top of page
Bernd Geiger

New Built-ins available for OntoBroker



New Built-ins available for OntoBroker


Built-ins are the data processing components of OntoBroker on whose processing results the defined logic functions then draw the appropriate conclusions (the reasoning). Built-ins can also be created by the user and can thus significantly extend the functionality of OntoBroker. All built-ins are written in Java and can be added hot-pluggable.

There are built-ins with interface functionality (connectors to external data sources and processing units) and those with pure data processing functionality. Furthermore, there are built-ins that access external processes for data processing, e.g. the Python engine, which can be used for preprocessing by means of machine learning.

A fourth category are so-called utility built-ins, which e.g. make such data structure contexts explicitly accessible, where a syntactic (in this case symbolic) realization seems too fiddly for the application. Such a case we have with the access to predicate functions. Here, one can now use the _predToMap function to obtain all components of a predicate of first or higher order. For example:

?x = P(a,b,c)(e,f).

?- _predToMap(?x,?y).

Result:

[0->P,1->[0->a,1->b,2->c],2->[0->d,1->e]]

Another built-in for data processing is _wordDist, which gives the normalized Levenshtein distance of two strings. The function is useful to find searched words e.g. despite typos or plural/singular form. For example:

_wordDist("material", "materials",?x) for ?x gives a value of 0.118

_wordDist("material", "machine",?x) for ?x gives a value of 0.625

If it does not give any proximity, the value is 1 and 0 for a complete match.

The built-ins are available from OntoBroker V 6.3.

Comentarios


bottom of page