From web directories to ontologies: natural language processing challenges

摘要: Hierarchical classifications are used pervasively by humans as a means to organize their data and knowledge about the world. One of main advantages is that natural language labels, describe contents, easily understood human users. However, at same time, this also one disadvantages these labels ambiguous very hard be reasoned software agents. This fact creates an insuperable hindrance for being embedded in Semantic Web infrastructure. paper presents approach converting into lightweight ontologies, it makes following contributions: (i) identifies NLP problems related conversion process shows how they different from classical NLP; (ii) proposes heuristic solutions problems, which especially effective domain; (iii) evaluates proposed testing them on DMoz data.