Proposal and Evaluation of Metadata Management Method for eDiscovery

Recent litigation has required the collection of electronic data evidence as well as paper document evidence. The collection process, called "eDiscovery," incurs long hours of searching for evidential files. To reduce the eDiscovery cost, it has been recently noticed that metadata-based search technologies play an important role. Although metadata, such as the name of author and the time and date of creation, usually contain useful information, it is unreliable because it can be overwritten by non-authorized personnel. In this paper, the risk of using metadata for eDiscovery is explained, and a metadata-management method for improving file-search efficiency is proposed. To reveal the risk of metadata, an experiment on connecting eDiscovery applications was performed. In the experiment, the files collected by one application were exported to the other application for reviewing them. The experiment shows that proper operations can create incorrect metadata. To correct the metadata, the proposed method extracts information from day-to-day operations and overwrites the metadata using that information. To evaluate the proposed method, metadata of files modified by 3 examinees were retrieved. Evaluation results show that the accuracy of metadata corrected by the proposed method is 100%, while the accuracy of that recorded by a conventional file system is only 60%. This result implies that the proposed method, using extracted information from day-to-day operations, is an important tool for assuring the files are provided as litigation evidence.