BLN Learning Tool

Start the tool from the command line with blnlearn.

The BLN learning tool learns the conditional probabilties and domains of random variables from a given training database and the fragment network. The tool allow you to invoke the actual BLN learning algorithms. Once you start the actual algorithm, the tool window itself will be hidden as long as the job is running, while the output of the algorithm is written to the console for you to follow. At the beginning, the tools list the main input parameters for your convenience, and, at the end, the query tool additionally outputs the inference results to the console.

The tool lets you choose the BLN fragment network which you can alter using the "show" button (this starts the Network Editor with the currently selected fragment network). The tool also features an integrated editor for the basic model description file. If you modify a file in the internal editor, it will automatically be saved as soon as you invoke the learning method. The new content can either be saved to the same file (overwriting the old content) or a new file, which you can choose to name as desired. Furthermore, the tool will save all the settings you made whenever the learning method is invoked, so that you can easily resume a session.

The training database to learn from can either consist of one single database or a number of databases that are given to the BLN learning tool by a file name pattern.

You have the option to not only learn the parameters but also the fixed domains from the database (by ticking the option ''learn domains'').

By default, it is assumed that your model can make use of all the data that appears in the training database(s). If functions/predicates should appear in the data that are not considered in the model, an exception will be raised. If you simply want to ignore this, tick the option ''ignore data on undefined predicates''.

In the field ''Add. params'' you may specify additional parameters to pass on to BLNlearn; for example:

• the option to apply uniform distribution by default (for CPT columns with no examples). (Add. Parameter: -ud)
• the option to keep counts in the CPTs without normalization. (Add. Parameter: -nn)
• the option to convert the learnt model to a Markov logic network (Add. Parameter -mln)