In a recent issue focusing on risk one article in Bloomberg Businessweek made the statement, “Such semantics may feel like distinctions without a difference. For underwriters in the specialty market of political-violence insurance, though, they’re crucial.” The semantic distinctions had to do with the labeling of political “protests” in Egypt and elsewhere in the Arab world.  I put “protests” in quotes to draw attention to the issue of semantics and its impact on insurance and risk assessment. 

Politic-violence and terrorism insurance hinges almost more on words than tangible, physical damage. If your building is burned to the ground, it’s burned to the ground and there are no doubts. It is a demonstrable fact, but political-violence insurers first look at the causes, not the consequences and this is where the semantics come to play such an important role. The Bloomberg Businessweek article chronicles the escalation in the rhetoric surrounding the developments in Cairo from “civil commotion” to “insurrection” to “mutiny”. It further reports on such event-distinctive classes as sabotage, terrorism, strikes, riots, civil commotion, insurrection, and outright war. Insurers assess risk for each class they will insure and include classes they will not insure. Risk assessment includes the likelihood of an event occurring and the associated costs.

The semantics are used to classify events for the purposes of setting rates and determining liabilities, and, of course, liabilities determine payouts and profits, or not. Insurers and the insured have a natural tendency to classify events differently. What does the policy cover and how was the incident labeled? Do you have insurance covering riots, but the event is classified as a civil disturbance? Is a civil disturbance the same as a civil commotion? Ones perspective will be very different if it is your building that burns verses your neighbors. (This harkens back to economic definitions and the old joke, “If your neighbor gets laid off, it’s a recession. If you get laid off, it’s a depression.”)

Using semantic principles and practices, taxonomists can create a description of a domain in the form of a taxonomy or thesaurus.  There are examples of taxonomies for the insurance industry here and here.  The classes illustrated in a taxonomy can then be used to classify an event, but classifying an event is very different than classifying an article, particularly when you’re talking about an insurance event. Classifying an article, even an article from an insurance industry publication, would typically receive multiple classifications. That is, you would assign several concepts from the taxonomy to the article. This process is really indexing by the assignment of subject terms from the taxonomy. Multiple concepts, or terms, are assigned because articles frequently discuss multiple topics. Also, keeping in mind that different users will come at a search requirement from different perspectives, multiple classes accommodate diverse routes to the same piece of content. 

An insurance event would normally get classified into only a single class. A destroyed building would not be classified into both the “civil commotion” class and the “insurrection” class. One classification may mean getting compensation and an alternate class would leave you high and dry. Does “flood” insurance cover damage by a “tsunami”? Both involve water, but are they the same?

A well-crafted taxonomy used by both the insurer and the insured would go a long way toward removing the ambiguity that leads to disputes. A powerful thesaurus and taxonomy management tool like our Data Harmony® Thesaurus Master® can conceptually and visually show relationships and convey meaning more accurately. Each term should have a tight definition. The relationships between terms can be clearly shown such as using a hierarchical display to indicate increasing degrees of severity and risk. Additional information about a concept and relations between concepts can be added to each term record. The results could be a system like xpeerient’s that I described in a recent blog. xpeerient developed a taxonomy and developed a customized version of Access Innovation’s Data Harmony M.A.I.™ that facilitates the mapping of IT buyers to IT vendor solutions at a level of specificity not possible before. Such an approach could do a much better job of matching insurance buyers with insurance vendors. What is the semantic risk here? 

What is your semantic risk?

Jay Ven Eman, CEO
Access Innovations