Creating Topic Sets

In K2, a topic is a named expression in the Verity Query Language (VQL), designed to locate documents related to a given concept or subject area. For example, a topic named DaimlerChrysler Cars might consist of a query expression that searches for that term itself, plus any corporate division names (like Mercedes Benz), plus the names of any of its automobile lines (like Chrysler), plus the names of any of its individual car models (like PT Cruiser). For this simple example, the body of the topic might contain VQL elements like these:

DaimlerChrysler Cars <OR> Mercedes Benz <OR> ... <OR> Chrysler <OR> PT Cruiser <OR> Caravan <OR> ...<OR> Jeep <OR> <WORD<CASE> Wrangler <OR>...

A more specific topic, such as hybrid cars, might consist of a query that searches for Honda Insight and Toyota Prius and possibly technical or legislative terms in documents relating to low-emission vehicles.

Topics can be created manually by domain experts or knowledge workers who understand how to express a concept in terms of search-query strings. They can also be created automatically, using machine-learning tools. An individual topic can be a simple, short expression, or it can be a long and complex one, involving many terms and sophisticated boolean and non-boolean search operators.

Topics can be combined and compiled into groups called topic sets. A topic set can have a flat structure or it can be hierarchical.

Administrators can create topic sets using either Intelligent Classifier or the mktopics command-line tool. K2 applications can use topic sets in several ways:

A topic set can be attached to an application to integrate it into the application’s search capabilities.

 

A topic set can be indexed into a collection to provide extra-fast searches over the collection for the topic set’s terms.

 

A topic set can also be used as the set of business rules for populating the taxonomy of a knowledge tree; see Creating Category Definitions.

 

Topics are also used as the basis for profile nets; see Creating Profile Nets.

 

 

 
Note   Topic sets are commonly used as sets of query expressions only; the topic names are used as search terms but do not necessarily appear in a K2 application’s user interface. If you want to implement a topic-set structure that users can directly browse, you can create it in a parametric index, as described in Relational Taxonomies.
 

 

For more information on topics and topic sets, see the Verity Collaborative Classifier Guide and the Verity Query Language and Topic Guide.