engleski

Corpus Analysis of Complex Names with Common Nounsin Croatian

The goal of this corpus-based researchis to see can the complex names with common nouns in their composition be extracted from Croatian hrWaC v2.2 corpus by using regular expressions, i.e. to what extent the capital letter (not the one after the full stop, the exclamation mark or the question mark)can be taken as an indication of a name. Common noun can be used as a regular noun or as a constituent of a complex name, which, on one hand, makes it difficult to tag them automatically, and on the other hand, affects the lexicographic description. With the help of regular expressions, we searched for capitalized common nouns and for sequences in which a capitalized attribute is on the first place and the common noun follows it. After analyzing 1000 examples in each search, we divided results into two groups: names and sequences with an uppercase letter that are not names. Some of the causes of extracting “false” names are technical (e.g. interpunction:separating sentences with paragraph mark(¶), lack of interpunction at the end of sentence ; whole parts of textwrit-ten in upper case...), and some of them lie in the texts crawled for hrWaC, which are not written in accordance with Croatian orthography.

Complex Names, Croatian Orthography, Corpus Search

nije evidentirano