How to Implement Large Complex Cloud Solutions

Ed Featherston

Subscribe to Ed Featherston: eMailAlertsEmail Alerts
Get Ed Featherston: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn


Related Topics: Cloud Data Analytics, Big Data on Ulitzer

Article

Finding the Right Little Data | @CloudExpo #BigData #ML #InternetOfThings

Even with the great strides technology has taken, data quality remains a tremendous challenge for genealogy researchers

Over the years, one of my favorite pastimes has been working on the family genealogy. I first started work on it in the early 1990s. At that time, records were not digitized and research involved going to libraries, newspapers, and various local, state, and federal archives. There one would have to sift through reams of old records, documents and microfiche. If you were lucky, someone had created printed indices of the information contained in those documents. These indices, when they existed, were based on manual transcription of the data, and prone to the data quality. This is inherent in transcribing information from what frequently were old handwritten documents. The challenge was then trying to find those nuggets of information that would relate and connect to individuals you were trying to locate in your research.

For anyone operating in the Big Data space of today, this may all sound very familiar. Genealogists, amateur and otherwise, have been dealing with the challenges of Big Data, unstructured data, and data quality long before the terms became technology buzzwords. There are tools and products today to help; who hasn't seen the commercials for ancestry.com? However, even with technology, there are still challenges. These challenges are not unique to genealogy, and make a good lens for viewing and discussing the business needs in general for Big Data. Let's take a closer look at some of them.

Data Quality
Even with the great strides technology has taken, data quality remains a tremendous challenge for genealogy researchers. Let's take U.S. Census data as an example. Every 10 years, the U.S. government conducts a census of the population, and the results of these censuses become public 72 years after they are taken (the 1940 Census material just recently became available). U.S. Census data is a gold mine for genealogy research.

Below is a sample from the 1930 Federal census. Census forms were filled out by individuals going door to door and asking the residents questions. However, there were a number of data quality factors you must take into consideration. The sample here has fairly good quality handwriting, although that's not always the case. Also, you are constrained by the census takers interpretation of the person's answers and pronunciation. For example, this could result in different variations on the spelling of names.

When this document gets transcribed, you could still have multiple sources of problems with the data quality. The original census taker could have written it down incorrectly or the person transcribing it could have made a transcription error.

This challenge is not unique to genealogy research. Data quality has been an issue in IT systems since the first IT system. In the world of Big Data, unstructured data (such as social media), and things like crowd-sourced data, can become a daunting challenge. As with any challenge, we must understand the impact of those issues, the risk, and what can be done to mitigate that risk. In my above example, Ancestry.com takes an interesting approach to the mitigation. Given they have millions of records based on scanned documents, checking each one is beyond reasonable expectations. Given that, they crowd-source corrections. As a customer, I locate a particular record for someone I am looking for, that little data in all the Big Data. If I notice there is some type of error I can flag that record, categorize the error and provide what I believe is the correct information. Ancestry will then look at my correction and, if appropriate, cleanse the transcribed data.

Data Pedigree
Even though we are discussing genealogy, data pedigree is not about the family tree. Data pedigree is ‘Where did the data comes from?' and ‘How reliable is that source?' If, as an organization, you own the data, that is not an issue. In today's Big Data world, many sources are outside of your direct control (unstructured social media data, crowd-sourced data). For genealogy research, data pedigree has always been an issue and concern. A date of birth is a lot more reliable from a town birth record than from say the census example above, where the information is ‘the age on last birthday, as provided in the interview' (I have seen variations of multiple years from sequential census forms for an individual). In my Ancestry.com example again, as well as source records, Ancestry members can make their research available for online search and sharing. When using others' data (i.e., crowd sourcing research), one must always feel comfortable with the reliability of the source. Ancestry allows you to identify what your source of information was, and can identify multiple sources (for example, I may source data of birth based on both a birth record, a marriage record, and a death record). That information is more reliable than a date of birth with no source cited. When I find a potential match (again, that little data I am truly looking for), I can determine if it truly is a match or possibly a false correlation.

Similarly, in any Big Data implementation, we must understand the pedigree of our data sources. This impacts any analytics we perform and the resulting correlations. If you don't, you run the risk of potentially false correlations and assumptions. For some entertaining examples of false correlations check out www.tylervigen.com.

Finding That Gem of Little Data in the Huge Oceans of Big Data
The ultimate value of Big Data is not the huge ocean of data. It's being able to find the gems of little data that provide the information you seek. In genealogy, it is wonderful that I have millions of public records, documents, and other genealogy research available to sift through, but that's not the value. The value is when I find that record for that one individual in the family tree I have been trying to find. Doing the analysis, and the matching of the data is very dependent on the challenges we have been discussing, data quality and data pedigree. The same is true for any Big Data implementation. Big Data without good understanding of the data is just a big pile of data taking up space.

No technology negates the need for good planning and design. Big Data not just about storing structured and unstructured data. It's not just providing the latest and greatest analytic tools. As technologists we must work with the business plan and design how to leverage and balance the data and its analysis. Work with the business to ensure there is the correct understanding of the data that is available, its quality, its pedigree and the impact of those. Then the true value of the Big Data will shine through as all the gems of little data are found.

This post is sponsored by SAS and Big Data Forum.

More Stories By Ed Featherston

Ed Featherston is VP, Principal Architect at Cloud Technology Partners. He brings 35 years of technology experience in designing, building, and implementing large complex solutions. He has significant expertise in systems integration, Internet/intranet, and cloud technologies. He has delivered projects in various industries, including financial services, pharmacy, government and retail.

Comments (0)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.