Home » Business Topics » Data Strategist

GAI, application sprawl, and the universal need for data-centric architecture

  • Alan Morrison 

Interview with Dave McComb, President of Semantic Arts

GAI, application sprawl, and the universal need for data-centric architecture

Image by Gerd Altmann from Pixabay

At the beginning of his career, Dave McComb was having fun as an IT consultant at Andersen Consulting — which later became Accenture — building and implementing enterprise application systems. But eventually he realized that IT departments and the consultants helping them were creating a serious problem. 

Early on, enterprises each ran only a handful of applications, and the applications were not well connected. Now enterprises are each running thousands of applications, even less well connected. The complexity has become overwhelming. 

Borrowing a term from Turing Award winner and Mythical Man Month author Fred Brooks, McComb asserts that, in terms of the typical large enterprise, “Accidental complexity has gone berserk — I think it’s 5,000 percent.”

How do you identify what accidental complexity is in information systems, and separate it from essential complexity? McComb thinks back to the example of the work he and his staff at Semantic Arts did at Schneider Electric, an industrial electronics equipment supplier.  There, the key question related to the company’s parts catalog database was which parts were compatible with one another in a specific country, given the different electrical requirements in each.

With the help of Schneider Electric’s project sponsor, knowledge graph consultancy Semantic Arts took the company’s relational database-oriented parts catalog and remodeled it to be transformed as a knowledge graph. (Full disclosure: I do some work for Semantic Arts.)

The result? A complex beast of a database with 700 tables and 7,000 attributes was reduced to 46 essential classes and 36 properties. Additionally, Semantic Arts dealt with the parts compatibility matching challenge, which had been stranded in separate spreadsheets maintained by engineers, by writing rules in SPARQL – SPARQL Protocol and RDF Query Language, a means querying standards-based semantic graphs  – that were then stored as a part of the knowledge graph.

The interview we recorded has a number of gems like this story, including thoughts on architectural transformation for GAI. I hope you enjoy listening to it.

Interview podcast with Dave McComb