Cases when the Deconstructor in lookup has no Construction equivalent in headwords

[vjjda]

Bhante @bdhrs,
To continue the discussion with you on email about whether deconstructor is necessary when headwords available.
I've made some sql queries to show the cases when construction for headwords is missing.

SELECT
    d.id,
    l.lookup_key,
    d.lemma_1,
    l.deconstructor,
    d.construction
FROM dpd_headwords d
LEFT JOIN lookup l
ON
    l.lookup_key = (
        CASE
            WHEN INSTR(d.lemma_1, ' ') > 0 THEN SUBSTR(d.lemma_1, 1, INSTR(d.lemma_1, ' ') - 1)
            ELSE d.lemma_1
        END
    )
WHERE l.deconstructor IS NOT NULL
AND l.deconstructor != ''
AND d.lemma_1 IS NOT NULL
AND d.lemma_1 NOT LIKE '% % %'
AND d.construction = '' -- comment out to see the result with construction
-- AND l.deconstructor LIKE '%,%' --uncomment to see deconstructor with multiple possibilities
;

The result?
There're like only 40 words not covered!
I've attached the result here:
construction missing.csv

Bhante can also comment out the line

AND d.construction = ''

to see how well the construction is, compared to the deconstructor.
I've attached here:
construction vs deconstructor.csv.

Add a line:

AND l.deconstructor LIKE '%,%'

and we can see the cases where the deconstructor offers two possibilities or more, compared with construction:
deconstructor possiblities vs construction.csv.

I found the results are rather interesting and I was thinking you might be interested in the analysis, too.

[bdhrs]

@vjjda thanks, finally got a chance to study this data. Most of it is expected, the way that the deconstructor tries to solve every possibility, whereas a headword is a specific possibility with an example to back it up. The [construction missing.csv](https://github.com/user-attachments/files/24360118/construction.missing.csv) is nice n easy low hanging fruit to add as headwords, ta.