Consider remove ontology definitions from pipeline representation serialization
Is your feature request related to a problem? Please describe. The current ontology serialized IR looks like the following:
components:
- configs:
serialize_method: jsonpickle
zip_pack: false
type: forte.data.readers.string_reader.StringReader
- configs:
overwrite: false
user_input: bye
selector:
configs: {}
type: forte.data.selector.DummySelector
type: advanced_pipeline_test.UserSimulator
forte_ir_version: 0.0.1
states:
attribute:
_check_type_consistency: true
_do_init_type_check: true
_enable_profiling: true
_initialized: true
resource:
merged_entry_tree:
attributes: []
children:
- attributes: []
children:
- attributes:
- lemma
- ud_features
- is_root
- pos
children: []
name: ft.onto.base_ontology.Token
- attributes: []
children: []
name: ft.onto.base_ontology.Document
- attributes:
The merged_entry_tree section is quite long and may not really be necessary. Especially if we are able to infer the ontology tree via https://github.com/asyml/forte/issues/583, we can drop this section and make the IR to be more focused on its main objectives.
Describe the solution you'd like After we finish https://github.com/asyml/forte/issues/583, during the pipeline save function, we should drop this part. Then we should make sure the init_from_config works correctly with this new change.
Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered.
Additional context Add any other context or screenshots about the feature request here.
As discussed here: https://github.com/asyml/forte/issues/583, it is unknown how to get the tree now, so we need to find an alternate solution
As discussed here: #583, it is unknown how to get the tree now, so we need to find an alternate solution
Based on the current status. Is it ok to remove me as the assignee?