Difference between revisions of "Help:Semantic tagging"
Line 32: | Line 32: | ||
The two general cases are: | The two general cases are: | ||
− | #When attribute ''values'' are variable/non-standard e.g. job titles vs movie titles, which means one searches on attributes themselves, not attribute ''values''. In other words, [[Centiare:Tutorial#Use_uniform_titles_when_identifying_any_Key_Person|job titles]] may vary between different key people, such as president, vice-president, director, etc., | + | #When attribute ''values'' are variable/non-standard e.g. job titles vs movie titles, which means one searches on attributes themselves, not attribute ''values''. In other words, while [[Centiare:Tutorial#Use_uniform_titles_when_identifying_any_Key_Person|job titles]] may vary between different key people, such as president, vice-president, director, etc., there is only one (the original) "Gone With the Wind". Here's an example from the [[Special:Attributes|attributes list]]: |
<pre> | <pre> |
Revision as of 16:53, 9 January 2007
Help:Contents |
See also Help:Annotation for more detailed background information on semantic tags. Refer to the Attributes or Relations lists to see examples of existing tags or check whether specific annotations are already in use.
Overview
Almost all of the semantic tagging that's taking place in Centiare will fall into two types of scenarios -- Relations and Attributes.
Suppose you were writing an article about the city of Berlin, Germany. You could easily type out in the article that "Berlin is the capital of the unified country of Germany, and there are nearly 3.4 million people living in its metropolitan area." That's really good encyclopedia information.
However, if someone searches for the exact phrase "capital of Germany" or "population of Berlin", your sentence that happens to answer both of those questions would not be returned by either of those particular text searches. What we hope to see in Centiare is active use of the semantic tagging process, so that such information is more likely to be found -- by either humans typing in commands, or machines programmed to find information.
So, the essence of semantic tagging is, somewhere in the Berlin article text, or in an infobox, or even in an addendum at the bottom of the article, if you want to create a semantic link that describes a "capital-relationship", this is done by writing:
[[capital of::Germany]]
Note the use of two (2) colons in succession. You've just created a semantic tag Relation.
Furthermore, if you want to create a semantic link that describes a "population-attribute", this is done by writing:
[[population:=3,396,990]]
Note the use of the colon and equal sign in succession. You've just created a semantic tag Attribute.
Custom Tags
Attributes
Now that you've created your first semantic tag (an attribute), you may wish to explore some of the advantages of using ordinal identifiers (Centiare convention is numbers) to further differentiate between certain related attributes.
The two general cases are:
- When attribute values are variable/non-standard e.g. job titles vs movie titles, which means one searches on attributes themselves, not attribute values. In other words, while job titles may vary between different key people, such as president, vice-president, director, etc., there is only one (the original) "Gone With the Wind". Here's an example from the attributes list:
Key Person1 Title Key Person2 Title
- When attributes have relations to other attributes e.g. key person<->job title. Expanding upon the example shown above, job titles can be associated with the respective individuals who hold those positions:
Key Person1 Key Person1 Title Key Person2 Key Person2 Title Key Person3 Key Person3 Title Key Person4 Key Person4 Title
In this way, friends/interests don't require ordinal identifiers, unless at some point there are secondary attribute relations. For example friends<->gender, interests<->summer/winter, etc.
Relations
Subjects such as interests can also be semantically tagged as relations, since common Directory or Main pages can be interlinked to multiple users via ASK queries (see Skiing, etc).
Search Queries
Centiare uses a parser function called ASK to perform free-form queries. ASK provides a ready complement to the Search Triple query-by-form facility.
When people use the "Search Triple" feature in Centiare, they will have utmost confidence and success in finding the articles they are looking for, if they correctly use the search forms, and you've correctly tagged your articles for semantic searching.
The possibilities for this are literally limitless, for both businesses and individuals. Imagine conducting a search for a male, born in Michigan between 1965 and 1968, who has interests in both skiing and poker. Do you think that would be easy with MySpace, Wikipedia, or Google? Fat chance. But on Centiare, it will be a piece of cake, if you learn to use the ASK function.