Exploring Concept Change and Transfer in Antiquity (eXChange)
The eXChange research project aims to provide innovative contributions to the systematical exploration of concept change and meaning shift within the textual heritage of the classical ancient world. eXChange is a joint research project of the NLP Group (ASV), Visualization Group (BSV) and the Classical Studies department (Ancient History, Ancient Philology) at the University of Leipzig.
The methodological interconnection of Text Mining, Visual Analytics and Historical Discourse Analysis allows for new perspectives on existing digitized corpora of ancient Greek and Latin texts, dating back to the 8th century BC. Besides the multilingual and very diachronic expanse of those voluminous textual resources there are further particularities, which arise from embedding them into the context of research: One issue is the fragmentary nature of metadata for single documents and their associated authors. The assignment of authors is by itself sometimes very vague and speculative in nature - as is the dating and geographical referencing. In most cases the metadata itself plays an important role within a historical hypothesis. For eXChange it is therefore important to allow for a dynamic metadata management. All metadata entries should be alterable on a per-user-basis at any time, resulting in corresponding "realtime" changes of the presented data view and text mining measures.
Therefore the acquisition and management of metadata has to be treated as separate task of user-centric knowledge management. As an immediate consequence of this flexible faceted approach to corpus annotation, there emerge significant technological challenges for Text Mining methods. This is especially observable in the areas of indexing and structuring of the data as well as in the design of efficient access patterns and interfaces.
The final open source licensed research system will consist of a visualization-driven web application backed by the above-mentioned Text-Mining-enabled data and metadata storage. It will allow for the creation, formalization, checking and adjustment of historical hypotheses concerning the concepts expressed in the ancient sources. Beyond that, the system will help researchers to explicate results and building blocks of their hypotheses and will allow them to collaboratively refine them in an appropriate form. This functionality is rooted in an adapted form of a concept-based corpus query, which allows for a domain-specific conceptualization of certain terms of interest beyond language boundaries.
Employing this abstraction of a simple full-text search yields inevitably a deeper critical examination of the recorded texts. The direct user interaction with concepts through corresponding textual occurrences, augmented by interactively displayed statistical measures in context allows for an immersive workflow known from other Visual-Analytics applications. This stimulates historical hypothesis building in way that embeds quantitative and qualitative concerns.