ATG Search uses rules to place documents and queries into a taxonomy of categories (described in the “Automatically Associating Content with Topics” section of the ATG Search Administration Guide). During a query, ATG Search can apply these rules to the query and return the categories as part of the result. ATG Search can also review the candidate document results for a query and collect the most relevant or frequent categories and return them as part of the result. The categorization of the query and the candidate documents may not be the same, the former containing the categories most relevant to the query and the latter containing the categories most relevant to the results. Both represent feedback for the end-user to enable further refinement of the query.
Because the categories are organized in a taxonomy, ATG Search can extend the categorization feedback to contain a tree of the categories of the returned document results. Instead of a flat list of the most relevant categories, ATG Search returns the portion of the taxonomy that contains the most relevant categories. The user interface that receives this tree can display it along with the results.
The figure that follows illustrates the process of extracting the category result tree. On the left-side of the figure is the taxonomy, indented to show the hierarchical relationships, Cat1 has two children named Cat2 and Cat3. On the right-hand side are the final results for the query, whose index items belong to one or more categories as shown by the arrows. Categories without any references to results are eliminated from the result tree if none of their descendants have references (Cat4 and Cat6). Categories without any references to results are included if some of their descendents have references to results (Cat1). Thus, the result tree contains the minimum portion of the taxonomy which references the result list. Note that a result can be referenced by multiple categories (Result5) and a category can reference multiple results (Cat2).
Category Result Tree
