User manual BUSINESS OBJECTS INTELLIGENT SEARCH XI 3.0 SP1 FOR WINDOWS AND UNIX CLASSIFIER TRAINER

[. . . ] Using Classifier Trainer, you can create a taxonomy of subjects representing the categorization scheme desired. For each category or subject in the taxonomy, you can provide a set of sample documents that represent content that matches the desired category. Classifier Trainer creates a taxonomy model containing fingerprints for each category for the run-time classification engine inside the BusinessObjects Intelligent Search XI 3. 0 Server to use. This vector-space classifier is tuned to work on the metadata of each result record. Audience for this guide This guide is intended for system administrators and others responsible for installing and configuring BusinessObjects Intelligent Search XI 3. 0 and assumes that you are familiar with your Intelligent Search installation. [. . . ] In the absence of such a meta tag, Classifier Trainer assumes the system default character set. In addition, HTML documents should not use the UTF-16 Unicode encoding; they should use UTF-8 instead. Organize your training documents in hierarchical folders where each folder represents a category node in the training taxonomy. Each folder should have roughly the same number of documents to ensure that the training run processes representative vocabulary within, and across, all categories. Figure 3-1 :Example of a typical hierarchy. Configuring the training run To configure a training run 1. Create a directory on a drive that has ample space for the training documents and the temporary files created during the training process. The drive should have free space equal to triple the size of your training files. Copy your training documents into a subfolder of this directory. Copy the configuration files stopwords and config from <INSTALLDIR>/intelligentsearch/conf/stdclassifier to the directory you created. · · The stopwords file contains a list of words that Classifier Trainer should ignore during the training process. Add to this file any words that you want your training run to ignore. The config file contains these optional settings used during the training process and during classification: BusinessObjects Intelligent Search XI 3. 0 Classifier Trainer User's Guide 13 3 Running Classifier Trainer Configuring the training run Parameter removesingle removenumbers stemlevel Description Set removesingle to 1 if Classifier Trainer should ignore single character tokens. Set removenumbers to 1 if Classifier Trainer should ignore numeric tokens. Stemming is the process of converting words to root forms by removing suffixes. Set stemlevel to: · 0 for no stemming · · 1 for minimal stemming 2 for aggressive stemming. For example, given the term "nationalizing:" This value: 0 1 2 maxtokenlen language Stems to: nationalizing nationalize nation Indicates the longest token Classifier Trainer should accept. Classifier Trainer ignores tokens longer than maxtokenlen. An identifier for the language to be used for linguistic processing. The language identifier should consist of a two-letter language identifier, optionally followed by a dash and a two-letter country identifier. See RFC 1766 for more information. removefromcategory Classifier Trainer does not use this value during training. The value name of this setting is a regular expression used during classification to remove any portion of a category name that matches this expression. Note: In most cases you should remove this setting from the config file since it may cause category names to be modified during classification. Configuring known phrases (Optional) Classifier Trainer will automatically detect frequently occurring phrases in your training documents. You can configure additional phrases by creating a file named phrases. txt that contains phrases (one per line) that the classifier should recognize. 14 BusinessObjects Intelligent Search XI 3. 0 Classifier Trainer User's Guide Running Classifier Trainer Running Classifier Trainer 3 Running Classifier Trainer To run Classifier Trainer 1. Open a console window and change to the directory containing your training files. 2. Choose a name for your new classifier model. For this example, assume you have chosen the name myclassifier. 2. [. . . ] Contact your customer support representative to obtain an ID. If you are redirected to the SAP Service Marketplace from the SAP Help Portal, use the menu in the navigation pane on the left to locate the category containing the documentation you want to access. https://boc. sdn. sap. com/developer/library Developer resources https://www. sdn. sap. com/irj/sdn/businessobjectssdklibrary SAP BusinessObjects articles on the SAP Community Network https://www. sdn. sap. com/irj/boc/businessobjects-articles These articles were formerly known as technical papers. 22 BusinessObjects Intelligent Search XI 3. 0 Classifier Trainer User's Guide More Information A Information Resource Notes Forums on the SAP Community Network Location https://service. sap. com/notes These notes were formerly known as Knowledge Base articles. https://www. sdn. sap. com/irj/scn/forums http://www. sap. com/services/education From traditional classroom learning to targeted elearning seminars, we can offer a training package to suit your learning needs and preferred learning style. Training http://service. sap. com/bosap-support The SAP Support Portal contains information about Customer Support programs and services. [. . . ]