?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Exploiting+context+when+learning+to+classify&rft.creator=Turney%2C+Peter&rft.subject=Artificial+Intelligence&rft.subject=Machine+Learning&rft.subject=Statistical+Models&rft.description=This+paper+addresses+the+problem+of+classifying+observations+when%0Afeatures+are+context-sensitive%2C+specifically+when+the+testing+set+involves+a+context%0Athat+is+different+from+the+training+set.+The+paper+begins+with+a+precise+definition+of%0Athe+problem%2C+then+general+strategies+are+presented+for+enhancing+the+performance%0Aof+classification+algorithms+on+this+type+of+problem.+These+strategies+are+tested+on%0Atwo+domains.+The+first+domain+is+the+diagnosis+of+gas+turbine+engines.+The%0Aproblem+is+to+diagnose+a+faulty+engine+in+one+context%2C+such+as+warm+weather%2C%0Awhen+the+fault+has+previously+been+seen+only+in+another+context%2C+such+as+cold%0Aweather.+The+second+domain+is+speech+recognition.+The+problem+is+to+recognize%0Awords+spoken+by+a+new+speaker%2C+not+represented+in+the+training+set.+For+both%0Adomains%2C+exploiting+context+results+in+substantially+more+accurate+classification.&rft.date=1993&rft.type=Conference+Paper&rft.type=PeerReviewed&rft.format=application%2Fpdf&rft.identifier=http%3A%2F%2Fcogprints.org%2F1863%2F3%2FNRC-35058.pdf&rft.identifier=++Turney%2C+Peter++(1993)+Exploiting+context+when+learning+to+classify.++%5BConference+Paper%5D+++++&rft.relation=http%3A%2F%2Fcogprints.org%2F1863%2F