Semantic Modeling for Data

Chapter 8. Bad Model Specification and Knowledge Acquisition

“Mulla, you lost your ring in the room, why are you looking for it in the yard?”

Mulla stroked his beard and said: “The room is too dark and I can’t see very well. I came out to the courtyard to look for my ring because there is much more light out here.”

Classic Tales of Mulla Nasreddin

As we saw in Chapter 5, before we start building a semantic model we need to decide what exactly we want to develop by specifying the model’s requirements. Moreover, while we are building the model, we need to design, implement, and apply appropriate knowledge acquisition mechanisms that will provide us with all the entities, relations, and other model elements that will satisfy these requirements.

Unfortunately, very often, we perform both these activities in a suboptimal way that results in expensive models that provide little value to their users. This chapter illustrates several problematic practices with respect to these activities, and provides useful insights on how to improve them. Many of these practices and insights (e.g., data specification and selection) are applicable to any kind of data science project, not merely semantic model development.

Building the Wrong Thing

When I joined Textkernel in early 2016, I was all too eager to start building the knowledge graph I had been hired for. Within my first month at the company I had already gathered the main requirements for the graph and specified the elements it should contain. ...