Social Process Mining

研究项目“社会过程挖掘”（SPM）旨在开发、实施和评估一种新的过程挖掘方法，该方法适用于企业协作系统（ECS）中基于日志的用户行为检测，并有助于自动检测使用模式（协作场景）。由于ECS的解释灵活性，其中可能的工作流比例如ERP系统中的工作流更加可变和非结构化。在ECS中缺乏相同活动序列的重复导致了远比ERP系统更复杂的过程模型，因此仅通过手工检查模型几乎不可能识别典型的行为模式。到目前为止，现有的流程挖掘方法不能自动支持识别在ECS流程中出现的行为模式。即使是针对弱结构过程量身定制的方法（例如，声明性过程挖掘或案例管理方法）在这里也被拉伸到了极限。SPM方法旨在通过自动识别生成的流程模型中的典型协作场景来纠正这种情况。因此，我们将开发和实现一种基于频繁子图挖掘的模式识别算法，作为SPM的一部分。该算法使用真实的ECS日志数据和已知的使用模式进行训练。近年来，研究小组在初步研究中建立了对识别和提供这些已知使用模式（领域知识）所必需的过程的深入理解。SPM的一般（技术）可行性也已在一些初步研究中得到证实。来自UniConnect协作平台的大量日志数据可用于该项目，该平台由科布伦茨-朗道大学托管，为DACH领域的3000多名注册用户提供了10多年的服务。在SPM开发完成后，新算法将在探索性的现场研究中进行测试，使用该领域实际公司的大规模协作系统的实际日志数据。其中一些系统有成千上万的用户，这意味着可以预期有意义的分析结果。计划中的SPM项目因此增加了理论和设计知识体系。设计目标是开发、实施和评估SPM方法和一个互补的软件工具。对理论的贡献是更好地理解ECS中的用户行为，一方面是通过识别和测试协作场景的形式，另一方面是通过揭示它们在真实公司环境中的使用模式的出现。所获得的知识可用于更好地理解工作场所中计算机支持的协作过程，并在此基础上制定更明智的措施供用户采用。

The research project "Social Process Mining" (SPM) aims to develop, implement and evaluate a novel process mining approach, which is suitable for log-based detection of user behaviour in Enterprise Collaboration Systems (ECS) and which facilitates the automatic detection of usage patterns (collaboration scenarios).Due to the interpretive flexibility of ECS, the workflows that are possible in them are much more variable and unstructured than, for example, in ERP systems. The lack of repetition of the same sequences of activities in ECS leads to far more complex process models than, for example, in ERP systems, so that it is hardly possible to identify typical behaviour patterns through merely examining the models manually. Existing process mining approaches do not provide automatic support to identify behavioural patterns as they occur in ECS processes so far. Even approaches that are tailored towards weakly structured processes (e.g., declarative process mining or case management approaches) are stretched to their limits here.The SPM approach is intended to remedy this situation by automatically identifying typical collaboration scenarios in the generated process models. Therefore, we will develop and implement a pattern recognition algorithm based on Frequent Subgraph Mining as part of the SPM. The algorithm is trained using real-world ECS log data and known usage patterns. The in-depth understanding of the processes necessary for the identification and provision of such known usage patterns (domain knowledge) has been established by the research team in preliminary studies in recent years. The general (technical) feasibility of SPM has also been demonstrated in several preliminary studies.Extensive log data from the UniConnect collaboration platform, which the University of Koblenz-Landau has been hosting for over 3,000 registered users in the DACH area for more than 10 years, is available for the project. After completion of the SPM development, the new algorithm will be tested in an explorative field study using the actual log data of highly scaled collaboration systems of real-world companies in the field. Some of these systems have hundreds of thousands of users, which means that meaningful analysis results can be expected.The planned SPM project thus adds to theory as well as to the design body of knowledge. The design goal to develop, implement and evaluate the SPM approach and a complementary software tool. The contribution to theory is the better understanding of user behaviour in ECS, on the one hand in the form of identified and tested collaboration scenarios and on the other hand by uncovering the occurrences of their usage patterns in the context of real companies. The acquired knowledge can be used to better understand computer-supported collaboration processes in the workplace and, based on this, to develop better-informed measures for user adoption.