filters = Filter.by_property(“project_id”).not_equal(“project_id_1”)
but still get these records
{‘file_id’: ‘file_id_1’, ‘auth_level’: 1, ‘text’: ‘text_7591’, ‘project_id’: ‘project_id_1’}
mean time, equal filter works
filters = Filter.by_property(“project_id”).not_equal(“project_id_1”)
but still get these records
{‘file_id’: ‘file_id_1’, ‘auth_level’: 1, ‘text’: ‘text_7591’, ‘project_id’: ‘project_id_1’}
mean time, equal filter works
Hi @shadowlin !
One configuration that directly influentiate this outcome is tokenization:
so by default the tokenization is word
, that will:
Keep only alpha-numeric characters, lowercase them, and split by whitespace.
So you will probably want to define the tokenization of that property to field
, that will:
Index the whole field after trimming whitespace characters.
Let me know if this helps
I did set this when create collection:
classes.Property(
name=“project_id”,
skip_vectorization=True,
index_filterable=True,
index_searchable=False,
data_type=classes.DataType.INT
)
should this disable tokenization?