This paper describes a data mining study of a set of ancient scripts in order to discover their relationships, including their possible common origin from a single root script. The data mining uses convolutional neural networks and support vector machines to find the degree of visual similarity between pairs of symbols in eight different ancient scripts. Among the surprising results of the data mining are the following: (1) the Indus Valley Script is visually closest to Sumerian pictographs, and (2) the Linear B script is visually closest to the Cretan Hieroglyphic script.

The recent surge in ancient scripts has resulted in huge image libraries of ancient texts. Data mining of the collected images enables the study of the evolution of these ancient scripts. In particular, the origin of the Indus Valley script is highly debated. We use convolutional neural networks to test which Phoenician alphabet letters and Brahmi symbols are closest to the Indus Valley script symbols. Surprisingly, our analysis shows that overall the Phoenician alphabet is much closer than the Brahmi script to the Indus Valley script symbols. CCS CONCEPTS • Computing methodologies → Machine learning; Machine learning approaches;

This work describes a general method of testing for redundancies in the sign lists of ancient scripts by data mining the positions of the signs within the inscriptions. The redundant signs are allographs of the same grapheme. The method is applied to the undeciphered Indus Valley Script, which stands out from other ancient scripts by having a large proposed sign list that contains dozens of asymmetric signs that have mirrored pairs. By a statistical analysis of mirrored asymmetric signs, this paper shows that the Indus Valley Script was multi-directional and the mirroring of signs often denotes only the direction of writing without any difference in meaning. For this and five other specific reasons listed in the paper, 50 pairs of signs, 23 mirrored, and 27 non-mirrored, can be grouped together because each pair consists of only insignificant variations of the same original sign. The reduced sign list may make decipherment easier in the future.

No abstract

This paper analyzes the relationships among eight ancient scripts from between Greece and India. We used convolutional neural networks combined with support vector machines to give a numerical rating of the similarity between pairs of signs (one sign from each of two different scripts). Two scripts that had a one-to-one matching of their signs were determined to be related. The result of the analysis is the finding of the following three groups, which are listed in chronological order: (1) Sumerian pictograms, the Indus Valley script, and the proto-Elamite script; (2) Cretan hieroglyphs and Linear B; and (3) the Phoenician, Greek, and Brahmi alphabets. Based on their geographic locations and times of appearance, Group (1) may originate from Mesopotamia in the early Bronze Age, Group (2) may originate from Europe in the middle Bronze Age, and Group (3) may originate from the Sinai Peninsula in the late Bronze Age.