I am new to "My Heritage" and I am still working my way around the program. I have not yet signed up with a subscription. I notice that some of the Smart match researchers have a LOT of people in their trees. (As many as 20,000 ) Are some of these reseachers proffessionals, just collecting names? I am interested in matching up with family members, how can I tell the difference?
I am quite new too but have come up to about 1800 in just 3 months. I have striktly only worked back in time and just added parents and lately all sibblings for each generation. Doing so I hope for more Smart Matches from people who has relativs among those siblings and maybe they have come furthur in the search which could help me. Other people, when they have come as far as possible, turns back and search forward for offspring of all siblings and looks for living relatives.
I can say that I found some historic persons in my heritage so I began to add all related persons even if they where not direct relatives just for fun. I come from Sweden where it is quite easy to come down to late 1700 for most branshes in the family tree as long as the parents are married. If you are lucky and your relatives owned their farms you can come down to early 1700 and possibly even 1600 and even more so if your relativ belongs to the nobles as one of my relatives did. And then almost all are in some way relative to early swedish kings and to in the european nobles. Of course some people may have bragged about there heritage way back in time to look better then they where so you have to take the information with caution.
So if you just add your parents and their parents and... you wont get so many people before a branch is ended. If you are lucky and fill all branches the number of people will double for each generation. After 10 generations the tree will exceed 1000 people with just direct relatives. If you add all siblings you will have thousends of people in your tree.
Very few people on here are professionals. As far as why a tree is so big, it depends on the tree. Some really will have a bunch of people on the tree. Others have copied many times and have the same people several times on the same tree. Other trees are complete gibberish and count mythical gods and fictional people as ancestors.
Smart Match is not so smart, it should be changed to Plain Stupid Match.
Why?
One example:
Serine Maria Hansdatter Rasch b. 1864 d. 1901
has the smart match:
Marta Marie Hansdatter b. 1822 d. 1851
Why should I need to deny this is not correct person.
The smart match says that a person dead before my person is even born is a good suggestion.
Wrong parents, wrong husband, wrong children.. u name it.
I have 25.589 "smart" matches awaiting to confirm. I do NOT want those stupid suggestions.. And even have to pay for them.
I agree "Smart Matches " are not so smart. Example: Mary O'Connor nee Lynch matched with Mary Lynch nee Conroy. Obviously to anyone with a functioning brain that Mary Lynch and Mary Conroy have nothing in common except that their forename is Mary. Oh wow , there's another Mary in the world !! One born in Ireland in 1926 and another born 10,000 miles away in Australia in .........1880 , I believe. How is this a match? And nothing else in common.
I get hundreds of so called "Matches" which take forever to delete/reject.
It also seems to me that I have requested the system to ignore matches from specific trees, but I keep getting them next month. Does anyone else find this happens to them?
The reason that you see wrong natches as well as correct ones is that we show you all matches, to reach mazimum research potential, like Shani explained.
I will forward your request to improve the quality of matches to our developers.
Can you please give me an example of a tree you chose to ignore, and got again?
In our Enhanced Smart Matches, we now show you ALL of your matches, the high quality matches and the lower quality ones.
The list is sorted by quality, so you’ll see the best Matches first, but we also show you the other matches below, so that no information will be hidden, and all your research potential will be met.
If you find that a match is wrong (sometimes siblings are matched with each other for example), just mark it as rejected. The Correct match will most often appear at the top of the list.
Marking the match as rejected will help you manage your matches, and will also help us understand that this should not be regarded as a Match in the future.
You can also choose to ignore all your Smart Matches for a family tree that you are matched with, so that you won't receive any results from this tree in the future.
Please see the following FAQ for more information:
I have now 43.000 matches.
Do you think it is a good way to sit and click "reject" on all those?
The sollution you mention is like having problem with a wheel on your car, and the sollution is "condemn your car".
It would be better if the Smart Match used exact match to even be listed as a match.
Today I even got a match on persons with different sex.
The sollutions mentioned later in this thread is only a cure for a sickness.
It is better to remove the cause of sickness.
A match should be in 3 levels based on importance:
1. Name and date of birth of selected person.
2. Names of parents / Names of children.
3. Additional info like place of birth, place of death and so on
A name is a match if 2 letters per 3 letters of name is correct.
Then you are safe in case of name Olav or Olaf.
I would like at least the opportunity to filter the smart matches, for example when I (soon) enable smartmatch for the bigger tree I have of 88.000 names, I would love to be able to (initially at least) filter that I do not want to see smart matches below 70% for example...
From i wrote that line i went to my smart matches, used 1 minute.. found this:
86%: Johan Mortenssøn Blix b1666 compared to Lars Mortensen Mortenssøn Blix b1671
Everyone knows that this is not the same person. >Only 2 of 4 infos is correct. That is 50%
Since they are siblings they have same parents. Thats why the 86%.
It is wrong to use parents as equal value as the name of main person.
Main person name and birth/death value should be 70%, children 20%, parents 10% of the percentage calculated..
These two people have lots of similarities, and that is why they were matched- their first names are similar and both contain Mortenssøn, they have similar birth years and similar relatives. This is the reason they were matched despite being siblings.
You are welcome to reject this match to help us understand that this should not be regarded as a Match in the future.
I could make you a complete algorithm that removed 95% of the wrong matches.
You need to use graduated points, calculated out of each of percentage simularity of each value stated of each individ.
Name - surname and dates is main criteria, secondary is parents - children - siblings, tertiary is place of birth and so on.
And why I'm sure? I'm as Sheldon Cooper from Big Bang Theory, but sadly without his eidetic memory.
I aced the Mensa-test in 13 minutes and I complained to the board why they couldn't do better...
If I had something like your company in my back, I would turn the whole system around. Making every person a customer, even those who don't want to do research. A product noone would say no to. I would double your return..
All systems available today is like having a watch (your person in your three). Then you know the time. If you have 2 watches (a match from another three, but not same values), you cant be sure which of them is correct.. There is way too much wrong information..
Well.. I have now over 60.000 matches. If I click once every second, I will need to click 120.000 times in next 33 hours!
I won't do that.. I wait until you have a product you could call smart match. For real..
And also would suggest a faster way of rejecting/accepting maches.
When I have one person with over 100 matches it takes a long time to reject all the wrong ones.
Often it is siblings beeing matched, then I would like to get a list of the different first names with checkboxes
and tick off all siblings, then all those matches would be rejected.
Same with the other data, as if two persons have the same name, but different birth year, I would want to reject all
with the wrong birth year
Why? It's like taking headache-pills for noises around you that gives you headache, when removing the cause of noise that gives you the headache should be done instead.
If I see the same "match" appear again after Ignoring and rejecting, then I go the my person, add spaces to the name and the computer can't figure it out.
Jacob Smith becomes J a c o b S m i t h. I can still read it with the spaces, computer can't. Viola! No more StupidMatch.
Maybe they could add a Levenstein distance to the name matching.
The user can then select a number of edits to be made to one name in order to match it to another, and below that threshold will consider it a variation of the same name (including a misspelling). This technique or another like it is likely already in use in Smart Match, so it would be nice to have that setting exposed on the website.
Note that in the FTB software you can tweak the sensitivity of the name matching. This would be similar.
