You might want to reconsider the way that you calculate the percentage of a match because of a couple of problems I've noticed.
Firstly, consider two brothers born in different years, with the first names Jack and John. Given the parents' details, and place of birth, all match, this scores very highly - over 95%, despite them being different people. I don't really have a suggestion on how to avoid that.
However, I'm also getting matches in the high 70%, where the names, dates and places of birth and death, and spouse's and children's names and details are all exactly the same, but my tree has the the parents' names and the other tree does not. I think that is a logic flaw, and also very annoying in that the other tree has nothing to contibute to my tree, even though I do have something to contribute to theirs. I would agree that if the parents' names were different, then maybe 70% is appropriate, but where one tree has no parents' names and the other tree does, that should not lower the match percentage. So the percentage matching should not treat a lack of information as a "no match".
A further problenm is in addresses - "England", "Wiltshire, England", "Mere, Witshire", "Mere, England", and "Mere, Wiltshire, England" are treated as five different addresses, and then tripled because some people put "UK", or "United Kingdom' at the end - so 15 ways to show the same piece of information. My only suggestion for this is to not just deal with an address as a single string, but to parse it into country, state shire or county, region, etc and to calculate the match on the basis oof consistent or inconsistent information - that is, if one has a country only and the other has a city, region and country, that is not actually a mismatch at all.
Last April (2015) I suggested improving the alpha listing of those site managers who have or have had matches with your tree. Every so often I would like to be able to find a site manager again after a having a match and cannot locate them due to the odd alpha listing. Usually site managers are listed by their first names, but not always, so finding them is sometimes nearly impossible. And, of course, the more there are the harder it gets. Please make the listing better. And also make a place where you can type in a site managers name and get sent to their name in the listing.
I would like to see how someone is related to me at all times. So if I go to Smart matches and I compare trees with someone, or if I bring a person up in my trees, or just anywhere I am looking at someone on my trees, I would like to see how they are related to me. On my tree, I don't want to add my great aunt's nephew's niece on to my tree as I am trying to keep it a direct/blood line. But I don't know if I am adding someone on that isn't directly related, as I do my research on this person
I have seen several other requests of this type, and I too would value this. I've got to the stage where I just don't bother checking matches that do not indicate relationships. I suggest that you add a field to each person in someone's tree that stores the number of steps that person is away from the the tree owner - a byte would probably do it. I know that in FTB, there is a column in the index that shows the relationship as the number of steps away that person is. Is this being recalculated continuously? I think probably not. That would mean that you already store that information. It is the work of a couple of minutes to add that number to the information being displayed on the "people" and "trees" tab of the smart match page. A simple (x steps) after the dates of birth and death or name would be enough, and would take one line of code. And while you are at it, why do the smart match pages only show the year and not the day and month of births and deaths? I'm sure I have missed getting more concise information as a result, since I have now checked over 34,800 smart matches and do not bother with "review match" unless I see there is some new information there. Also, how about making the default for number of matches to show at a time 10 instead of 50, so it doesn't jam up everyone's download time (and your computer time too!)?
Being a programmer myself, I know those 3 changes could be made in minutes, with no need for extensive testing. I've been using FTB 7.0 and the website for around 5 years now - when is an update coming out?
In the "Find a person" entry block if I enter a persons name that was born before the year 1000 (four digits) the names come up with "?" for the date. If DOB and DOD are both before the year 1000 then they both come back with "?" in the date fields. This makes it difficult finding the right person if they lived or died before the year 1000.
I suggest that you need to tweak your results table field to recognise dates that are less than 1000.
I have found a lot of matches where the tree matched to mine seems to have taken the information from my tree - and obviously it is a match. I guess this from the fact that the matching tree has a photo that I took and put in my tree - so I am the origiinal source.
Now I accept that the owner of the new tree may have subsequently added information that I did not know, but I have yet to find a case where this has actually happened - in all cases, there is no new information to be had. But I still get notification of a match, have to wait for the matches to download, go through them, press 3 buttons to confirm, confirm the confirm, and get back to the smart match page to look at the next one. This is getting tedious. I do, however, occasionally find matches with some interesting or pertinent data, so I am not willing to give up entirely (but maybe soon !).
So, I propose a couple of options :
1. Automatically confirm in my tree when the other person has already copied the information from my tree through a match that inserted the person from my tree into theirs. I don't think there is a danger of them adding information that I will miss - it will still turn up if I check the people matches and get the combined data. Maybe this could be an option, so those with a lot of info already could let the matches be automatically confirmed, whereas those just starting out could get hem all.
2. Provide an option on the smart match screen (the one that lists all matches) to filter - like any database should have. One would be to filter out already confirmed matches - why make me wait fora list of 50 matches when I have already confirmed or rejected all but a handful? Then add a filter that I can activate (and happen by default) such that smart matches does not bother showing matches with trees that have the same or less data than my tree already has.
3. Alternatively, or as well, provide a button that goes through all matches and automatically confirms those that fit a selectable filter option - e.g. >90% match, no extra information, and less information.
Then maybe I wouldn't waste an hour a week confirming matches without finding anything new.
I, too have asked for the percentage filter, as well as a 'last modified' filter. If a tree hasn't been modified for a couple years, it is obviously stagnant. I'm hesitant to opt the 'ignore this tree' in case they come back and wish to expand. As soon as they log back in to their tree, the 'Last Updated' field will show current, therefore allowing the matches to once again present themselves.
Tofay I had quite a few single smart matches that I worked through.
This involved going to the tree tab of the smart match page, clicking on one of them, waiting for the match page to appear, reviewing the information, and confirming the match, AND THEN - I have to go back to the smart match page and wait for the whole 50 matches shown on the page to reload and appear before selecting the next one.
It would be better if I could click on a "go to next pending match" button so I wouldn't have to go back to the main page every time.
Try loading two copies of MH in separate TABS then you will already have the SmartMatches loaded in their own TAB. That's pretty much what I do.
P.S. But I have an even better advantage. My grandson works in IT and suggested my computer was very tired and maybe sick and needed a new motherboard. He said,'You would really like it to be faster...'
And now when I start it (seven second start-up...) I get a screen display with a vicious big dragon and the legend, 'Gaming motherboard'. It goes like the clappers!
I recently added 10 people using the "Add Discovery" only to find that one was a duplicate. The Discovery page is a great idea, but it would be so much better if there was some way of viewing and selecting which of the 10 people were to be added. For example, in FTB, when using the smart match page "merge" facility, we get a popup window allowing us to select which people, and even (through the "step-by-step") which data, to accept. That is a MUCH better way to do it.
Also, I had four separate discoveries in my list, and when I accepted one, and returned to the discoveries page, the other three had disappeared, leaving a message that discoveries were being sourced, and I should come back later.
In general your idea is great - EVENT can be a general genealogy element which does let to map participation of selected PERSONS. Same like maping PERSONS with PHOTO. This would open number of possibilities and would convert the system into HISTORY MANAGEMENT SYSTEM.
The problem is - MH intend to be GENEALOGY INFOMATION SYSTEM, it would be nice enough if MH would focus to make the GIS working smooth, inlcuding:
- fix of Tag loss incidents
- fix of GEDCOM export fails
- fix of improper URL interpretation
- fix of lost screen "News and Articles"
- fix of privacy issues
- fix of customer support process
- new function of record changes audit
- new function of duplicate persons merge
Keeping in mind, that none of a/m topics gets MH attention, it is very low probability that your great idea will be vitalized in the nearest 10 years.
The problem - if someone deletes a person after a smart match has been identified, it sometimes appears in your smart match list but with no information showing, and no way to delete/confirm/reject the match.
MH solution, as advised in email - wait, and it will eventually correct itself. (Can take some weeks.)
My workaround - go to a "previously confirmed" match and "undo confirm". Go back to "pending", and click on "confirm all". The blank match will disappear off the pending list.
When using the Smart Matches feature, it often suggests adding new relatives for the matched person.
In many cases, a relative is suggested even though he is alreay in my tree. This usually happens because of naming differences for that relative.
It is not unusual that Jewish immigrants have a Hebrew first name, as well as a local first name.
In one of the cases I've come across, Smart Matches keeps suggesting a relative named 'Joseph' even though I already have him as 'José' in my tree. Smart Matches thinks these are two different people even though they have the same dates of birth, spouse, children, etc.
Therefore, I suggest adding an option of manually merging such duplicates.
This is a minor irritant, but maybe has a simple fix. As a programmer myself, i know it only needs a few lines of code.
When copying information from matched people in a smart match (smart matches - "by people" tab, "view matches" button, "copy and edit details" button, the only data that can be easily copied, using the arrow button, is the most common entry. Usually this is a great and simple feature, however ...
Sometimes the most common data is not the most detailed. For example, I might get 130 matches to various trees where a location is recorded as "Wiltshire, England" and 40 where it is "England", and 10 where it is "Mere, Wiltshire, England".
Obviously, I want the "Mere, Wiltshire, England", but clicking the arrow would give me "Wiltshire, England" and then I have to type the extra bit. . I get around this by highlighting, copying, and pasting (click,drag,release,CTRL-C,move,click,CTRL-V) but it would be SO much easier if all the variations of address (and other details) appeared on separate lines, each with an arrow button, and I could choose which one to click and copy. Alternatively, I would like to click on one of the entries to indicate the one I want to copy.
Sorting by the longest entry instead of the most common entry would work much, but not all of the time.
There have been several requests, including from me, to alter the date/time information about posts to Forums, which presently are fixed to the topic's original author's entry. I am again requesting a change, but with a specific programming suggestion.
The date/time information shown is clearly a varying length character string, worded and formatted nicely. It would seem fairly easy to enhance that string by appending date/time information of the most recent reply posting. E.g., "One day ago, replied 8 hours ago". Let us know what you think!