I got to read an issue of Scientific American as I was waiting in the dentist's office the other day. One of the feature articles is an update on the status of "Web 3.0" also referred to as the "Semantic Web."
The Semantic Web was first envisioned by the founders of the internet in 2001. By then they realized that there was a necessary piece missing for the internet to reach its full potential. I call Web 1.0 the "Burma Shave Era" - companies and organizations put web pages up along the "electronic superhighway" much like roadway billboards. Web 2.0 heralds in the "Interactive Web" - with the ability to not only access information but also provide feedback and update information stored on the internet. Web 3.0 envisions the day when the data on the internet is organized in such a way that software applications will be able to explore the internet much as humans do today. These applications will use standardized object relationships to make new inferences and recommend novel solutions to problems.
Web 3.0 depends on the development of a taxonomy that defines the possible relationships between objects and an ontology that carries the specific relationships that exist. Retailers have forever struggled with developing product classifications (taxonomies) for all the items they sell. We have merchandising hierarchies, operating department hierarchies, sales tax hierarchies, etc. The industry effort conducted by the Global Commerce Initiative to develop a Global Product Classification is an indication of how difficult it is to get agreement on how items should be organized.
Web 3.0 envisions defining all the objects found on the internet in an ontology of "triples" where each entry consists of three data elements; the subject, verb and predicate. These relationships will be used to make inferences from relationships found on different websites that represent different disciplines. So computers can look at a triple with "Joe" as the subject, "is a" as the verb, and "person" as the predicate. Somewhere else on the internet is a site that tells the computer that "persons", "are", "human." The computer can now infer that Joe is human. Across the internet, a retail website may contain all the fast moving consumer goods and their ingredients. The National Institute of Health computer may identify certain diseases that are exacerbated or suppressed by certain compounds. An application could be written that would explore these two databases and identify the items which should or should not be eaten by sufferers of these diseases.
The internet is a big place and we're right to be skeptical of the ultimate vision of Web 3.0. The consensus that is needed across industries will be very difficult to achieve, just as we have seen within the retail channel for product classifications. But the potential benefits are huge. If my replenishment or merchandising applications understand what items are substitutes for one another and what items share the same price point or meet similar consumer needs, they can begin to make really meaningful recommendations for assortment planning, pricing, and inventory allocation. With a clear understanding of the relationship between various items, computer applications designed for the retail supply chain could become the true assistants everyone wishes them to be.
Discussion
Questions: How much potential benefit do you see in the concept of a Semantic
Web? Is it hype or reality? Could a smaller scope be implemented for the retail
channel?
[Author's commentary] I have great hopes for the semantic web. I think there
is actually more potential here than for RFID in the short term. By understanding
item relationships, automation can help decide what items to put on sale and
avoid conflicting promotions. Inventory levels can be monitored and substitutions
made on the fly. Competition analysis becomes much easier.
When I read about it, I wondered why I hadn't heard more about the semantic web from the technology community. The cynic in me says there's not enough hardware to sell. The semantic web needs agreement on the taxonomy, a few more disk drives to hold the data relationships, and some application software that uses the data to make decisions. This pales when compared to the hardware revenue from RFID tags, readers and software, but I really believe applications that are sensitive to item relationships and understand the need for additional hotdog rolls when the hotdogs go on sale will make a huge difference in the effectiveness of retail applications. Then again, maybe the technology vendors realize how difficult it is to get all the retail channel players to agree.