Lightweight Ontologies

Classifications are perhaps the most natural tool humans use to organize information content. Information items are hierarchical arranged under topic nodes moving from general ones to more specific ones as long as we go deep in the hierarchy. This attitude is well known in Knowledge Organization as the principle of organizing from the general to the specific, called synthetically the get-specific principle in “Encoding Classifications As Lightweight Ontologies”.

Classifications content is usually described using natural language labels (see Figure 1), which has been proved to be very effective in manual tasks (for example, to index documents, to search and navigate the tree). However, natural language labels show their limitations when one tries to automate reasoning over them, for instance for automatic indexing and semantic matching or when dealing with multiple languages.

Two example course catalogs classifications
Figure 1: Two example course catalogs classifications

Therefore, a fundamental preliminary step is to translate classifications into their formal alter-ego, namely into lightweight ontologies. Following the approach described in “Encoding Classifications As Lightweight Ontologies” and exploiting dedicated Natural language processing (NLP) techniques tuned to short phrases (for instance, as described in “From Web Directories To Ontologies: Natural Language Processing Challenges”), each node label can be translated into an unambiguous formal expression, i.e. into a propositional Description Logic (DL) expression. As a result, lightweight ontologies, or formal classifications, are tree-like structures where each node label is a language-independent propositional DL formula codifying the meaning of the node. Taking into account its context (namely the path from the root node), each node formula is subsumed by the formula of the node above. As a consequence, the backbone structure of a lightweight ontology is represented by subsumption relations between nodes. “Encoding Classifications As Lightweight Ontologies” provides some examples of lightweight ontologies. “Semantic Matching” and “Computing Minimal Mappings” show how lightweight ontologies can be used to automate important tasks, in particular to favor interoperability among different knowledge organization systems.