Creating an Example-Based Taxonomy from Scratch (SAP Library

Creating an Example-Based Taxonomy from Scratch

Purpose

You want to create an example-based taxonomy and corresponding folder structure from scratch. That is, you do not want to use an existing folder hierarchy for the taxonomy.

Process Flow

1. Create an index

You need to create an index before you can create an example-based taxonomy. You do this in the index administration iView in index administration (KM Admin ® Index Administration; see Creating an Index).

2. Assign data sources to the index

You now assign data sources to the newly created index. You want to index these data sources and create a taxonomy for them (see Assigning Data Sources). These data sources need to contain the example documents that you want to use to train your new taxonomy (see step 5).

3. Create the root nodes of the folder hierarchy

You now create the root nodes for the folder hierarchy of your taxonomy in index administration under Taxonomies. You create a new taxonomy and define the type of taxonomy at the same time. In this instance, it is an example-based taxonomy (see Creating Taxonomies for an Index).

4. Create the categories of the folder hierarchy

Now create the categories of the taxonomy manually in the form of individual folders. You do this in the Category Browser iView (Content Managers ® Repositories ® Taxonomies). Name the folders and create a new folder hierarchy targeted to your needs.

5. Train the categories of the taxonomy using example-documents

When creating an example-based taxonomy, you train the categories of your new taxonomy using example documents. Afterwards, the system can classify documents that are similar to the example documents into the categories of your taxonomy. You can train a taxonomy manually or using search queries.

¡ Training a taxonomy manually: You train a taxonomy manually by assigning example documents to the selected folder of your taxonomy’s folder hierarchy manually. You can do this in the Navigation in Documents iView (Content Manager ® Repositories ® Documents) or in the Classification Inbox (Content Manager ® Classification Inbox; see Structure link Classification Inbox).

¡ Training a taxonomy using search queries:You can also train a taxonomy by looking for suitable example documents by using search queries. To do this, you execute a search query and manually assign the documents found to a category as the category’s example documents. You do this in the Taxonomy Trainer iView (Content Manager ® Taxonomy Trainer) . This iView is only for training the categories of example-based taxonomies once (see Taxonomy Trainer).

6. Reindex by data source

After the training of categories using example documents, you need to reindex all data sources that are assigned to the index in question. This is the only way to classify the document belonging to the index into the categories that you have just trained (see Reindexing and Incremental Update).

7. Assign further data sources to the index
(optional)

You can also assign further data sources to the index (see Assigning Data Sources).

Result

When you have created an example-based taxonomy, you can begin the example-based classification of documents (See Structure link Classifying Documents).

This graphic is explained in the accompanying text

Note that the paths specified for iViews are valid for the standard delivery of Knowledge Management (KM). The paths given can be different to the paths in your portal. This depends on your configuration.