Back to Top

Duplicate Profile Photos Show Integrity Issues

Post Image: 
Duplicate Profile Photos Show Integrity Issues


Sometimes in a digital world it can be confusing whether the lack of quality stems from poor programming (and lack of adequate testing) or from poor data quality. In this post we'll help you identify the possible reasons for quality breakdown that appear in popular social networking sites.

Recently I noticed in my (professional) social media feed that one of my contact's pictures was showing up twice (as seen in screenshot 1 (below). There are a few reasons why this could happen, such as a programming error (e.g. when programming with loops sometimes the results can show the same item twice if it loops one too many times). If a profile photo has been associated with more than one person's profile it may reveal poor data quality. Typically functional testing (or user acceptance testing for that matter) would catch this error. If it isn't a functional problem (e.g. the programming is correct but the underlying data is wrong), it may be ignored by programmers and moved into production with the intent to correct the data later (or that the production data will be correct and the issue will not appear).

My guess is that the issue shown in screenshot #1, below, is a programming error that occurs when a person updates their profile twice which creates two rows of data that is used in this feed. Because the title of change only states a change in position (e.g. first to "Marketing Director" and then to "Director of Marketing", for example), my connection's update to her position (twice) caused there to be two news feed rows of data. The programmers didn't take this into account, and when looping through the news feed data, their program identified those as two different update events (thus showing my contact's photo twice), even though to us humans, this isn't really two separate changes in one's position.
News Feed Events
Event ID Person's Position Title Profile Pic #
1 Marketing Director 33348
2 Director of Marketing 33348


If I am correct, regarding screenshot #1, the resolution of the issue is to revise the program to only show profile pictures once for all event change of a type, and/or set period of time. In other words, a true change in job role, not a simple data change. If however, the issue was caused by two of my contacts having the same picture it could be a data quality problem similar to that described in screenshot #2, described next.

Screenshot #1 Screenshot #2
Screenshot 1 Screenshot #2
Screenshot #1 Screenshot #2


In another example, shown in screenshot #2 (on the right above), which is from a mobile application, the profile picture of one person (shown at the top and bottom of the screen), shows two times in the list, but under two different names. In this case it isn't clear if the issue is programmatic in nature or related to data quality. My guess is that during creation of the profile the uploaded picture was associated with the wrong person, and without listing each profile and picture (as a test) the issue was overlooked until I found it by scrolling through the profiles as an end-user.

If this was the case, we might find the profile data to look like the following. In this case clearly the "lin-picture.jpg" profile picture is incorrectly associated with Lin and Susan. It should only be shown for Lin (Person ID 1).

Person ID Person Name Profile Picture File Name
1 Lin lin-picture.jpg
2 Susan lin-picture.jpg


In terms of the Conformed Dimensions of Data Qualiity, we'd say, in this case that the Accuracy, and specifically, the concept of "Agree with Real-world" has been broken. In a database context, the user's profile pictures are likely stored outside of the database (depending on how it is setup) with foreign keys linking the picture file to the person's record in the database. In this case the "Referential Integrity" linking the picture with the profile may be incorrect, thereby exhibiting poor quality whithin the Integrity dimension.

Another possibility is that, during the testing of the application, the person identified earlier as "Lin" created two accounts (under two different names), but used the same profile picture, because she was in a hurry. Unfortunately one, or both of those test accounts were not later deleted before the application released for the event.


Blog tags: