Soundex Search bugs
Brought to you by:
canajun2eh,
yalnifj
1. Soundex search does not work on 2 ROMN names and
probably neither on 2 PHON names.
2. In DM Soundex when a name consists of more than one
word, it should be coded as if one word, such as Nowy
Targ, which should be treated as Nowytarg.
Today it seems that each word it searched separately.
Logged In: YES
user_id=959928
3. When the DM Soundex coding array has a zero in the 1st
position of a group of 3 values, to be used at the begining
of the name, this zero should be part of the calculated DM
Soundex code.
Today it is not included in a similar way to the -1 values
that should not be included in the calculated DM Soundex
code.
Logged In: YES
user_id=959928
point 3 is corrected
Logged In: YES
user_id=300048
#1 should be fixed with the new soundex code in the SVN as
the soundex is calculated for every name both in the
individuals and the names table.
#2 I think it should calculate each name seperately. For
example if we take the name Joseph John Finlay. I would
expect to find him by searching on any of the following
examples:
Joseph Finlay
John Finlay
John Joseph Finlay
Juhn Jusoph Finley
If we combined the names together we wouldn't be able to
search for the names in reverse order.
--John
Logged In: YES
user_id=918839
Dear user,
Can you please test the provided solution and let us know if
this
solved your issue? We can then either continue working on
your
bug or close it.
With kind regards,
The PGV Team
Logged In: YES
user_id=959928
2. will verify each name separately and all the name words
together so that for instance for Roche Lea both Rocke, Lea
and Rochelea will catch the name.
The solution did write into the database only one GM-code.
I have sent John offline comments and testdata for the after
4.0 development.
Will test it as soon as it is ready for testing.
Logged In: YES
user_id=634811
any update?
Logged In: YES
user_id=959928
I sent the following to John on June 18 before I knew the
students finished their semester (I have adjusted the note a
little).
John,
A few issues that I have seen so far on my 4.1 Future site
.../future/phpGedView/.
1. I think that we should save any DM or Basic Russell code
only once in the soundex DB table and in the search name code.
I see now for Jankeloff I118 the following DM codes
165870:465870:565870:658700:165870:465870:565870:658700
It is not clear to me why we also see two identical Basic codes.
2. How do you know which soundex codes relate to which
individual names?
3. I search on Andrej on the Soundex.ged file
In 4.0 two Latin and 2 Hebrew names are returned
In 4.1 only the Hebrew names are returned.
4. I search on Zizilnitzky only on the Soundex.ged.
I receive errors with index 1, 2 or 3
ERROR 8: Undefined index: 1
0 Error occurred on in function soundexsearch
1 called from line 949 of file search_ctrl.php in function
soundexsearch
2 called from line 308 of file search_ctrl.php in function init
3 called from line 2035 of file search_ctrl.php in function
require_once
4 called from line 28 of file search.php
I have on my 4.1 code site also the TT A.ged file.
The 4.1 shows me also all the individuals from TT A.ged
whose name has a DM soundex code of 448645.
The 4.1 shows me also all the individuals from Soundex.ged
whose ID is the same as the ID of the 448645 individuals
from TT A.ged.
Their name is shown as (unknown) (unknown) although this is
not their name.
The shown birth and death data of these Soundex.ged
individuals is that of the parallel TT A.ged individuals.
4.0 returns only I1, I4 and I8.
5. I search on Szczuczyn
Receive error messages
ERROR 8: Undefined index: 1
0 Error occurred on in function soundexsearch
1 called from line 949 of file search_ctrl.php in function
soundexsearch
2 called from line 308 of file search_ctrl.php in function init
3 called from line 2035 of file search_ctrl.php in function
require_once
4 called from line 28 of file search.php
4.0 returns also I9173, I5004 and Hebrew I4717. These are
not returned in 4.1.
6. I search for Jankeloff
I am shown 2 correct Soundex.ged individuals with code 165870.
I am also shown the same Ids from TT A.ged
In 4.0 code I am also shown individuals with code 658000.
4.1 code does not return them.
7. Search on Джеймс returns nothing in 4.1.
In 4.0 I see I4864.
The code of the name should be 344640.
This individual has this code in the soundex table.
I have not yet looked at the Places.
Regards,
Meliza
Logged In: YES
user_id=959928
I have posted the DM Soundex improvement request in RFE 1511090.
Logged In: YES
user_id=959928
8. When I search on my test site English page for a place
only I see
SELECT i.i_id, i.i_file, i.i_name FROM pgv_places JOIN
pgv_placelinks ON p_id = pl_p_id JOIN pgv_individuals as i
ON pl_gid = i_id WHERE p_file = 10 AND (p_dm_soundex =
'664600' )
This should be commented out.
8. The query does return all the individuals with the same
ID as the person in our selected gedcom file that has a
soundex match. Also i_file should be added to the query.
This problem exists both for name and for place DM Soundex
searces.
9. When I search for Latin or Hebrew first name only,
nothing is returned.
Logged In: YES
user_id=959928
seems there are two number 8s.
10. With the many additional search results from other
gedcom files it is a little hard to tell, but it seems to me
that the results are no longer sorted.
The results should be sorted (at least in most cases and as
a default, if this will be optional) by the given name.
Logged In: YES
user_id=959928
In the places table I see only one DM code value and there
are cases where it is not the correct one.
Some samples:
Kauhava - DM code should give 557000, we see 500000.
Seems that the soundex algorithm returns 500000, 550000 and
570000 instead of the correct value 557000.
We agreed that names that contain two words would be coded
so that each word is done separately and they are coded
together.
Here we see only the combined code.
Karlovy Vary - DM code should give 598700, 790000 and
598779, we see 598779. Same for the Basic codes - we see
only K641, should also have V600.
Sosnowiec - should give 446750, 446740, 446500 and 446400,
we have only 446400 in the table.
Seems that the Soundex algorithm returns only 446400.
Vyborg should return 779400 and 779500. We see only 779400.
Wyszogrod should return 744930 and 745930. We have only
744930 in the table.
Kajaani, should return 516000, we have 560000.
Seems a soundex algorithm problem.
Logged In: YES
user_id=959928
11. It would ease testing if the debug texts would print
when DEBUG=1 is added to the URL.
I do not know how to do this.
Logged In: YES
user_id=959928
Points 4. 5. and 8. are corrected.
The 4.1 Soundex Search results are still off and require
correction.
Logged In: YES
user_id=959928
Originator: YES
When can we test the Soundex corrections?
I see that some issues are already solved.
For instance I no longer see results of other, not selected Gedcoms,point 6.
8. I no longer see the error.
But when I search for place Helsinki without filling anything in Surname, I am returned 3 Hebrew names that do not have any place.
and so on
Logged In: YES
user_id=959928
Originator: YES
Much better!
How should I see the Soundex codes of the names/places that I fill when I debug? Adding &DEBUG=1 seems not yet to work.
What is the status of the Soundex RFE?
1. When I search on a name AND place or a place AND year etc., I expect to see in the result only data that fulfills all the search criteria. This does not seem to work.
2. When I search for place Karlovi I do not see records with place Karlovy Vary.
We agreed that we would search each word separately and the words combined together.
I see still only codes in the table for the combined text.
3. My active Gedcom is Soundex and I search on it and on the big Gedcom file.
I imported today both these files.
I fill Helsinki in the place – nothing is returned!!
Helsinki, Finland returns neither anything.
When the active Gedcom is the big file, lots of records are returned for Helsinki.
4. Searching onמליצה does not return Meliza. מליסה and Meliza return both the Hebrew and the Latin names???
5. I search on Andrej on the Soundex.ged file In 4.0 two Latin and 2 Hebrew names are returned In 4.1 only the Hebrew names are returned.
6. Soundex file search on surname Джеймс should return I4864. It is not returned. The soundex code of the name exists in the table.
In 4.0 I see I4864.
The code of the name should be 344640.
Logged In: YES
user_id=634811
Originator: NO
There have been some soundex fixes recently. Is this fixed now?
Logged In: YES
user_id=959928
Originator: YES
2. We agreed that we should be able to find the place or name by combined names or by just one of the names.
The combined name seems to work, but not one name of the combination.
Logged In: YES
user_id=959928
Originator: YES
We are almost there
2. 3. on place does not yet work.
11. debugging?
RFE 1511090
I've just rewritten the soundex search function. Provided the underlying soundex() and DMSoundex() functions work, the output should work.
This Tracker item was closed automatically by the system. It was
previously set to a Pending status, and the original submitter
did not respond within 100 days (the time period specified by
the administrator of this Tracker).