crop but had little effect on weed control or corn yield. The bioeconomic model was responsive to differing weed populations, maintained weed control and soybean yield and often increased economic returns under the weed species and densities in this research.

The beef cattle investment decision provides an excellent opportunity to increase the economic efficiency of beef cattle production. The investment questions that face beef cattle producers are of interest to beef cattle producers, educators, and financial institutions involved in lending to beef cattle producing firms. This study develops a decision support aid utilizing expert system technology to assist beef cattle producers in making well-founded investment decisions with respect to the firm's beef cattle herd.

This paper argues that most of the problems that actuaries have to deal with in the context of non-life insurance can be usefully cast in the framework of computational intelligence (a.k.a. artificial intelligence), the discipline that studies the design of agents which exhibit intelligent behaviour. Finding an adequate framework for actuarial problems has more than a simply theoretical interest: it also allows a technological transfer from the computational intelligence discipline to general insurance, wherever techniques have been developed for problems which are common to both contexts. This has already happened in the past (neural networks, clustering, data mining have all found applications to general insurance) but not in a systematic way. One of the objectives of this paper will therefore be to introduce some useful techniques such as sparsity-based regularisation and dynamic decision networks that are not yet known to the wider actuarial community.

Whilst in the first part of this paper we dealt mainly with data-driven loss modelling under the assumption that all the data were accurate and fully relevant to the exercise, in this second part of the paper we explore how to deal with uncertain knowledge, whether this uncertainty comes from the fact that the data are not fully reliable (e.g. they are estimates) or from the fact that the knowledge is “soft” (e.g. expert beliefs) or not fully relevant (e.g. market information on a given risk). Most importantly, we will deal with the problem of making pricing, reserving and capital decisions under uncertainty. It will be concluded that a Bayesian framework is the most adequate for dealing with uncertainty, and we will present a number of computational intelligence techniques to do this in practice.

Introduction. Farm typologies and cropping practice typologies generally aim at seeking determinants of existing crop management strategies. They constitute the first step for setting improvement goals for cropping systems. Though there are a host of farm typology methods, few deal specifically with farmers’ practices, and even fewer investigate the correlations between practices. We propose here a framework for analysing the determinants of crop management, based on a vision of a crop management sequence condensed into logical combinations of cropping techniques. Materials and methods. This analytical framework was applied to the case of Guadeloupian citrus production, using a representative sample of 41 producers. Three stages were necessary to implement our analytical framework. At stage 1, logical and ordered combinations of cropping practices (CCPs), constitutive of observed as well as reference crop managements (RCMs), were identified through expert analysis. Based on measurements of deviations between farmers’ CCPs and RCMs’ CCPs, a typology of cropping practices was next built. At stage 2, the performances of farmers’ crop managements were evaluated using relevant indicators. Finally, at stage 3, constraints – either related to the environment or to the whole farm management – that determined producers’ cropping practices were identified for making, with the stakeholders, proposals for further technical improvements. Results. Crop management sequences were condensed into five CCPs. A technical profile was then determined for every producer, before a multiple correspondence factorial analysis was run. It identified two groups of producers with contrasting technical profiles. The collective analysis of these results pointed out “weed management” as a major constraint on the cropping systems, revealing that the RCM was inadequate in a context of impossible mechanisation. Discussion. Restructuring complex sequences of cropping techniques into five logical combinations of techniques enabled the comparison with a reference crop management. The cropping systems’ constraints and the objectives for further improvements were then set up collectively by the farmers and social stakeholders, along with the researchers. This analysis constitutes the first stage of a process of redesigning cropping systems, and its result provides a sound basis for a participatory approach

This paper gives a theoretical foundation and implementation details of data and control structures in a diagnostic expert system with adaptive uncertainty management. The considered problem covers a class of expert systems and the application domain in which the operating personnel actively monitor and control the outcome of an industrial manufacturing process. Based on subjective ranking of the latent fault in the final output product, the operator queries the expert system for a recommended corrective action. Subsequent successful or unsuccessful corrective actions consistently modify confidence measures of all corrective recommendations from the same fault set, according to the embedded analytical model. The described approach is based on the relational data representation and procedural control structures, as opposed to a more traditional declarative, production-based system.

Research in computer-aided design/engineering (CAD/E) has focused on enhancing the capability of computer systems in a design environment, and this work has continued in this trend by illustrating the use of the Dempster-Shafer theory to expand the computer’s role in a CAD/E environment. An expert system was created using Dempster-Shafer methods that effectively modeled the professional judgment of a skilled tribologist in the selection of rolling element bearings. A qualitative and symbolic approach was used, but access to simple quantitative models was provided to the expert system shell. Although there has been significant discussion in the literature regarding modification/improvement of the Dempster-Shafer theory, Shafer’s theories were found adequate in all respects for replicating the expert’s judgment. However, an understanding of the basic theory is required for interpreting the results.

A multifaceted approach to evaluating expert systems is overviewed. This approach has three facets: a technical facet, for “looking inside the black box”; an empirical facet, for assessing the system’s impact on performance; and a subjective facet, for obtaining users’ judgments about the system. Such an approach is required to test the system against the different types of criteria of interest to sponsors and users and is consistent with evolving lifecycle paradigms. Moreover, such an approach leads to the application of different evaluation methods to answer different types of evaluation questions. Different evaluation methods for each facet are overviewed.

Surgical waiting lists are of high importance in countries, where the national health system is unable to deliver surgical services at a rate that would allow patients to avoid unnecessary periods of waiting. Prioritization of these lists, however, is frequently arbitrary and inconsistent.

The objective of our research was to analyze the medical decision-making process when prioritizing patients with congenital cardiac malformations for cardiac surgical procedures, identifying an appropriate representation of knowledge, and transferring this knowledge onto the design and implementation of an expert system (“PrioHeart”).

The medical decision-making process was stratified into three stages. The first was to analyze the details of the procedure and patient to define important impact factors on clinical priority, such as the risk of adverse events. The second step was to evaluate these impact factors to define an appropriate “timing category” within which a procedure should be performed. The third, and final, step was to re-evaluate the characteristics of individual patients to differentiate between those in the same timing category.

We implemented this decision-making process using a rule-based production system with support for fuzzy sets, using the FuzzyCLIPS inference engine and expert system shell as a suitable development environment for the knowledge base.

The “PrioHeart” expert system was developed to give paediatric cardiologists a tool to allow and facilitate the prioritization of patients on the cardiosurgical waiting list. Evaluation of “PrioHeart” on limited sets of patients documented appropriate results of prioritization, with a significant correlation between the prioritization made using “PrioHeart” and those results obtained by the individual consultant specialist.

We conclude that our study has demonstrated the feasibility of using an expert system approach with a fuzzy, rule-based production system to implement the prioritization of cardiac surgical patients. The approach may potentially be transferable to other medical subspecialities.

Searching for information is an ubiquitous need in today's data-oriented environments. However, a request for search often entails the service and cooperation of tools managing a diversified set of tasks. In this article, we explore how tools in the form of cooperating agents can be deployed for information management. We describe an agent framework called GAME (goal-oriented, agent-managed environment), and focus on how GAME agents search cooperatively for information requested by a user. Cooperative search entails several issues such as coordinating agent activities, maintaining transparency to agent heterogeneity, and designing information formats to be shared among the agents that require examination. This article analyzes these issues and describes how they are handled in the GAME framework. Cooperative search effectively supports collaboration and information sharing not only among agents in a domain, but also among GAME agents developed across domains. We illustrate the application of cooperative search in task-oriented domains such as Manufacturing and Front Office showing how GAME promotes intradomain and interdomain collaboration in a Factory environment.

Production systems have traditionally been used in engineering to aid the designer in generating designs in domains not significantly influenced by geometry. However, in geometric-based domains such as mechanical engineering, their popularity has been limited. This work proposes the use of shape grammars as the framework for engineering expert systems. Shape grammars are shown to have all the generation and analysis capabilities of traditional production systems, while being able to represent knowledge about both the functionality and the form of a product. Additionally, the parametric nature of shape grammars, their ability to deal with physical form rather than abstract elements, and their ability to recognize emergent shapes give them significant advantages over traditional production systems for geometry-based engineering design.