+
Overview
+
+ Welcome to GaLAHaD: Generating Linguistic Annotations for Historical Dutch. This page provides an overview of the
+ GaLAHaD application.
+
+
+ GaLAHaD is a platform that enables (non-technical) users to add linguistic annotation to (primarily) historical
+ Dutch documents, which can then in turn be inspected, evaluated and exported. Here follows a short introduction of
+ what GaLAHaD entails and what users can expect from this platform. Thereafter follows a short step-by-step guide
+ that takes you through the platform.
+
+
Context of the application
+
+ Linguistic annotation of (historical) texts is invaluable for various kinds of research and, of course, a
+ consistent and apt annotation with the least amount of errors is necessary if one wants to research (words in)
+ large amounts of text. However, lack of a central place where researchers were able to look at the possibilities
+ of linguistically enriching historical Dutch texts made it hard to find a good fitting model without substantial
+ effort and technical knowledge. So instead, researchers might use (or develop) a model for their own documents
+ without looking at other models.
+
+
+ Although researchers use their own models, there was a wish to be able to look at all the possibilities of
+ linguistically enriching historical texts and also to compare the different models used and to be able to use the
+ one(s) most suited for their specific data. Therefore, in the context of
+ CLARIAH (Common Lab Research Infrastructure for the Arts and
+ Humanities), GaLAHaD has been created at INT to fulfil in that need.
+
+
Intended use
+
For linguistic users
+
+ GaLAHaD aims to be a central platform that allows users to do various things. The main thing for researchers is to
+ linguistically enrich (historical) data (corpora). GaLAHad enables a user to upload documents (in several formats)
+ to a corpus. This corpus can then be enriched with one or more taggers; this is also possible if a corpus has
+ already been enriched. Each of these taggers (and also enrichment that was added prior to uploading the documents)
+ can then be compared and evaluated.
+
+
+ With several (visually helpful) criteria a user should be able to decide which tagger is the best for their data
+ in a reasonable timespan. It is also possible to download all evaluation criteria in separate Excel-files.
+ Furthermore, a user is then able to export their corpus to
+ BlackLab, to add some manual corrections (CoBaLT) or to download the enriched data (either as new documents or appended to the old documents). It should be noted
+ that individual steps take longer the more data are uploaded.
+
+
It also possible to make your corpus/corpora publicly available. TBD
+
For technical users
+
+ For technical users it is also possible to look at the performance of all taggers that exist in GaLAHaD.
+ Furthermore, it is also possible to add your own tagger to see how it compares to the other taggers. TBD
+
+
What can you expect of the Galahad?
+
+ GaLAHaD as a platform thus serves as a central hub for enriching historical Dutch. For instance, it allows users
+ to compare different taggers, to annotate data and to add your own material. Users can expect all of this to be
+ done rather easily and fast, though it should be noted that each operation takes more time, the bigger the
+ corpus/corpora (or documents) is/are. Besides ease of access, you can also expect good quality of linguistically
+ enriched data, though it of course depends how suited a tagger is for the specific data. Some taggers are more
+ suited (more trained) for older texts and some more for newer texts.
+
+
+ Even with the current state of the art, GaLAHaD can at best help to minimize the amount of manual corrections
+ needed. Depending on your use case, you should still expect that a significant post-processing step is required.
+
+
+
+
+
+
+
+ Help
+ About
+ Contribute
+ Lancelot
+
+
+