2422. Each Decision Matters: Creating Word Embeddings as Part of a Historical Research Workflow
Invited abstract in session TA-12: AI for Optimization Modeling, stream Artificial Intelligence, Machine Learning and Optimization.
Thursday, 8:45-10:15Room: H10
Authors (first author is the speaker)
| 1. | Sophie Jasmin Spliethoff
|
| Department of History, Bielefeld University |
Abstract
This contribution aims at exploring the incorporation of word embeddings in historical research. It demonstrates how decisions made during the working process may affect resulting interpretations and emphasises the importance of designing interdisciplinary projects that combine expertise from different relevant fields throughout all research steps.
With the introduction of the printing press in the late 15th century, authors were suddenly able to publish smaller books, much cheaper and quicker than ever before. The emergence of this new medium and its usage to spread invectives in the context of the Reformation were mutually dependent. From a historian’s perspective, it is of great interest to find out to what extent contemporaries already linked ideas of media forms and functions during the 16th century. Rather than drawing conclusions from individual historical statements, word embeddings prove to be a suitable method in order to thoroughly explore links made between concepts.
Working with historical texts poses two main challenges: The available corpora are comparably small secondly and spelling was not standardised yet. Selecting and pre-processing input data are therefore essential to yield meaningful results. However, this entails many different options, for instance lemmatising vs. stemming terms, that must be carefully considered depending on individual research interests and resources. While digital methods may create the appearance of objectivity, this contribution sheds light on the impact of individual decisions on resulting interpretations and conclusions.
Keywords
- Artificial Intelligence
Status: accepted
Back to the list of papers