Japanese Word Sketches: Advances and Problems
|
| Title | Japanese Word Sketches: Advances and Problems |
| Authors | |
| Abstract | In this paper, we present results of an evaluation of Japanese word sketches and address in detail issues that were observed by the evaluators. A word sketch presents a list of salient collocates of a word, organized by the grammatical relations holding between the word and its collocate. The word sketch functionality is incorporated into the Sketch Engine corpus query system and has been created for more than twenty languages so far, including Japanese. The issues that have been discovered in the evaluation of word sketches in Japanese are to be addressed for further enhancement of the word sketch functionality. Other tools and resources which are combined for use and influence the performance of the word sketches should also be looked over. We divide the issues into the following: 1) the lemmatizer and tagger in use, 2) the sketch grammar that is specifically written for Japanese, and 3) the corpus and statistical methods.-----V prispevku predstavljamo rezultate ocenjevanja japonskih besednih skic in podrobno prikazujemo probleme in težave, ki smo jih opazili ocenjevalci. Besedna skica je seznam izstopajočih kolokacij neke besede, ki ga organizirajo slovnične relacije med besedo in drugimi besedami, ki skupaj sestavaljajo kolokacije. Funkcije besedne skice so vgrajene v korpusno orodje Sketch Engine in na voljo trenutno že v več kot dvajsetih jezikih, med njimi tudi v japonščini. Problemi in težave, ki smo jih odkrili med ocenjevanjem besednih skic v japonščini, moramo dalje proučiti za okrepitev funkcij besednih skic. Problemi in težave so pri naslednjih: 1) pri sistemu ugotavljanja osnovne oblike besede in označevalcu besednih vrst v rabi; 2) v slovnici za skice, ki je napisana posebej za japonščino; 3) pri korpusu in statističnih metodah. |
| Publisher | Faculty of Arts |
| Date | 2011-10-20 |
| Source | Acta Linguistica Asiatica Vol 1, No 2 (2011) |
| Rights | Attribution 3.0 Unported (CC BY 3.0) |