| Abstract: |
This paper is devoted to the Compreno multilingual lexical database – an integral model of natural language description designed for the semantic analysis of texts and the various tasks based on such an analysis, for example machine translation or fact extraction. Semantic data is stored in the form of the universal semantic hierarchy, which is currently filled with English, Russian, and, partly, French, German and Chinese contents. The description of each word meaning includes its positioning in the semantic hierarchy, definition, all the necessary semantic and grammatical information, collocations, and all possible semantic links for the word, both actant and circumstantial. The paper gives a detailed description of the Compreno semantic pattern and focuses especially on the universality problem, namely, on the difficulties that arise when creating a semantic model appropriate for processing different languages. |