Lebenszyklusinformationen von Wissensdokumenten: Erfassung, Verwaltung und Validierung

With the growing number of documents which are digitally available, users tend to have more and more problems to organize these documents personally or collaboratively. Specifically for knowledge workers, fast retrieval of relevant documents is essential. In many cases, users have problems to rediscover documents that they or members of their working group once stored somewhere. Often, users rather download documents from the Internet once again instead of searching them in their local file system. Often they do not even know that documents where members of their working group documented relevant knowledge even exist. One reason for the bad retrievability of documents organized in file systems is the lack of information about such documents. Metadata is hardly maintained and usually consists of information automatically provided by the operating system only. This information is barely distinctive and does not help to improve the situation described above. Furthermore, existing tools for retrieval and management of documents like the Windows Filesystem Explorer do not support users sufficiently. The present thesis is based on the observation that a multitude of information emerges from actions performed on a document during its lifecycle. Users open, read, edit or use documents several times. These processes provide for the emergence of information that can be utilized to support both the management and the retrieval of these documents. However, most of the information is lost if it is not captured during those processes. Manual creation of such information would be too much effort and too costly. The underlying approach of this thesis is the automatic capture of information emerging from processes conducted on a document during its lifecycle. The thus acquired information should then be organized, processed and made accessible for utilization. Initially, we analyze which kinds of information emerge during the lifecycle of a knowledge document. Based on existing lifecycle models from other domains we develop a lifecycle model for knowledge documents. We define the concept of lifecycle information and categorize it further into relation and usage information. On this basis we identify the information that emerges during the different phases of a document’s lifecycle. Hereby, we focus on relation information emerging from reuse processes conducted on knowledge documents. Before lifecycle information can be utilized, it has to be captured, managed and made accessible across different systems and applications. Finally, we have to make sure that the relations captured stay valid. All of these aspects are addressed in the given thesis. We have designed, implemented and evaluated a framework for automatic capture, management and utilization of lifecycle information. The framework deploys a plug-in-based concept for the capture of information that is portable to arbitrary applications. We have identified two different means to capture valid relations and implemented both in three different capture components. The management and provision of captured information is done in a server-based manner. We propose a scheme for the organization of lifecycle information, which specifically covers the capture and management of relation information, for which no sufficient solution existed so far. Furthermore, we have designed various scenarios for the utilization of lifecycle information and implemented two of them prototypically. Especially for relation information it is necessary to ensure the validity of the information captured. If there is an action that provides for the emergence of a relation there might also be actions that cause the relation to become invalid. To address this issue we have designed two algorithms for the automatic validation of relation information and have evaluated them on different corpora. On the given corpora our algorithms perform better quality-wise than state of the art approaches while maintaining a lower storage consumption. We furthermore show that the proposed algorithms can be applied in various additional scenarios. The user-based evaluation of the proposed framework we have conducted shows that the capture of valid lifecycle information is achievable with high reliability. Through the automatic capture of lifecycle information of knowledge documents, this thesis creates a basis and prerequisite for the utilization of this new kind of information in various scenarios.