278, 1980.SCREENED RESULTS WERE APPROPRIATE. THE MRRE. H. SHORTLIF...

231-278, 1980.

screened results were APPROPRIATE. The MRR

E. H. Shortliffe, “Computer-Based Medical Consultations:

for manual transcription (TRS) with no recognition

MYCIN,” Elsevier/North Holland, New York NY, 1976.

errors was 0.43. In addition, we could improve the

MRR from 0.25 (REC) to 0.28 (DQ) by using the

B. Lowerre et. al., “The Harpy speech understanding system,”W. A. Lea (Ed.), Trends in Speech recognition, pp. 340, Pren-

DQs only once. Experimental results revealed the

tice Hall.

potential of the generated DQs in compensating for

L. D. Erman et. al., “The Hearsay-II Speech-Understanding

the degradation of the QA performance due to recog-

System: Integrating Knowledge to Resolve Uncertainty,”

nition errors.

ACM computing Survays, Vol. 12, No. 2, pp. 213 – 253,