Wednesday, September 19, 2012

Introducing Record Matching - MyHeritage.com - English blog

19????Sep 20120 comments

We're pleased to introduce today a new technology - Record Matching - that automatically finds relevant historical records for every family tree on MyHeritage!

This is an add-on feature for SuperSearch, our global search engine for historical records, that was successfully launched in June. We're very excited about Record Matching, and believe it is a breakthrough that can bring value to almost every user of MyHeritage and to people not using MyHeritage who are curious about their family history. Read the details below and we hope you'll share our excitement.

Introducing Record Matching Technology

What is Record Matching?
If you're like many of us who love genealogy but don't have lots of spare time to invest in it, you'll love Record Matching. While you're busy with other things - or even?sleeping - Record Matching does much of the work for you. It works behind the scenes on a new server farm set up by MyHeritage, constantly comparing every family tree on MyHeritage to more than 4 billion?historical records on SuperSearch, looking for matches to bring to you. A Record Match is a document relevant to your family's history, such as a birth record of one of your ancestors, a tombstone photo of a relative in your family, or a newspaper article describing how your great-grandfather met and fell in love with your great-grandmother. Record Matches are found automatically and delivered directly to you. New discoveries await you!

What's unique about Record Matching?
Record Matching is the world's first and only technology to?find?family tree matches in newspaper articles, books and other free text documents, using semantic analysis. Based on the world?s largest historical newspaper collection dating back to the 18th century that we have under license, relevant newspaper articles found using this technology are incredibly valuable in shedding light on the lives, personalities and achievements of our ancestors. We call it "adding color to family history".
Tombstone
This is done in addition to matching structured data such as birth, marriage, death and census records. Of particular note is the collection of tombstone photos that we have under license from Find-A-Grave. It's extremely useful to receive matching gravestone photos and burial information for the individuals in one?s family tree, automatically and effortlessly, and in one consolidated location, as provided by Record Matching.

Synonyms and phonetics
Record Matching is the first technology to translate names between languages, to find documents for you even in languages different than the one your family tree is in. The technology is particularly good with synonyms and phonetics, so you can expect matches ranging from the obvious (William in record vs. Bill in the family tree) to the subtle (Alessandro in record vs. Sasha in the family tree).

Each match has a Confidence Score, ranging from half a star to five stars, indicating the likelihood that the historical record found correctly belongs to the associated individual in the family tree.

Record Matching runs periodically?in order to cover new individuals recently added or edited in the family trees, and to cover new data collections of historical records that we keep adding.

As you can see, Record Matching is another great family history research?tool for MyHeritage users.?If your family tree is on MyHeritage, we'll continuously compare it ? for free ? to the billions of historical records we have, so that you?ll have a better chance to make new discoveries with no effort. The technology's high level of accuracy will save time, and you won't need to do a thing except sit back, relax, go through the Record Matches we deliver to you and enjoy the discoveries they bring.

How to use Record Matches
Record Matches are calculated automatically for all family trees on MyHeritage.

When you?log in?and visit your family site on MyHeritage, you will see a box on the side bar listing the number of Record Matches found for you. Click it to enter the Record Matches page. You can also access this page from the "Family tree" tab in your family site.

Record Matches on family site (click to zoom)

Record Matches on family site (click to zoom)

People and Collections
The page for viewing your Record Matches allows you to go over them by person or by collection. So you can view all matches that were found for a particular person in your tree, or view all matches found for everyone in your tree in a particular collection, such as the 1940 USA census.

Collections

Record Matches viewed by collections (click to zoom)

Record Matches viewed by collections (click to zoom)

Click the blue buttons to view the matches in any of the collections. For example, click the "Find A Grave" collection in order to see tombstone photos of deceased people in your family tree. We call this the "collection page" and it displays all Record Matches from one collection. The information from your family tree is displayed on the left and information from the matching historical record, on the right.

Reviewing Find-a-Grave Record Matches (Click to zoom).

Reviewing Find-a-Grave Record Matches (Click to zoom).

Click any "Review match" button to drill-down and see the full record for any match. This is displayed in the Record page.

Record page
In the example Record Page below, the Record Match was an interesting newspaper article about a railroad engineer in the family tree. The date of death of this person was known in the family tree, but the newspaper reveals the unfortunate circumstances of his death - a coal shed explosion! This particular Record Match was found for Randy Seaver's family tree (more on this below).

Record Match - Tragic death in coal shed explosion (click to zoom)

Record Match - Tragic death in coal shed explosion (click to zoom)

Note that you can save any newspaper article image and add it to your family tree by clicking the Download icon in the bottom right corner (shaped like an arrow pointing down).

People
If you prefer, you can also go over your Record Matches by people, instead of by collection. To do so, click the "By people" tab in the top right corner.

Record Matches viewed by people (click to zoom)

Record Matches viewed by people (click to zoom)

Click the blue buttons to view all matches of any of the people. We call the page that will open the "person page" and it displays all Record Matches of a single individual in your family tree. Information about this individual from your family tree will be conveniently displayed on the left side, allowing you to compare it quickly to the information found in the historical records.

Person page

Person page: Record Matches of a single person (click to zoom)

Person page: Record Matches of a single person (click to zoom)

In the example above, the person has a match in the Social Security Death Index and another one in the 1940 USA census. Click any "Review match" button to drill-down and see the full record for any match. This is displayed in the Record Page.

Record Match in 1940 US Census (click to zoom)

Record Match in 1940 US Census (click to zoom)

Shown above is a Record Match for?a person in the 1940 US census collection.?Our users can click the census image to view it full-screen in our special viewer.

Confirming or rejecting Record Matches
You are encouraged to confirm or reject the matches as you go over them, indicating that they are correct or incorrect, respectively. This can be done in the Record pages, Collection pages and Person pages. It is helpful as a way to mark matches you have already reviewed in order to easily work your way through the matches. Confirmed matches will have a green checkmark next to them, and you can later change your mind and undo the confirmation if necessary. Incorrect matches will move out of the way and become hidden from view when you refresh the page. Your feedback on the accuracy of the matches will also help us improve the technology. As shown below, you can set the filtering options to temporarily hide confirmed matches from view, leaving only the pending (unconfirmed) matches displayed.

Before confirming a correct match, or shortly after, you are encouraged to extract the information from the match (scanned image, dates, events, facts, stories, etc) and add it to your family tree. At present this needs to be done manually in another window, as the first release of Record Matching does not include a save wizard for adding information directly from a match into the family tree on MyHeritage. This, along with an option to create a source citation in the family tree with the matching record, will be added soon in the first update of Record Matching. We recommend that you create a source citation for any piece of information that you add to your family tree from a historical record, so you and other researchers will know later on exactly where you got it from.

At the bottom of each list of matches you have the option to Confirm All or Reject all if you wish to apply these actions on all the matches displayed on the page instead of doing it individually.

Filtering Record Matches
In the Collection pages and Person pages, you can click the "Filtering options" link. The following screen will open, allowing you to control which matches are displayed.

Filtering Record Matches (click to zoom)

Filtering Record Matches (click to zoom)

You can use filtering to hide confirmed matches (thus showing you only pending, unconfirmed matches) and you can always bring all matches back by changing the filtering options. You can view only matches to structured records (such as vital records, census records), text records (newspapers and publications), or both. You can also request to display only matches with a minimum Confidence Score of your choice.

Confidence Scores
Record Matching is highly accurate, but its accuracy doesn?t come at the price of finding fewer matches (i.e. only the obvious ones). Record Matches assesses the likelihood of each match being relevant to a family tree individual, and in the process assigns a Confidence Score for each match based on the degree of similarity of the information in the record (names, dates, facts, relationships, etc) to the information in the family tree. We have prepared an additional sub-system which we codenamed "Devil's Advocate" which goes over the matches and reduces the score or completely rejects those that contradict information in your family tree, to reduce false positives to the minimum.

This Confidence Score, which is a unique feature of MyHeritage, is displayed next to each match and ranges from half a star to five stars. You can sort matches by score on every page, or filter matches by the score and look only at higher probability matches, or extend the scope to look at all matches, likely and unlikely.

Please note an interesting conundrum: the higher the score, the less interesting the match might be! This is because a high-scoring match means that the historical record contains and matches a lot of information in the tree, hence your tree probably contains a great deal of information about this individual, so the new match may possibly not add much new information. There may be 'genealogy gold' particularly in the lower scoring matches where the system is less confident about the relevancy of the match, but that doesn?t mean the match is less interesting or brings less information, rather there may not be much information in the family tree about this person to lift up the score.

Delivery of Record Matches
You can visit your family site at any time to see the Record Matches found for you, including the latest ones. In addition, we'll be sending our users a weekly email (depending on their preference) like the one shown below. The email will list new Record Matches that were found, and provide the option to review any match directly by clicking on the links provided. This is similar to the weekly email that delivers Smart Matches?. If there are no new matches to speak of, no email will be sent.

Record Matches email notification (click to zoom)

Record Matches email notification (click to zoom)

Record Matching and Smart Matching?
Record Matching augments our flagship Smart Matching? technology. Smart Matching? finds matches in family trees of other users, whereas Record Matching finds matches in all other (non-family-tree) historical records. The two technologies work together in a cycle that constantly pushes forward your knowledge of your family history. As you collaborate with other users and enhance your family tree using Smart Matches?, Record Matching receives more leads and information with which to find more historical records, which in turn facilitates more Smart Matches? with other family trees. For example, when you confirm a Smart Match? for an individual in your family tree, with an individual in another user?s family tree, you are rewarded as all Record Matches found for the other individual automatically get applied to your family tree as well.

Enhancements on the way
There are many exciting sub-features and enhancements that couldn't be included in the first release and will be added by us in the next few months. These include the ability to save matches directly into the tree, highlighting of matches that add new info to the family tree, real-time matches displayed immediately whenever a new person is added to the tree, and integrating the Record Matching technology into our profile pages and Family Tree Builder genealogy software.

Please note that matches that are currently shown in profile pages and in Family Tree Builder 6.0 are not Record Matches, but are simple search results in World Vital Records. These will be replaced with real Record Matches very soon.

How much does it cost?
The new Record Matching feature runs for free for everyone who has one or more family trees on MyHeritage.

Viewing the matches is free in extract (some information will be hidden), but viewing the full records and their scanned images or the newspaper articles requires a Data Subscription which is the same subscription used to view records on MyHeritage SuperSearch.

For users who prefer it, pay-as-you-go credits may be purchased to view specific records in smaller quantity, in lieu of a subscription. Credits can be used to review a few matches of high interest, but if there are a great deal of interesting matches, a Data Subscription is a more economical way to review them all -?see details on data subscription and credit options.

We're currently offering Data Subscriptions with a 36% discount, costing US$76 for the first year instead of the regular price of US$120. A Data Subscription provides unlimited access to all records in MyHeritage SuperSearch and to all Record Matches.

Some Record Matches, such as those coming from partners like Find-a-Grave and Ellis Island, are always free and viewing them does not require payment of any kind. For viewing the full records, users are referred to the Find-a-Grave and Ellis Island websites, which are free.

Record Matches in Action
To?demonstrate?how effective and interesting Record Matches can be, here are?some examples from the family tree of well-known genealogy expert and author of the Genea-Musings blog, Randy Seaver. We'd like to thank Randy for allowing us to use examples from his family tree in order to demonstrate the technology. Almost 4,000 Record Matches were found by MyHeritage for Randy's tree including hundreds of newspaper articles that will add a lot of color to the family history. As Randy nicely put it, he now has "a lot of work to do!" :-)

Randy's Family Tree - Profile of Frederick Thomas Blanchard

Randy

Randy knew that Frederick Thomas Blanchard, his first cousin twice removed, had?married Mary Helen Webster?in 1912.

What Randy had never seen before was a 100-year-old newspaper article (see below) which showed that Dr. Frederick Blanchard was marrying?his?college sweetheart; they met when they were both?students. The?article?lists his achievements, and?indicates that the couple plan to make their home in a?"cozily arranged bungalow"...

Like all Record Matches, this match was found automatically.

Record Match: university romance (Click to zoom)

Record Match: university romance (Click to zoom)

Here's another example from Randy's tree. Laura (below) has no children and not much is known about her husband.

Randy's Family Tree - Profile of Laura May Underhill (click to zoom)

Randy

With automatic Record Matches, Randy discovered that she was the wife of a mayor who died in a plane crash! The article also mentions that they had a son who was a judge, a fact missing in the family tree and now the door opens to find out more about this previously unknown person in the family.

Record Match Discovery: wife of a mayor who died in a plane crash (click to zoom)

Record Match Discovery: wife of a mayor who died in a plane crash (click to zoom)

Finally, here is an example of an unexpected discovery found for the family tree of MyHeritage founder & CEO, Gilad Japhet. In the tree we see his relative, Gertrude Sarah Levin, who married Solomon Deitch but little else is known about them.

Gilad's family tree - Gertrude and Solomon (click to zoom)

Gilad

The Record Match was a newspaper from Kansas in the 1930's. It includes their wedding photo. The bride is beautiful. The caption is intriguing: "Love needed no words". It turns out that the marriage service was conducted in sign-language as both were deaf mute... Another great example of how Record Matches adds color to what we know about the lives of our ancestors and relatives.

Record Match: Not your usual wedding (click to zoom)

Record Match: Not your usual wedding (click to zoom)

These are just a?few?of the thousands of Record Matches found for Randy?s and Gilad's family trees, and in fact of the hundreds of millions of Record Matches that we have already found for our users, which are available as of today on MyHeritage.

Getting your Record Matches
Wondering what Record Matches we can find for you? Check your family site now to see what discoveries are already waiting for you there.?If you're using our Family Tree Builder genealogy software, be sure to publish your latest family tree to your family site on MyHeritage, in order to have Record Matches calculated for you.

If no Record Matches were found for your tree yet, worry not, all you'll need is some patience. Record Matching works continuously?and new data collections are added daily. We will bring the results?to you automatically?once?found, and we are specifically working on adding data collections for non-English speaking countries.

If you're not yet a member of MyHeritage, you're more than welcome to sign up for free at MyHeritage, import your family tree or build a new one using the site?s friendly tools, and receive your matches in less than 24 hours.

We're delighted to invite you to a webinar about Record Matches in which we'll go step-by-step through the process of accessing and reviewing your matches and explain how to use the information to add family members or details to your family tree. Please visit the webinar registration page.

We look forward to hearing your experiences with Record Matches and welcome your comments below.

Source: http://blog.myheritage.com/2012/09/introducing-record-matching/

jr martinez melasma jimmy rollins jimmy rollins let it snow jason trawick jerry lewis

No comments:

Post a Comment