Runs - Spam 2006¶

B53S3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: B53S3F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: online
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

B53S3pcd¶

Run ID: B53S3pcd
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

B53S3pci¶

Run ID: B53S3pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

B53S3ped¶

Run ID: B53S3ped
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

B53S3pei¶

Run ID: B53S3pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

BASA2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: BASA2F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: active
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

BASA2pci¶

Run ID: BASA2pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

BASA2pei¶

Run ID: BASA2pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

BASS2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: BASS2F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: online
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

BASS2pcd¶

Run ID: BASS2pcd
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/17/2006
Type: online
Task: run

BASS2pci¶

Run ID: BASS2pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/17/2006
Type: online
Task: run

BASS2ped¶

Run ID: BASS2ped
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/17/2006
Type: online
Task: run

BASS2pei¶

Run ID: BASS2pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/17/2006
Type: online
Task: run

CRMS1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: CRMS1F
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This is essentially unchanged from last year's submission; it is the Markov Random Field classifier with OSB (Orthogonal Sparse Bigram) features of length 2,3,4,and 5, and uses a thick-threshold training algorithm with a base thickness of 10 units +/- the ambivalent centerpoint. This is basically the same as last year's better submission, but now we get to see how it fares against delayed training.

CRMS1Fchi¶

Run ID: CRMS1Fchi
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS1Fchidly¶

Run ID: CRMS1Fchidly
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS1Fdelay¶

Run ID: CRMS1Fdelay
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS1Ffull¶

Run ID: CRMS1Ffull
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: CRMS2F
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This is it is the Markov Random Field classifier with OSBU (Orthogonal Sparse Bigram) unique features of length 2,3,4,and 5, and uses a thick-threshold training algorithm with a base thickness of 10 units +/- the ambivalent centerpoint. Additionally, it keeps a log of all previously trained files and their known types, and retrains these periodically. The goal is to prevent the "dip rebound" that previous versions of CRM114 OSBU. The periodic retraining is incremental - for each new text, two old ones are retrained.

CRMS2Fchi¶

Run ID: CRMS2Fchi
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS2Fchidly¶

Run ID: CRMS2Fchidly
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS2Fdelay¶

Run ID: CRMS2Fdelay
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS2Ffull¶

Run ID: CRMS2Ffull
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: CRMS3F
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This is the Markov Random Field classifier with OSB (Orthogonal Sparse Bigram) unique features of length 2,3,4,and 5, and uses a thick-threshold training algorithm with a base thickness of 20 units +/- the ambivalent centerpoint, which is twice the base thickness of the basic default system. This configuration uses more memory buckets than the default system as well.

CRMS3Fchi¶

Run ID: CRMS3Fchi
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS3Fchidly¶

Run ID: CRMS3Fchidly
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS3Fdelay¶

Run ID: CRMS3Fdelay
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS3Ffull¶

Run ID: CRMS3Ffull
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS4F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: CRMS4F
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This is the Hyperspace (radiance KNN) classifier with OSB (Orthogonal Sparse Bigram) unique features of length 2,3,4,and 5, and trains best as here, in a very thin threshold. This filter is believed to be not as accurate as the other configurations, but it runs very VERY fast.

CRMS4Fchi¶

Run ID: CRMS4Fchi
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS4Fchidly¶

Run ID: CRMS4Fchidly
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS4Fdelay¶

Run ID: CRMS4Fdelay
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

CRMS4Ffull¶

Run ID: CRMS4Ffull
Participant: mitsubhishi.yerazunis
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

dalS1F¶

Results | Participants | Summary | Appendix

Run ID: dalS1F
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 7/14/2006
Type: online
Task: filter
Run description: Variation on a byte n-gram technique.

dalS1pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS1pcd
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS1pci¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS1pci
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS1ped¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS1ped
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS1pei¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS1pei
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS2F¶

Results | Participants | Summary | Appendix

Run ID: dalS2F
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 7/14/2006
Type: online
Task: filter
Run description: Variation on a byte n-gram technique.

dalS2pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS2pcd
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS2pci¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS2pci
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS2ped¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS2ped
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS2pei¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS2pei
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS3F¶

Results | Participants | Summary | Appendix

Run ID: dalS3F
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 7/14/2006
Type: online
Task: filter
Run description: Variation on a byte n-gram technique.

dalS3pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS3pcd
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS3pci¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS3pci
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS3ped¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS3ped
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS3pei¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS3pei
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS4F¶

Results | Participants | Summary | Appendix

Run ID: dalS4F
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 7/14/2006
Type: online
Task: filter
Run description: Variation on character n-gram technique

dalS4pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS4pcd
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS4pci¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS4pci
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS4ped¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS4ped
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

dalS4pei¶

Results | Participants | Input | Summary | Appendix

Run ID: dalS4pei
Participant: dalhousieu.keselj
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

hitA1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hitA1F
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: active
Task: filter
Run description: This system explores the feasibility of constructing an SVM (Support Vector Machines)classifier for Spam Filter task. In this task, we adopts the IG (Information Gain) to select feature so that the feature noise is reduced. In active learning, the Informativeness and Diversity Criteria are adopted to achieve the high performance.

hitA1pci¶

Run ID: hitA1pci
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/21/2006
Type: active
Task: run

hitA1pei¶

Run ID: hitA1pei
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/21/2006
Type: active
Task: run

hitA2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hitA2F
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: active
Task: filter
Run description: This system explores the feasibility of constructing an SVM (Support Vector Machines)classifier for Spam Filter task. This system presents a new feature selection algorithm with the category information analysis in text classification. The algorithm is distinguished from others by providing a pre-fetching technique for classifier while it is compatible with efficient feature selection.In active learning, the Informativeness and Diversity Criteria are adopted to achieve the high performance.

hitA2pci¶

Run ID: hitA2pci
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/21/2006
Type: active
Task: run

hitA2pei¶

Run ID: hitA2pei
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/21/2006
Type: active
Task: run

hitS1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hitS1F
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This system explores the feasibility of constructing an SVM (Support Vector Machines)classifier for Spam Filter task. In this framework, we adopts the IG (Information Gain) to select feature so that the feature noise is reduced. The feasibility of the approach has been checked on spam assassin corpus and Trec05 spam trecs. The results show that it can outperforms the baseline algorithm with high performance and efficiency.

hitS1pcd¶

Run ID: hitS1pcd
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/21/2006
Type: online
Task: run

hitS1pci¶

Run ID: hitS1pci
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS1ped¶

Run ID: hitS1ped
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS1pei¶

Run ID: hitS1pei
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hitS2F
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This system explores the feasibility of constructing an SVM (Support Vector Machines)classifier for Spam Filter task. In this framework, we adopts the IG (Information Gain) to select feature so that the feature noise is reduced. The feasibility of the approach has been checked on spam assassin corpus and Trec05 spam trecs. The results show that it can outperforms the baseline algorithm with high performance and efficiency.

hitS2pcd¶

Run ID: hitS2pcd
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS2pci¶

Run ID: hitS2pci
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS2ped¶

Run ID: hitS2ped
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS2pei¶

Run ID: hitS2pei
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hitS3F
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This system explores the feasibility of constructing an SVM (Support Vector Machines)classifier for Spam Filter task. This system presents a new feature selection algorithm with the category information analysis in spam detection. The algorithm obscure or reduce the noises of text features by computing the feature contribution with word and document frequency and introducing variance mechanism to mine the latent category information.

hitS3pcd¶

Run ID: hitS3pcd
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS3pci¶

Run ID: hitS3pci
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS3ped¶

Run ID: hitS3ped
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hitS3pei¶

Run ID: hitS3pei
Participant: harbin.zhao
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubA1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubA1F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: active
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with othogonal sparse bigrams.

hubA1pci¶

Run ID: hubA1pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubA1pei¶

Run ID: hubA1pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubA2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubA2F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: active
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with orthogonal sparse bigrams.

hubA2pci¶

Run ID: hubA2pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubA2pei¶

Run ID: hubA2pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubA3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubA3F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: active
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with orthogonal sparse bigrams.

hubA3pci¶

Run ID: hubA3pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubA3pei¶

Run ID: hubA3pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubA4F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubA4F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: active
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with othogonal sparse bigrams.

hubA4pci¶

Run ID: hubA4pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubA4pei¶

Run ID: hubA4pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

hubS1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubS1F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with orthogonal sparse bigrams.

hubS1pcd¶

Run ID: hubS1pcd
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS1pci¶

Run ID: hubS1pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS1ped¶

Run ID: hubS1ped
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS1pei¶

Run ID: hubS1pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubS2F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with othogonal sparse bigrams.

hubS2pcd¶

Run ID: hubS2pcd
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS2pci¶

Run ID: hubS2pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS2ped¶

Run ID: hubS2ped
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS2pei¶

Run ID: hubS2pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubS3F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with orthogonal sparse bigrams.

hubS3pcd¶

Run ID: hubS3pcd
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS3pci¶

Run ID: hubS3pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS3ped¶

Run ID: hubS3ped
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS3pei¶

Run ID: hubS3pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS4F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: hubS4F
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This filter was developed by the Knowledge Management Group of Humboldt University, Berlin in association with Strato AG. It is based on a winnow classifier with othogonal sparse bigrams.

hubS4pcd¶

Run ID: hubS4pcd
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS4pci¶

Run ID: hubS4pci
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS4ped¶

Run ID: hubS4ped
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

hubS4pei¶

Run ID: hubS4pei
Participant: humboldtu.haider
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

ijsA1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: ijsA1F
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: active
Task: filter
Run description: Active learning using an extremely simple heuristic The most recent N messages in the email stream are trained on.

ijsA1pci¶

Run ID: ijsA1pci
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: active
Task: run

ijsA1pei¶

Run ID: ijsA1pei
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: active
Task: run

ijsA2pci¶

Run ID: ijsA2pci
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: active
Task: run

ijsA2pei¶

Run ID: ijsA2pei
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: active
Task: run

ijsS1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: ijsS1F
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: This filter contains a straightforward implementation of the PPM compression scheme. Messages are classification to the most probable class, as determined by the compression rate (i.e. probability of target document) exhibited by PPM models of ham and spam. The PPM model is adaptive, i.e. it is updated with statistics from the target document at each character position as the target document is scanned. Uses an order-6 PPM model, escape method D, full alphabet (all 256 symbols) and exclusion of seen symbols when estimating backoff probabilities. This system uses essentially the same algorithm as our 2005 submission labelled ijsSPAM2.

ijsS1pcd¶

Run ID: ijsS1pcd
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

ijsS1pci¶

Run ID: ijsS1pci
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

ijsS1ped¶

Run ID: ijsS1ped
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

ijsS1pei¶

Run ID: ijsS1pei
Participant: jozef-stefan-inst.bratko
Track: Spam
Year: 2006
Submission: 8/23/2006
Type: online
Task: run

KB3A1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: KB3A1F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/12/2006
Type: active
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

KB3A1pci¶

Run ID: KB3A1pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

KB3A1pei¶

Run ID: KB3A1pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

KB3S1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: KB3S1F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/12/2006
Type: online
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

KB3S1pcd¶

Run ID: KB3S1pcd
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

KB3S1pci¶

Run ID: KB3S1pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

KB3S1ped¶

Run ID: KB3S1ped
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

KB3S1pei¶

Run ID: KB3S1pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

KB9A3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: KB9A3F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: active
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

KB9A3pci¶

Run ID: KB9A3pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

KB9A3pei¶

Run ID: KB9A3pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

KB9S4F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: KB9S4F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: online
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

KB9S4pcd¶

Run ID: KB9S4pcd
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

KB9S4pci¶

Run ID: KB9S4pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

KB9S4ped¶

Run ID: KB9S4ped
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

KB9S4pei¶

Run ID: KB9S4pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/16/2006
Type: online
Task: run

oflA1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: oflA1F
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: active
Task: filter
Run description: OSBF-Lua is a typical Bayesian classifier, but enhanced with OSB, a feature extraction technique, as its front-end and EDDC for feature selection.

oflA1pci¶

Run ID: oflA1pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflA1pei¶

Run ID: oflA1pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflA2pci¶

Run ID: oflA2pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflA2pei¶

Run ID: oflA2pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflA3pci¶

Run ID: oflA3pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflA3pei¶

Run ID: oflA3pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflA4pci¶

Run ID: oflA4pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflA4pei¶

Run ID: oflA4pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/7/2006
Type: active
Task: run

oflS1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: oflS1F
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: online
Task: filter
Run description: OSBF-Lua is a typical Bayesian classifier, but enhanced with OSB, a feature extraction technique, as its front-end and EDDC for feature selection.

oflS1pcd¶

Run ID: oflS1pcd
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS1pci¶

Run ID: oflS1pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS1ped¶

Run ID: oflS1ped
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS1pei¶

Run ID: oflS1pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: oflS2F
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: online
Task: filter
Run description: OSBF-Lua is a typical Bayesian classifier, but enhanced with OSB, a feature extraction technique, as its front-end and EDDC for feature selection.

oflS2pcd¶

Run ID: oflS2pcd
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS2pci¶

Run ID: oflS2pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS2ped¶

Run ID: oflS2ped
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS2pei¶

Run ID: oflS2pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: oflS3F
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: online
Task: filter
Run description: OSBF-Lua is a typical Bayesian classifier, but enhanced with OSB, a feature extraction technique, as its front-end and EDDC for feature selection.

oflS3pcd¶

Run ID: oflS3pcd
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS3pci¶

Run ID: oflS3pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS3ped¶

Run ID: oflS3ped
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS3pei¶

Run ID: oflS3pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS4F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: oflS4F
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: online
Task: filter
Run description: OSBF-Lua is a typical Bayesian classifier, but enhanced with OSB, a feature extraction technique, as its front-end and EDDC for feature selection.

oflS4pcd¶

Run ID: oflS4pcd
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS4pci¶

Run ID: oflS4pci
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS4ped¶

Run ID: oflS4ped
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

oflS4pei¶

Run ID: oflS4pei
Participant: fidelis.assis
Track: Spam
Year: 2006
Submission: 8/3/2006
Type: online
Task: run

tamS1F¶

Results | Participants | Summary | Appendix

Run ID: tamS1F
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 7/12/2006
Type: online
Task: filter
Run description: SpamBayes 1.1a2 (http //spambayes.org), with modifications to improve filtering of wide-character email, particularly Asian. No pre-learning, chi-squared combination, with a semi-arbitrary 0.4 ham and spam cutoff.

tamS1pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS1pcd
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS1pci¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS1pci
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS1ped¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS1ped
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS1pei¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS1pei
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS2F¶

Results | Participants | Summary | Appendix

Run ID: tamS2F
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: SpamBayes (spambayes.org) 1.a2 chi-squared combining, semi-arbitrary 0.4 cutoffs, bigram option enabled, otherwise defaults. Mostly split-on-whitespace tokenization. Trains on classify. Train-on-error. Doesn't use classification if fewer than 10 ham and 10 spam trained.

tamS2pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS2pcd
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS2pci¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS2pci
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS2ped¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS2ped
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS2pei¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS2pei
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS3F¶

Results | Participants | Summary | Appendix

Run ID: tamS3F
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 7/12/2006
Type: online
Task: filter
Run description: SpamBayes 1.1a2 (spambayes.org). No pre-learning, chi-squared combination, split-on-whitespace (mostly) tokenization. Automatically adjusts the cutoff rate based on the number of ham & spam trained. (Since train-on-error is used, this also effects which messages are trained).

tamS3pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS3pcd
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS3pci¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS3pci
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS3ped¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS3ped
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS3pei¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS3pei
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS4F¶

Results | Participants | Summary | Appendix

Run ID: tamS4F
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 7/12/2006
Type: online
Task: filter
Run description: SpamBayes 1.1a2 (spambayes.org). Train-on-everything, split-on-whitespace (mostly), chi-squared combination, semi-arbitrary ham & spam cutoff at 0.4. Also adds tokenization of attached images (which plain SpamBayes ignores).

tamS4pcd¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS4pcd
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS4pci¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS4pci
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS4ped¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS4ped
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tamS4pei¶

Results | Participants | Input | Summary | Appendix

Run ID: tamS4pei
Participant: masseyu.meyer
Track: Spam
Year: 2006
Submission: 8/6/2006
Type: online
Task: run

tufS1F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: tufS1F
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: Filter is TuftSpam filter. No prior training. Flags set to -N 4 and -T 100.

tufS1pcd¶

Run ID: tufS1pcd
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS1pci¶

Run ID: tufS1pci
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS1ped¶

Run ID: tufS1ped
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS1pei¶

Run ID: tufS1pei
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS2F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: tufS2F
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: Filter is TuftSpam filter. No prior training. Flags set to -N 5 and -T 100.

tufS2pcd¶

Run ID: tufS2pcd
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS2pci¶

Run ID: tufS2pci
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS2ped¶

Run ID: tufS2ped
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS2pei¶

Run ID: tufS2pei
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: online
Task: run

tufS3F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: tufS3F
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: Filter is TuftSpam filter. No prior training. Flags set to -N 6 and -T 100.

tufS4F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: tufS4F
Participant: tufts.sculley
Track: Spam
Year: 2006
Submission: 7/13/2006
Type: online
Task: filter
Run description: Filter is TuftSpam filter. No prior training. Flags set to -N 7 and -T 100.

WEIA4F¶

Results | Participants | Proceedings | Summary | Appendix

Run ID: WEIA4F
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 7/11/2006
Type: active
Task: filter
Run description: Welcome to KidultPRIS! Nowadays most of us are tired of the constant bombardment of their inboxes by unwanted email. It is high time for us to construct a robust spam filter which can detect spam efficiently without regarding our legal mails as spam. Fortunately, KidultPRIS is one of the excellent spam filters that meet our needs.KidultPRIS is a command line program which is exploited by the members in the lab of Pattern Recognition and Intelligent System, Beijing University of Posts and Telecommunications.

WEIA4pci¶

Run ID: WEIA4pci
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run

WEIA4pei¶

Run ID: WEIA4pei
Participant: beijingu-posts-tele.weiran
Track: Spam
Year: 2006
Submission: 8/22/2006
Type: active
Task: run