BeritJanssen
commented
1 month ago SCoT is not an algorithm, but an application, with multiple component, and heavy dependecy on JoBimText for pre-processing. I don't think it's worth investing time in applying the methods to current I-Analyzer corpora, especially seeing as the Jyväskylä group is building collaborations with Turku and Helsinki machine learning groups to investigate LLMs for parliamentary data.
Is your feature request related to a problem? Please describe. Word embeddings can be considered one way of finding synonyms, but give no way of distinguishing between antonyms and synonyms other than human inspection. A visiting scholar of the People & Parliament team suggested SCoT as an alternative, see this paper
Current plan For now, investigate the technique and give feedback to the Jyväskylä team about the feasibility of integrating this method.