Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
A Computational Framework for the Analysis of Ancient Scripts
Archaeologists and researchers have studied the relationship among ancient scripts since their discovery. The vast image libraries which have recently become available facilitate the computational study of the evolution of these ancient scripts. In this research, we see how various scripts may be classified to belong to particular script families. We focus on the Indus Valley Script, which is an entirely undeciphered script and whose origin is highly debated. This dissertation makes three significant contributions. The first contribution is to the analysis of ancient script similarities using deep learning methods. We designed an algorithmic solution solely based on computer recognition of the scripts; thereby eliminating human bias. Our programs can take as input script image sets and classify them using their similarity to standard symbols. The second contribution is to the statistical analysis and refinement of the Indus Valley Script. In the script set of around seven hundred, we were able to reduce the sample size by ten percent. In addition, we algorithmically remove symbols which were once thought to be different entities. Numerous scholars have debated over the script size and whether to scale the symbol set down. However, none have used an algorithmic process to remove and reduce the dataset. The third contribution of this dissertation is to the hierarchical tree showing the similarity between the ancient scripts. We also create a plausible hierarchical tree to indicate the evolutionary relationship of the ancient scripts.
Daggumati, Shruti R, "A Computational Framework for the Analysis of Ancient Scripts" (2019). ETD collection for University of Nebraska - Lincoln. AAI13861605.