Natural language processing is an important area of computational linguistics and artificial intelligence . Mostly, its existing applications are developed for and based on English texts. We developed an application for the statistical analysis of large text corpora, which takes into account the unique characteristics of Slovene as a strongly inflected language. Since modern text corpora consist of several billion words, we paid special attention to efficient parallel algorithms that are capable of processing these collections in a relatively short amount of time. We analyzed the Gigafida corpus - consisting of 1.2 billion words - on multiple levels: string level, word level, n-gram level, prefix and suffix level, as well as word formation processes of Slovene.
|
