engleski

Developing text retrieval system using robust morphological parsing

The paper describes the way in which a text retrieval system for inflected languages, such as Croatian, can be improved. Based on a robust morphological analysis, it can search for all word forms without using a dictionary. Using all the possible combinations of morphological alternation and filtering final results through words in inverted index, a satisfactory level of precision can be obtained.

text retrieval; robust natural language processing;document indexing; free-text searching

nije evidentirano