You want to integrate two disparate pieces of software, so you set up message interactions between them. You hammer out a common data format for the messages. You poke and prod and get all the technology to work.
But your software still can’t communicate. The problem? Semantics.
Semantics refers to the meaning of data. You can have the right information, but if you don’t know what it means, you can’t do anything with it. So, in order to solve our integration problem, we buy a semantic integration tool and hook that up to our system. Problem solved, right?
Not so fast. Even the best semantic integration tool on the market probably doesn’t understand what the information in your messages mean either. After all, computers don’t understand meaning. In the final analysis, computers understand only zeroes and ones. Only humans understand meaning.
Yes, we have many semantic tools, standards, and models at our disposal. The Semantic Web. The Uniform Business Language (UBL). The Universal Data Element Framework (UDEF). Not to mention literally hundreds or even thousands of industry-specific semantic models. But just when we think we’ve cracked the semantic nut, a complete solution is just out of reach. We’re chasing the pot of gold under the rainbow, and just as we think we’re getting close, we realize it’s still out of reach.
Perhaps our tools, standards, and models simply have to mature a bit. We’ve made so much progress that fully automated semantic integration is just around the corner, some say. But I disagree. I don’t think we’ll ever solve this problem with digital computing. Semantics is something only human brains can deal with, and until we can replicate our own brains, our technology will fail in this task.
The problem? Ambiguity. The goal of semantic integration is to eliminate ambiguity, so if two pieces of software are both talking about a business entity, say for example, customer or invoice, then they both unambiguously define the meaning of that entity. But when humans converse, ambiguity is an inherent part of our conversations. As long as we seek to eliminate ambiguity rather than embrace it, the semantic pot of gold will remain out of reach. But ones and zeroes don’t allow for ambiguity. Only humans can do that.