Modelling Resource Allocation in a Decentralized Organization with an AI‐Based, Goal‐Directive Model*
整合三种人工智能技术（过滤波束搜索、世界模型、黑板模型）为分权组织中的资源分配决策过程建模，解决了传统数学规划方法难以处理大规模问题的局限，并展示了在三层分权组织中的应用。
ABSTRACT Modelling the resource allocation decision process in a decentralized organization using mathematical programming decomposition approaches has proven intractable for all but the smallest of problems. In this paper three artificial intelligence (AI) techniques are integrated to model the resource allocation decision process, and the solution for a three‐level decentralized organization is illustrated. The techniques are (1) the filtered beam search, which selects a list of potential projects at the subordinate level, (2) the world model, which describes the ordinate and superordinate's worlds and their decision‐making processes, and (3) the blackboard model, which allows for global, but selective, storage of information. Benefits of the new approach include: several different mechanisms of managerial control may be modelled, the true information flow in an organization is more closely mirrored, and problems of a more realistic size are now viable. There are three primary ways in which the new model could be useful to management: The desirability and effects of different organizational structures can be modeled, the efficacy of various coordination mechanisms along the organizational structure can be examined, and a subset of projects from a wider list of possibilities can be selected. Also, an example is given showing how the model may be extended to scenarios with asymetric information and divergence of preferences, whereby subordinates have different objectives than their managers and where there is no way for managers to verify the truthfulness of their subordinates' responses.