This paper introduces a novel method for addressing the challenge of deciphering ancient scripts. The approach relies on combinatorial optimisation along with coupled simulated annealing, an advanced technique for non-convex optimisation. Encoding solutions through k-permutations facilitates the representation of null, one-to-many, and many-to-one mappings between signs. In comparison to current state-of-the-art systems evaluated on established benchmarks from literature and three new benchmarks introduced in this study, the proposed system demonstrates superior performance in enhancing cognate identification results. Keywords: ancient script decipherment, combinatorial optimization, k-permutations, coupled simulated annealing, evaluation benchmarks status released display-pdf yes is-olf no is-manuscript no is-preprint no is-journal-matter no is-scanned no is-retracted no Received 2025 Feb 21; Accepted 2025 Apr 23; Collection date 2025. 1 Introduction Numerous ancient scripts around the world remain undeciphered, with many of them dating back millennia. The challenges in deciphering these scripts stem from factors such as insufficient inscriptions, the absence of known language descendants utilizing these scripts, and uncertainty about whether the symbols truly form a writing system. In literature, numerous contributions address these subproblems, offering computational methods tailored to each, frequently focusing on a particular script. The sequential tasks typically involve: (a) determining if a set of symbols genuinely constitutes a writing system, followed by (b) devising procedures to segment the symbol stream into individual signs. Subsequently, (c) reducing the set of signs to the minimal collection for the given writing system, thereby forming the alphabet (or syllabary, or sign inventory), and identifying all allographs. This task proves intricate due to variations introduced by scribe writing styles and the evolution of symbols over time, complicating the identification and management of allographs. In addressing this challenge, Skelton ( 2008 ) and Skelton and Firth ( 2016 ) applied phylogenetic systematics to the realm of writing systems. Their focus was particularly on Linear B, a pre-alphabetic Greek script. Through this method, they scrutinized the evolution of the Linear B script over time, taking into account scribal hands as an additional source of variation. This application showcased the efficacy of phylogenetic analysis in understanding the development of writing systems. Born et al. 1.5 Define signs values and match sign sequences with a known language Every contemporary endeavor to decrypt ancient scripts using computational tools relies on contrasting a missing script or language wordlist with words from a deciphered and known language. In this system, a recurrent Neural Network (NN) is employed to establish the mapping between lost and known signs, despite the advantage of using contextual information to perform the task, it lacks the adaptability necessary for addressing two practical decipherment challenges. Firstly, paleographers often possess partial knowledge about the mapping of certain signs, and this information needs to be incorporated into the system. Secondly, real inscriptions are frequently broken or damaged, leading to unreadable signs, requiring the incorporation of uncertainty into the system, potentially through the use of wildcards or other special symbols. We reduced the number of non-cognate words, creating a dataset of 2,214 cognate words, 1,119 unpaired Ugaritic words, and 1,108 Old Hebrew words without corresponding cognates. These words were randomly selected from the dataset proposed in Snyder et al. ( 2010 ). Linear B/Mycenaean Greek - LB/MG . Linear B, a syllabic writing system employed for Mycenaean Greek dating back to approximately 1450 BC. Luo et al. ( 2019 ) curated a dataset by extracting pairs of Linear B and Greek words from a compiled lexicon, eliminating some ambiguous translations and resulting in 919 cognate pairs. Inside round parentheses, the maximum Accuracy value obtained in our experiments is indicated. † Results for NeuroCipher computed or recomputed by us simulating a real setting and using the code in Luo et al. ( 2019 ). ‡ To enable the system to converge toward meaningful results we had to provide the number of cognates in the dataset, information not available in real settings. Our system exhibits superior accuracy compared to any other work across almost all benchmark datasets, with a substantial margin. (b) Access to an extensive cognate list is crucial, yet in most real cases, only two word lists are available for matching, without any assurance that cognates from the lost language truly exist in the lexicon of the known language. (c) In natural language processing (NLP), evaluations are typically conducted on well-established test beds and the studies discussed earlier focused on well-known correspondences to demonstrate system effectiveness. On the contrary, testing these systems on real cases involving unknown writing systems and their corresponding languages presents an entirely different set of challenges and uncertain comparanda. We aim to contribute insights that may finally address longstanding problems unresolved for centuries. Funding Statement The author(s) declare that no financial support was received for the research and/or 4 https://github.com/ftamburin/EditDistanceWild 5 https://github.com/structurely/csa 6 The Tower of Babel, https://starlingdb.org . Data availability statement The datasets and codes presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://github.com/ftamburin/CSA_OptMatcher . Author contributions FT: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Validation, Writing – original draft, Writing – review & editing.

Deciphering Ancient Chinese Oracle Bone Inscriptions Using Case-Based Reasoning | Springer Nature Link Skip to main content Advertisement Deciphering Ancient Chinese Oracle Bone Inscriptions Using Case-Based Reasoning Conference paper First Online: 10 September 2021 pp 309–324 Cite this conference paper Save conference paper View saved research Case-Based Reasoning Research and Development (ICCBR 2021) Abstract Ancient oracle bone inscriptions (OBIs) are important Chinese cultural artefacts, which are difficult and time-consuming to decipher even by the most expert paleographers and, as a result, a large proportion of excavated OBIs remain unidentified. In practice, OBIs are deciphered by translating between different writing systems; Chinese writing systems have evolved over time and ancient OBIs can be deciphered by translating their inscriptions to a known inscription in an adjacent writing system, but this is a complex and time-consuming process. In this paper we propose a novel case-based system, to support this task, allowing a paleographer to present an unknown inscription (image) as a query, to receive a set of similar images from an adjacent writing system with associated scholarly information, and so help guide the deciphering of the query. One important contribution of this work involves the use of an auto-encoder to learn suitable image representations to capture the relationship between two adjacent writing systems. We demonstrate the effectiveness of this approach using a novel, purpose-built case base, and discuss its use in a paleographic setting. This is a preview of subscription content, log in via an institution to check access. Access this chapter Log in via an institution Subscribe and save Springer+ from €37.37 /Month Starting from 10 chapters or articles per month Access and download chapters and articles from more than 300k books and 2,500 journals Cancel anytime View plans Buy Now Chapter EUR 29.95 Price includes VAT (Indonesia) Available as PDF Read on any device Instant download Own it forever Buy Chapter eBook EUR 64.19 Price includes VAT (Indonesia) Available as EPUB and PDF Read on any device Instant download Own it forever Buy eBook Softcover Book EUR 74.99 Price excludes VAT (Indonesia) Compact, lightweight edition Free shipping worldwide - view details Buy Softcover Book Tax calculation will be finalised at checkout Purchases are for personal use only Institutional subscriptions Similar content being viewed by others An open dataset for oracle bone character recognition and decipherment Article Open access 06 September 2024 Puzzle Pieces Picker: Deciphering Ancient Chinese Characters with Radical Reconstruction Chapter © 2024 Recognition of Oracle Bone Inscriptions by using Two Deep Learning Models Article 05 April 2022 Explore related subjects Discover the latest articles, books and news in related subjects, suggested using machine learning. Archaeological Methodology Forensic Archaeology Palaeography Prehistoric Archaeology Reverse engineering Computational Anthropology Deep Learning Techniques for Oracle Bone Inscriptions Recognition Notes 1. http://dx.doi.org/10.17632/ksk47h2hsh.2 . 2. https://hanziyuan.net/ . 3. https://www.zdic.net/ . 4. https://github.com/ICCBR/AncientDiscovery . References Artetxe, M., Labaka, G., Agirre, E., Cho, K.: Unsupervised neural machine translation. arXiv preprint arXiv:1710.11041 (2017) Boltz, W.G.: Early Chinese writing. World Archaeol. 17 (3), 420–436 (1986) Article Google Scholar Brown, P.F., et al.: A statistical approach to French/English translation. In: RIAO, pp. 810–829 (1988) Google Scholar Collins, B., Cunningham, P., Veale, T.: An example-based approach to machine translation. In: Conference of the Association for Machine Translation in the Americas (1996) Google Scholar Derpanis, K.G.: Mean shift clustering. In: Lecture Notes, p. 32 (2005) Google Scholar Dorr, B.J., Jordan, P.W., Benoit, J.W.: A survey of current paradigms in machine translation. Adv. Comput. 49 , 1–68 (1999) Article Google Scholar Feng, G., Jing, X., Yong-ge, L.: Recognition of fuzzy characters on Oracle-bone inscriptions. Editor information Editors and Affiliations Universidad Complutense de Madrid, Madrid, Spain Antonio A. Sánchez-Ruiz Knexus Research Corp., National Harbor, MD, USA Michael W. Floyd Rights and permissions Reprints and permissions Copyright information © 2021 Springer Nature Switzerland AG About this paper Cite this paper Zhang, G., Liu, D., Smyth, B., Dong, R. (2021). Deciphering Ancient Chinese Oracle Bone Inscriptions Using Case-Based Reasoning. In: Sánchez-Ruiz, A.A., Floyd, M.W. (eds) Case-Based Reasoning Research and Development. ICCBR 2021. Lecture Notes in Computer Science(), vol 12877. Springer, Cham.

The Phaistos Disk is an ancient artifact from Crete. At each side of the disk, a series of unknown signs is written along a spiral. Professional archaeologists expect that we will only learn what it is until similar objects are found. A statistical analysis in this article shows what it is not: it is not a one‐dimensional text, since there are relations between the signs in adjacent windings of the spiral. Three patterns of such relations have been identified. A Monte Carlo simulation of one of them has been performed, using a model of the spiral form. It is concluded that the probability of this pattern being coincidental is small, well below the conventional threshold.

Oracle bone inscriptions (OBIs) are ancient Chinese scripts originated in the Shang Dynasty of China, and now less than half of the existing OBIs are well deciphered. To date, interpreting OBIs mainly relies on professional historians using the rules of OBIs evolution, and the remaining part of the oracle’s deciphering work is stuck in a bottleneck period. Here, we systematically analyze the evolution process of oracle characters by using the Siamese network in Few-shot learning (FSL). We first establish a dataset containing Chinese characters which have finished a relatively complete evolutio