Given the problems of uncertainty and vagueness we argue that it is appropriate to use Text Mining methods to the semantic analysis of historical texts.
After addressing the challenges of historical spelling variants and OCR errors, we show that document classification attains high accuracy, and that the feature weights can be interpreted historically and linguistically, although with a high level of noise. Further, we were surprised how accurately topic models allow us to trace socio-historical changes, for example the change from scholastic thinking to empirical science in medical studies, and how professional health care replaced medieval quackery. Conceptual maps using kernel density estimation also led to clear results, with the disadvantage that topics are less clearly apparent. Both of these approaches are robust to parameter details, as long as stopword lists are used and OCR errors or historical spelling variants are addressed. Our results on using fasttext are mixed, however. ...
Dimensions of textual preprocessing: normalization of spelling variants, lemmatization, stop word removal, OCR error treatment.
Quantitative method: classical probabilistic topic modeling realizing an unsupervised soft-clustering of textual regions. Using the byproducts of supervised text classification mapping  text regions into their epoch of origin: for logistic regression on word unigrams, the byproducts are word weights, for modern distributional approaches with task-specific word embeddings, the byproducts are the similarity of the derived dense continuous word representations.
Dimensions of distributional approaches are windows of context (sentence, page, article, book). Dimensions of text classification approaches are the binning of periods (10-60 year) and the amount of sub-sampling (stratification) of the available material. Older periods typically have a lot less text material available.
Given that each method has its strengths and its idiosyncrasies, applying a broad variety of quantitative approaches and being able to compare and inspect the different outputs 
SC: Outlook: Methoden, welche LDA und EMbeddings kombinieren? LDA2vec? https://lda2vec.readthedocs.io/en/latest/
\cite{Blei2006}
NOTES:
CFP WAS:
Call for Proposals
Workshop on Computational Methods in the Humanities 2018 (COMHUM 2018)
Workshop date: June 4–5, 2018 Location: University of Lausanne,
Switzerland
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
It is often said that the digital humanities are “situated at the
intersection of computer science and the humanities,” but what does this
mean? We believe that the point of using computers in the humanities is
not just to automatically analyze larger amounts of data or to
accelerate research. We therefore prefer to understand digital
humanities as (1) the study of means and methods of constructing formal
models in the humanities and (2) as the application of these means and
methods for the construction of concrete models in particular humanities
disciplines. The central research questions are thus correspondingly (1)
which computational methods are most appropriate for dealing with the
particular challenges posed by humanities research, e.g., uncertainty,
vagueness, incompleteness, but also with different positions (points of
view, values, criteria, perspectives, approaches, readings, etc.)? And
(2) how can such computational methods be applied to concrete research
questions in the humanities?
The goal of this workshop is to bring together researchers involved with
computational approaches in the humanities with the objective of
stimulating the research and exchange around innovative,
methodologically explicit approaches, to encourage discussion among
researchers and developers from different communities, and to help
bridging the divide that still exists between the different disciplines
involved in this field.
The program will consist of invited and contributed talks on
computational methods for and in the humanities. The official language
of the workshop is English. Contributions can be submitted in English or
French.
The workshop is organized by the Department of Language and Information
Sciences at the University of Lausanne, with the support of the Faculty
of Arts. The workshop underlines the commitment of the Department of
Language and Information Sciences to the computational dimension of the
digital humanities, including formal and mathematical methods.
Topics
══════
The topics of the workshop encompass formal and computational aspects
related to the development and use of computational methods in the
humanities (in particular the disciplines represented in the Faculty
of Arts of UNIL - such as literature, linguistics, history, history of
art, cinema studies, game studies).
Topics include, but are not limited to:
• Theoretical issues of formal modeling in the humanities
• Knowledge representation in the humanities
• Data structures addressing specific problems in the humanities
   (including text and markup)
Quantitative methods in the humanities
• Computer vision and image analysis in the humanities
• Spatial analysis in the humanities
• Network analysis in the humanities