Finding stories in the archive through paragraph alignment
Wenjian Xu, María Esteva · Literary and Linguistic Computing · 2011
Referring to the concept of archival bond, we define stories as formed by documents that relate to a target activity and developed a method called paragraph alignment to find these documents. The method computes archival bond by measuring the cosine similarity between document paragraphs. We tested the method in a chaotic case study collection created in a shared server by different authors. Results demonstrate that this method is more efficient to find stories than calculating the cosine similarity between entire documents. This research helps archivists make sense of collections that are considered inaccessible and whose stories may otherwise be lost.