Potential Problems with Data Mashups

Below is more from Information Management’s May 2010 Webcast by DM Radio, “The Last Mile: Data Visualization in a Mashed-Up”. This Webcast was hosted by Eric Kavanagh and included BI consultants William Laurent and Malcolm Chisholm, and InetSoft's Product Manager Byron Igoe.

Eric Kavanagh (EK): Let’s talk some more about mashups. William, what would you say about the potential problems with mashups?

William Laurent (WL): If we’re talking about weakness about mashups, one weakness is that I have seen in the publishing capabilities of a mashup or a mashup product. How do we go from conveying this wonderful set of information on the person’s computer screen, which is essentially a 3D world view to producing that as a hardcopy. There’s no click-through capabilities. You’re going to lose a certain amount of information. I think that really is, from what I’ve seen, is a challenge for the actual developer to take that information that we have in mashup form and incorporate that into the hardcopy world.

EK: And Malcolm, what are your thoughts about all that?

Malcolm Chisholm (MC): Well, my thoughts are around the semantic challenges. You have some users who sit in very narrow silos and understand a little bit of data, but with a mashup obviously you are bring together disparate data and creating a whole picture, and you have other users who sit at a much higher level and see it more broadly but don’t understand all the nuances that are in there. So how do we address this semantic challenge of understanding the data that we are bringing together sufficiently so there are no issues about interpretation when the mashup is rendered to them.

EK: Yeah, and that’s a very interesting point, Malcolm. Is the relationship causal? Is it related? Is it merely a fluke that the way you are looking at these things, they seem related? That is definitely an issue, but that is not just limited to mashups. It’s limited to ten years ago when folks were doing multidimensional analysis on OLAP cubes. They slice and dice. If they see a pattern, is it? Is it causal? Is it correlated? Is it not? That’s almost a separate issue all together. How to visualize the data and then accurately communicate with users.

MC: I agree that it is a very old issue, but it will just reappear in a different form. It is still something that we have to face somehow. Going back to the earlier discussion, maybe there’s something we can do with that which will make it easier for data mashups.

Previous: How Data Mashups Work