The subversion of traditional disciplines by new technology: Pedigree network using large data tracing asked Jo

Family tree enthusiasts may think it's interesting to find information about relatives through census records, birth certificates and other documents on the Ancestry Web site. It is less persuasive when it comes to showing off personal records to friends and relatives, and it is not easy to tell the social files of an ancestor.

The people behind the Ancestry.com service are aware of this. They are now making maximum use of their 4PB databases (including official personal records, user submissions, and other new feature data) to provide users with a computer-generated but editable ancestor Information Digest.

The story view, launched by ancestry, was targeted at a small number of customers earlier this quarter, and now 10% of customers can enjoy the service. Eric Shoup, the company's product executive vice president, said in a recent interview that they plan to further refine story view by comparing the Stroy view feature before and after the story view feature is formally released. Ancestry has enhanced the interactivity of this feature by allowing users to surround a single page of document images and edit the text portions of a file.

How it works

The Story view feature is based on a more sophisticated tool that enables data mining of kinship data including handwritten records. But sometimes there are only key fields, such as name and place of residence. Customers can access handwritten records, locate a relative's description, and view data that has not been processed, such as the person's occupation.

By progressively directing "keyers" to parse handwritten records and convert records to searchable text, ancestry is trying to get more information from handwritten records. Street addresses have been added in this way, and other fields will be added later. At the same time, as ancestry continues to expand its database, social archives will have more sources.

To extract information from multiple documents to generate a paragraph summary, ancestry turned to narrative science, which was founded in 2010 to focus on the use of machines to generate readable copies (the legendary technology that would make us all unemployed in small parts). The early application of sports events and the earnings reports of listed companies, now narrative science technology is being used more for personal information processing.

Reed McGrew, chief developer of the ancestry narrative (narrative) and content (context) service team, said that when ancestry first used narrative science technology, it could only produce data in batches. They generate a lot of financial reports, and that's not what we're trying to offer because this batch is really slow.

Within a few months, narrative science developed a new API that could work at a more granular level. "They generate social archives based on individual users," McGrew said.

Ancestry specializes in the processing of family tree information, the company's editors provide editorial standards, or "rules" that specify the format of narratives data. McGrew explained the ancestry standard: "For example, the record of a child 10 years younger than his mother is more like a typing error, although it will happen in reality, but not in most cases, so we will treat this record as if it were wrong."

A record that contains information about a relative Shoup

In Story view, an ancestor's picture and life summary below is a scaled document picture, not a discrete field of structured text. Next to the picture, ancestry provides the lead that is generated from the document information. Once ancestry finds that all records are related to a person, it will be assembled into a complete sentence according to the ancestry's editing rules. Once the document based lead is displayed in the browser, users can edit and save them before they are shared.

Difficult to share

Ancestry's CIO Scott Sorenson says the challenge is not to create and store new data and Web pages for users. Storage is getting cheaper, and accurate handwriting processing is not a problem. Usually keyers are found in China, the Chinese character set is much larger than our alphabet, they are very good at typing these records.

The real hard part is ensuring high availability of services, millions of users providing the right documents and text, and ensuring that site traffic peaks without crashing, but one of the goals of story view is to allow more people to browse the content of the site and eventually register.