Sphinx Stemming "finicky"?

Question

I just installed the en.pak and (as per a recently posted question) I matched

Radiography

to

Radiograph
Radiographic
Radiograper

However I rotated again and now somehow it stopped working. The only error message I see is

"WARNING: index 'idx_X': dict=keywords and prefixes and morphology enabled, forcing index_exact_words=1"

I commented out morphology (#) just in case but still somehow and all of a sudden Radiography only finds itself.

Interesting update. I tested on

Run

and it found

Run
Ran
Running

and tested on

Office

and it found

Office
Offices

But it failed to get

Proctologist

When I searched on

Proctology

My settings is

morphology = lemmatize_en_all

and installed the en.pak so is there some 'advanced' setting needed to get stems from words like 'ology' since it clearly is getting basic plurals and lems for Ran/Run/Running?

Answer 1

It turns out that lemmatize is not ideal for the types of suffix matches I am looking for and that stem_en is a better choice for this. When I changed morphology to stem_en radiology matches radiologist and also did not match radiography