Seminar on Dec 19 th Abstracts & speaker information The First Online 3D Epigraphic Library: The University of Florida Digital Epigraphy and Archaeology Project Eleni Bozia (USA) Angelos Barmpoutis (USA) Robert Wagman (USA) Speaker information Eleni Bozia Department of Classics University of Florida Gainesville, USA bozia@ufl.edu Research fields Classics, Epigraphy, Archaeology In our presentation we intend to introduce the Digital Epigraphy Toolbox which is a crossplatform web-application designed to facilitate the digital preservation, study, and electronic dissemination of ancient inscriptions1. It allows epigraphists to digitize in 3D their epigraphic squeezes using our novel cost-effective technique, which overcomes the limitations of the current methods for digitizing epigraphic data in 2-dimensions only. Our toolbox contains several options for 3D visualization of inscriptions, as well as a set of scientific tools for analyzing the lettering techniques and performing quantitative analysis of the letterform variations. Furthermore, the users have the option to share their data, as well as search other uploaded collections of 3D inscriptions in a semi-supervised dynamic library. This dynamic library is organized thematically according to language, area of origin, and date and contains a comprehensive record of the inscription in the form of plain text, 3D model, 2D photographs, and other epigraphic information. Enhancing Digital Humanities Epigraphic analysis thus far has been limited to the study of published inscriptions in the form of simple text. This form of study, however, leaves room for subjective or even incorrect readings and interpretations of the text of the inscription. Also, the inaccessibility of the source text prohibits the correction of those mistakes and possibly the better identification of partially corrupted characters. A common way to overcome this limitation is the publication of photographic evidence along with the text. Even though this technique seemingly provides access to the inscription itself, the lighting conditions and the inability to capture the
tridimensional shape of the letters compromises the study of detailed characteristics, such as the lettering technique and even the dating of the inscription. Albeit the 3D anaglyph of the inscription can be captured in a squeeze, the key disadvantage of this method is the inaccessibility of the squeezes to the academic community. In our prototype technique the tridimensional digitization of squeezes is achieved through the bidirectional scanning of the squeeze using a typical 2D office scanner. The scanned images are then being processed by the algorithm that we developed which analyzes the depicted shading in the images and reconstructs in 3D the original inscription. The advantages of this process are numerous. It does not require any additional expensive equipment. The squeezes can now be preserved safely in a digitized form thus eliminating the possibility of deterioration of the squeezed paper. They can also be distributed electronically enhancing epigraphic studies. Furthermore, the digital squeezes can be visualized more effectively compared to 2D images, as they can be viewed from different angles, under different artificial lighting conditions, and in different zooming scales. More specifically, the user of the Digital Epigraphy Toolbox has the option to upload an inscription in various formats, such as scanned images of squeezes, photographs of inscriptions, or even 3D object files produced by 3D scanners, laser and depth scanners etc. He then has the option to reconstruct the tridimensional shape of the inscription from images, view, rotate, and zoom on the 3D model of the inscription, and apply different virtual lighting conditions. The user can also perform automatic post-processing analysis, namely segment the letters automatically and statistically compare the letterforms in a group of inscribed characters. The variability of letter forms is plotted as a comprehensible dendrogram. This tool is very useful especially in cases where the epigraphist needs to compare and analyze the letterforms of a large group of inscriptions. Finally, we have created a semi-supervised dynamic library of 3D epigraphic material that is organized thematically according to language, area of origin, and date. Each database entry contains a comprehensive record of the inscription in the form of plain text, 3D model, photograph of the original inscription, and other textual information about the inscription. Authorized users may upload more material, or edit existing records. Impact on the Humanities In conclusion, we would like to emphasize that the Digital Epigraphy Toolbox will serve uniquely epigraphists, archaeologists, classicists, as well as other scholars in humanities, as it will enhance the digital dissemination and study of inscriptions minimizing the disadvantages of the lack of physical contact between the scholar and the actual inscription. It will also promote comparative studies and collaboration between individual scholars and institutions that carry collections of squeezes by providing them a unique tool for sharing epigraphic material in a tridimensional electronic form with their colleagues. Preliminary results of our work have been presented in the 14th Congress of Greek and Latin Epigraphy hosted by Humbolt University in August 2012. Our results have already attracted attention from other institutions that led to collaborative research with the University of California, Berkeley and the University of Milan. Future Directions We plan to populate the database with 3D models of inscriptions and squeezes in collaboration with various universities and museums in Europe and North America. Furthermore, we are in the process of seeking funding so as to extend our 3D object analysis technique to other archaeological artifacts, such as Greco-Roman statues and vases. Footnotes 1 This project was in part funded by the National Endowment for the Humanities (Grant: HD-51214-11 June 2011-November 2012).
Relevant publications E. Bozia, A. Barmpoutis, R. S. Wagman, "The First Online 3D Epigraphic Library.", 14th International Congress of Greek and Latin Epigraphy, September 2012, Humboldt University, Germany A. Barmpoutis, E. Bozia, and R. S. Wagman. "A novel framework for 3D reconstruction and analysis of ancient inscriptions." Journal of Machine Vision and Applications, 21(6), 2010, pp. 989-998. Relevant links http://plaza.ufl.edu/bozia/epigraphy/ https://securegrants.neh.gov/publicquery/main.aspx?f=1&gn=hd-51214-11
Automatic Analysis and Synthesis of Complex Linguistic Networks Alexander Mehler (Germany) Speaker information Alexander Mehler Computer Science Goethe University Frankfurt am Main Frankfurt/Main, Germany Mehler@em.uni-frankfurt.de Research fields history, literary studies, German studies, Latin studies, linguistics, sociology With the advent of the Web 2.0 the rise of social-semiotic networks can be observed which because of their size, structure and complexity have been widely unknown so far. This relates to linguistic networks and their collaborative generation by large communities of interacting agents as, for example, the Wikipedia, social tagging systems and social networking sites. These networks induce a further level of information structuring far away from the focus of prevalent approaches to computational linguistics and computer science. With the availability of Web 2.0 corpora, complex networks as a whole and, thus, macro level structures come into the fore as a new research object hardly accounted for in the past. This seminar introduces into the analysis and synthesis of such social-semiotic networks. It is divided into three parts: 1. Starting from a short introduction into graph theory, the text-technological representation of complex linguistic networks is explained and demonstrated in detail. Using several large-scale examples (amongst others, numerous releases of the Wikipedia, special wikis, social ontologies as well as numerous literary texts) it is shown how to automatically induce graphs based on these corpora and how to process them. 2. The second part of the seminar demonstrates the quantitative analysis of such networks. Amongst others, models of stochastic physics (e.g., models of network motifs), quantitative sociology (e.g., center-periphery models) and quantitative linguistics (e.g., scale-free distributions) are introduced as well as explained with respect to their interplay. Special emphasis is put on an example-driven evaluation and practical demonstration of all models by means of large-scale data and its usage in the course of solving computational linguistic tasks (e.g., text classification). In this way, the seminar builds a bridge from complex network analysis to information retrieval by means of several applications such as lexical chaining, semantic search and document classification. In so far the seminar makes use of data of the humanities (e.g., corpora of literary texts), it demonstrates how insights into complex networks can be utilized to implement systems in computational humanities.
3. The third part of the course looks on complex networks from the point of view of generative models.it introduces, exemplifies and criticizes models of language evolution that have recently been proposed to model distributed cognition and language learning. As our reconstruction of these models starts from formal semantics, the seminar finally bridges to models that are prevalent in linguistics. All in all, this seminar brings together three perspectives on social-semiotic networks: their representation, analysis and synthesis. It puts special emphasis on historical literary documents in order to exemplify authorship attribution in terms of complex network analysis. This also relates to a network theoretical operationalization of the linguistic notion of intertextuality. In this sense, the seminar refers to computational linguistics and computational humanities as the preferred domains of exemplification and utilization of this content. Relevant publications Lücking, Andy und Alexander Mehler (2011). "A Model of Complexity Levels of Meaning Constitution in Simulation Models of Language Evolution". In: International Journal of Signs and Semiotic Systems 1.1, S. 18-38. Lücking, Andy und Alexander Mehler (2012). "What s the Score of the Naming Game? Constraints on Semantic Categorization". In: Proceedings of the 9th International Conference on the Evolution of Language (Evolang IX), Kyoto, Japan, S. 196-203. Mehler, Alexander (2008). "Structural Similarities of Complex Networks: A Computational Model by Example of Wiki Graphs". In: Applied Artificial Intelligence 22.7&8, S. 619-683. DOI: 10.1080/08839510802164085. Mehler, Alexander (2011). "A Quantitative Graph Model of Social Ontologies by Example of Wikipedia". In: Towards an Information Theory of Complex Networks: Statistical Methods and Applications. Hrsg. von Matthias Dehmer, Frank Emmert-Streib und Alexander Mehler. Boston/Basel: Birkhäuser, S. 259-319. Mehler, Alexander und Andy Lücking (2012). "WikiNect: Towards a Gestural Writing System for Kinetic Museum Wikis". In: Proceedings of the International Workshop on User Experience in e-learning and Augmented Technologies in Education (UXeLATE 2012) in Conjunction with ACM Multimedia 2012, 29 October - 2 November, Nara, Japan. Mehler, Alexander, Andy Lücking und Peter Menke (2012). "Assessing Cognitive Alignment in Different Types of Dialog by means of a Network Model". In: Neural Networks 32, S. 159-164. DOI: 10.1016/j.neunet.2012.02.013. Relevant links http://www.hucompute.org/