Search technologies seek patterns within text and then rank them based on algorithms. Since natural language text is imprecise and nuanced, it is difficult if not impossible to derive prices meaning from NLT with any great confidence. But when documents contain a meaning model that relates to an open directory of meaning, expressed in precise semantic terms and grammar, it will be possible to connect queries directly to answers.
For instance, the English word "cardinal" can refer to a bird, a church official, a baseball team, a football team, a number, a matter of particular importance, etc. Search technology may use other factors to improve the contexual relevancy of its returns, but a site that makes use of an explicit, inline meaning model within the metadata of its pages will tell search bots exactly what things mean.
This will allow programmers to develop special programs for pages and sites that declare a semantic directory within their metadata. If the semantic model meets a known standard, it would be possible to include it within a utility that accepts structured queries to produce direct answers.
For example, if my directory of meaning includes the triple...
Barack Hussein Obama | has birthdate | 0524Z04081961
... then a semantic search utility would theoretically be able to find this statement and parse it into a direct answer.
This might require a different interface than the ubiquitous Google box, but its advantages would be considerable. A person seeking to learn exactly where and when and in what context Barack Obama made his comment about "at a certain point, you've made enough money," would have have to wade through screen after screen of bullshit to derive a useful answer, since search doesn't run on meaning, but on popular usage.
However, it's not necessary to develop new utilities to benefit from semantic structure. Even if we stuck to the same search tools we use today, a page with embedded inline semantic meaning will rank higher than the same page without that embedded metadata. That's because "searchbots love structure." They understand it, they make use of it, and they reward it.
Ironically, the first benefit for those who use these SCMS tools might well be good old SEO.
Recent Comments