TY - GEN
T1 - A new topic filter based on maximum entropy model
AU - Chen, Chen
AU - Liu, Huilin
AU - Wang, Guoren
AU - Yu, Lili
PY - 2009
Y1 - 2009
N2 - Because of the large web scale and the information requirement for special field, focuse2825453011d search has attracted more and more people. For the complexity of natural language, there are ambiguous for a word itself, and which will take some trouble for topic filter. For the two main problems, false positive and false negative, this paper proposes two new methods separately. By machine learning, we construct a guide model with the maximum entropy principle, by which we can filter the noise pages out easily and by KNN method, the false negative problem will be solved easily. The experiment shows that our model or method really outperforms the base-line method.
AB - Because of the large web scale and the information requirement for special field, focuse2825453011d search has attracted more and more people. For the complexity of natural language, there are ambiguous for a word itself, and which will take some trouble for topic filter. For the two main problems, false positive and false negative, this paper proposes two new methods separately. By machine learning, we construct a guide model with the maximum entropy principle, by which we can filter the noise pages out easily and by KNN method, the false negative problem will be solved easily. The experiment shows that our model or method really outperforms the base-line method.
KW - Focused search
KW - KNN
KW - Maximum entropy
KW - Noise pages
KW - Topic filter
UR - http://www.scopus.com/inward/record.url?scp=76549093344&partnerID=8YFLogxK
U2 - 10.1109/FSKD.2009.709
DO - 10.1109/FSKD.2009.709
M3 - Conference contribution
AN - SCOPUS:76549093344
SN - 9780769537351
T3 - 6th International Conference on Fuzzy Systems and Knowledge Discovery, FSKD 2009
SP - 495
EP - 499
BT - 6th International Conference on Fuzzy Systems and Knowledge Discovery, FSKD 2009
T2 - 6th International Conference on Fuzzy Systems and Knowledge Discovery, FSKD 2009
Y2 - 14 August 2009 through 16 August 2009
ER -