What question did this study set out to answer?

To develop and assess effective text classification methods trained exclusively on positive and unlabeled data in the complete absence of negative training examples.

April 23, 2004

Building text classifiers using positive and unlabeled examples

View Full Paper

Ask AI

Bookmark

Authors

BLB. LiuNanchang Institute of Technology YDY. DaiChongqing University XLXiaoli LiInstitute for Infocomm Research

Discussion

Member takes

Implication

Experimental evaluation demonstrates superior text classification using biased support vector machines without negative examples, highlighting improved accuracy for partially labeled data.

Key Points

To develop and assess effective text classification methods trained exclusively on positive and unlabeled data in the complete absence of negative training examples.
Introduced novel methods for the two-step classification framework and systematically evaluated all combinations of methods across both stages.
Formulated and implemented a theoretically principled approach based on a biased formulation of support vector machines (SVM).
Comparative testing established the relative effectiveness of different method combinations across the two-step framework.
The biased SVM formulation consistently outperformed existing two-step techniques in classification accuracy across experimental evaluations.

Cite This Study

Liu et al. (2004) studied this question.

synapsesocial.com/papers/6a1222fdc031bb6829a5f89c https://doi.org/10.1109/icdm.2003.1250918

View Full Paper

Ask AI

Bookmark

Implication

Experimental evaluation demonstrates superior text classification using biased support vector machines without negative examples, highlighting improved accuracy for partially labeled data.

Building text classifiers using positive and unlabeled examples

Discussion

Member takes

Implication

Key Points

Cite This Study

Building text classifiers using positive and unlabeled examples

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider