Efficient allocation of resources to activities is pivotal in executing business processes but remains challenging. While resource allocation methodologies are well-established in domains like manufacturing, their application within business process management remains limited. Existing methods often do not scale well to large processes with numerous activities or optimize across multiple cases. This paper aims to address this gap by proposing two learning-based methods for resource allocation in business processes. The first method leverages Deep Reinforcement Learning (DRL) to learn near-optimal policies by taking action in the business process. The second method is a score-based value function approximation approach, which learns the weights of a set of curated features to prioritize resource assignments. To evaluate the proposed approaches, we first designed six distinct business processes with archetypal process flows and characteristics. These business processes were then connected to form three realistically sized business processes. We benchmarked our methods against traditional heuristics and existing resource allocation methods. The results show that our methods learn adaptive resource allocation policies that outperform or are competitive with the benchmarks in five out of six individual business processes. The DRL approach outperforms all benchmarks in all three composite business processes and finds a policy that is, on average, 13.1% better than the best-performing benchmark.

翻译：资源向活动的有效分配是执行业务流程的关键，但仍面临挑战。尽管资源分配方法在制造业等领域已相当成熟，但其在业务流程管理中的应用仍十分有限。现有方法在处理包含大量活动的大规模流程或多案例优化时往往扩展性不足。本文旨在通过提出两种基于学习的业务流程资源分配方法来解决这一差距。第一种方法利用深度强化学习（DRL）通过在业务流程中采取行动来学习近优策略。第二种方法是一种基于评分的价值函数近似方法，通过学习一组精选特征的权重来优先分配资源。为评估所提方法，我们首先设计了六个具有典型流程模式与特征的独立业务流程，随后将这些流程连接成三个实际规模的复合流程。我们将所提方法与传统启发式算法及现有资源分配方法进行了对比。结果表明，我们的方法能学习到自适应的资源分配策略，在六个独立业务流程中有五个优于或与基准方法相当。DRL方法在所有三个复合流程中均优于所有基准方法，且找到的策略平均比最优基准方法高出13.1%。