Perform Named Entity Recognition (NER) on french novels from the roman18 corpus with the help of SpaCy in the context of the "Mining and Modeling Text" project (University Trier). www.mimotext.uni-trier.de
The roman18-corpus of about 100 eighteenth-century French novels in full text is analysed via SpaCy’s named entity recognition. Named entity recognition (NER) is a popular information retrieval technique “to identify and segment named entities and classify or categorize them under various predefined classes” (Sarkar, 2019).
A sentence from Denis Diderot's novel La Religieuse (1796), proceeded with Spacy / displacy
Within the French language package of SpaCy one can distinguish the following types of named entities: LOC, PER, MISC and ORG entities. The five most common “LOC” (location) entities within each novel and their numerical occurences per text are extracted.
The script which performs the NER and its technical documentation is under scripts/. The latest results of this script are stored under raw_results/, although this is configurable. These results still need manual corrections and supplements to be suitable for import into the MiMoText Wikibase. The latest version of this is under edited_for_import/. In this folder there is also a file named edits_openrefine.txt. This has been generated by OpenRefine which is used for semi-automatic reconciliation and includes all the editing steps in this tool.
We utilise the roman18-corpus, which contains french novels 1750-1800 in XML/TEI and plain text. We utilised the novels in modernised plain text.
Software in this repo, unless specified otherwise, is made available under the MIT license. We don’t claim any copyright or other rights on the metadata. If you use our scripts or results, for example in research or teaching, please reference this repository using the citation suggestion below.
NER on Eighteenth-Century French Novels (1750-1800), edited by Julia Röttgermann, Johanna Konstanciak and Henning Gebhard. Trier: TCDH, 2022. URL: https://github.com/MiMoText/NER_novels.