Reporting a bag: problem with surname page.

[54]? - A simpler more understandable example - Click on the "Learn more about the Чернявский Surname" link on Юрий Петрович Чернявский's profile.

[54] == Mike Stangel

Thanks @Bjørn P. Brox

Thanks, we noticed that as well on Friday. The problem isn't with the surname page itself but rather with out search engine, which has trouble with non-Latin characters (you've probably noticed that searching for those names is a bust, too). We're discussing a couple possible solutions.

You should also discuss the removal of accents in the mapping of profiles to a surname page (anglicisation).

In my opinion you should NOT.

As in the Sunde/Šunde discussion we have had, finding Norwegian Sunde profiles linked to the Croatian Šunde surname page is a bit strange but it is not strange for ex-Croatians which have converted their name to Sunde to get linked to the Sunde surname page (or even Šunde).

It is much better to maintain two such surname pages where we cross-link to the other page if needed, as we have done here:

http://www.geni.com/surnames/Sunde
http://www.geni.com/surnames/%C5%A0unde

Mike Stangel, Private User Please consider for solution to automate similar and familiar surname linking.

Look at my surname example: http://www.geni.com/surnames/chernyavskyy (and this list is not completed)

To many similar surname pages to update manually. To many surnames to search for relatives. And, it is an issue to find the tree matches automatically.

What should be the rules?

I and most of my family would protest loudly if Brox was automatically linked to Brooks surname page and opposite.

It is much better to do it manually by someone familiar to the variations like you have done in your http://www.geni.com/surnames/chernyavskyy example.

I agree with you Private User, that linking of similar surnames is better to make by community, not by Geni. But I argue that Geni should be able to use thous links in search and tree matches features.

Another point: Currently, this surname page http://www.geni.com/surnames/chernyavskyy has more then 15 links on similar surname pages, probably in future it will be more. So, do users should update manually "similar surnames section" ON EVERY SURNAME PAGE? I don' t think so. Clearly, it is better to automate this routine. But, again, I agree that links should be defined manually, by community.

I agree that Geni and users should be able to add and maintain similar names rules, even including automatic rules like surnames ending -sen/-søn/-son/-sønn in Scandinavia should be treated equally and a similar approach like your example with variations on -avsky.

Using Metaphone/Soundex or removing/accent folding to match surnames as "rules" should however be avoided.

If rules should be extended to common nicknames is however more difficult, like Robert versus Bob.

Bjørn: I'm missing the connection between nicknames and surnames.

Not necessary a connection, but from an earlier discussion about searching profiles people have argued that common US nicknames should automatically have a match both on hot-matches and in search, for example Bobby Kennedy should match Robert. Kennedy.

Bjørn P. Brox, Absolutely.

As for Metapone and other algorithms you mentioned - I do not know a lot how it works on Geni now, interesting to know more about and why you think it should be avoided? If it can help to find more tree matches, why to avoid that?

Because it is only semi-accurate, and that only for English spellings of names. There IS an adaptation of Soundex that is better suited for Slavic and Hebrew names, called Soundex-DM, but Geni does NOT yet support it.

Юрий Петрович Чернявский you are missing some surname variations still:
http://www.eki.ee/varia/transkribaja/

Rus-Est transcription: Tšernjavski
Rus-international transliteration: Černjavskij
Rus-Eng Library of Congress (USA): Cherniavskii
Rus-Eng BGN/PCGN (UK, maps): Chernyavskiy
Rus-Franc (USSR Zagran-Pasport): Tcherniavskii

Lauri Kreen,
Is it official transcription, not machinery?

@Shmuel-Aharon Kam (Kahn).
Hmm, ok but it can help to find tree matches, right? It is only suggestions and not adding relatives automatically.

Lauri Kreen,
What is "international transliteration"? It is not English transliteration, can't understand the meaning.

No it is best NOT to be used for finding matches, because too often people will see a match and complete the merge without really checking it.

The 'international transliteration' might perhaps refer to what is commonly known as 'scientific transliteration'. Info on scientific transliteration and a table of ISO 9 with diacritics (Ukranian, Russian etc.) can be found here: http://en.wikipedia.org/wiki/Scientific_transliteration_of_Cyrillic

This seems to be conflicting point. From one side, it is URGENT to match relatives across different names spelling, languages, etc, but in other hand, people often complete wrong merger.

My prediction is that Geni will allow to use more algorithms but has some mechanism to prevent wrong merger. For example GENi can alert users if matching probability is low.

Olav Linno Poëll, Lauri Kreen,
I see that 'scientific transliteration' is can't be using as native transliteration for any user/profile. Users should use native kind of surnames on native language.

Юрий Петрович Чернявский YES - All these are official. This site is managed by Estonian Language Institute and all these are based on well known and agreed transcription tables.

The "International Russian-Latin transliteration is according to 1987 Montreal agreement.
http://www.eki.ee/books/ekk09/index.php?p=2&p1=6&id=36
it is the same as GOST (1983) / UN (1987)
http://en.wikipedia.org/wiki/GOST_16876-71

The Russian-Estonian one is used by Estonian translators, newspapers etc:
http://www.eki.ee/books/ekk09/index.php?p=2&p1=6&id=37
(it is included in all Russian-Estonian and Estonian-Russian dictionaries).

The Rus-Eng Library of Congress (USA) is according to ALA-LC table from here:
http://en.wikipedia.org/wiki/Romanization_of_Russian

The Rus-Eng BGN/PCGN (UK, maps) is according to BGN/PCGN table also from here:
http://en.wikipedia.org/wiki/Romanization_of_Russian

Rus-French (USSR Zagran-Pasport)
This is according to:
"по французским правилам транскрибируются" in:
http://www.vgd.ru/ENGLISH/translit.htm

Lauri Kreen, Thank you very much.

Unicode, anyone?

The problem is still exist :-(

Problem with spammers as above? ;-)

The surname section are by the way not visible on claimed profiles, - I have reported that as an error because that would be the first place you would click on such a link.

Yes, the problem with surnames.

I not fully understand what you mean due to my English is not native.

I discovered that this is a common problem for all long non-Latin surnames, like my Чернявский

Spammers must die :-)

@Mike Stangel, Surnames still not works for long non-Latin names

Can you post an example that we can work with, to fix the problem?