The corpus used for citation labelling and NER worked well with various examples, but text classification poses a few unique problems when using different types of decisions across different court levels. One of those problems is with how clearly the FIRAC elements come out in different types of decisions.
Working on the assumption that appellate decisions exemplify these elements better than trial decisions, a new corpus containing only SCC cases should work better for text classification.
The corpus used for citation labelling and NER worked well with various examples, but text classification poses a few unique problems when using different types of decisions across different court levels. One of those problems is with how clearly the FIRAC elements come out in different types of decisions.
Working on the assumption that appellate decisions exemplify these elements better than trial decisions, a new corpus containing only SCC cases should work better for text classification.