Runs - Blog 2010¶

BIT10bl1fd1¶

Run ID: BIT10bl1fd1
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 170c0bbb262076d01ccbe57fa37f0e02
Run description: BIT10bl1fd1 uses SVM with linear-kernel to classify blogs in our own baseline1, followed by re-ranking blogs by topic-facet mixture model.

BIT10bl1fd2¶

Run ID: BIT10bl1fd2
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 9854eb5aaf3b5c2faf323ed97b51a320
Run description: BIT10bl1fd2 uses SVM with linear-kernel to classify blogs in our own baseline1, followed by re-ranking blogs by topic-facet mixture model. The parameter settings are different to BIT10bl1fd1.

BIT10bl1fd3¶

Run ID: BIT10bl1fd3
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: a1bb531073744428a2b5aa080be55ef3
Run description: BIT10bl1fd3 uses SVM with linear-kernel to classify blogs in our own baseline1, followed by re-ranking blogs by topic-facet mixture model.

BIT10bl1fd4¶

Run ID: BIT10bl1fd4
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: e4853289602490cf8ca4faa7cde87a7a
Run description: BIT10bl1fd4 uses SVM with linear-kernel to classify blogs in our own baseline1, followed by re-ranking blogs by topic-facet mixture model.

BIT10bl2fd1¶

Run ID: BIT10bl2fd1
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: afea4c084777633733053f1bb25ac8a4
Run description: BIT10bl2fd1 uses SVM with linear-kernel to classify blogs in our own baseline2, followed by re-ranking blogs by topic-facet mixture model.

BIT10bl2fd2¶

Run ID: BIT10bl2fd2
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 292bf5a127d055fa9005f6fc66cabd82
Run description: BIT10bl2fd2 uses SVM with linear-kernel to classify blogs in our own baseline2, followed by re-ranking blogs by topic-facet mixture model. The parameter settings are different to BIT10bl2fd1.

BIT10bl2fd3¶

Run ID: BIT10bl2fd3
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 8699b2f8045c55c04a359f266d33f3ff
Run description: BIT10bl2fd3 uses SVM with linear-kernel to classify blogs in our own baseline2, followed by re-ranking blogs by topic-facet mixture model.

BIT10bl2fd4¶

Run ID: BIT10bl2fd4
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 62e7deaaaf00c299fee401d2f325fafa
Run description: BIT10bl2fd4 uses SVM with linear-kernel to classify blogs in our own baseline2, followed by re-ranking blogs by topic-facet mixture model.

BIT10std1fd1¶

Run ID: BIT10std1fd1
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 63297b8e2dffb1e19234563e30f73b63
Run description: BIT10std1fd1 uses SVM with linear-kernel to classify blogs in stdbaseline1, followed by re-ranking blogs by topic-facet mixture model.

BIT10std1fd2¶

Run ID: BIT10std1fd2
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: cf14eaf836ac000de02a20ef5404a46e
Run description: BIT10std1fd2 uses SVM with linear-kernel to classify blogs in stdbaseline1, followed by re-ranking blogs by topic-facet mixture model. The parameter settings are different to BIT10std1fd1.

BIT10std1fd3¶

Run ID: BIT10std1fd3
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 4632897ef21d8ad7bd7ee88f995b4ba6
Run description: BIT10std1fd3 uses SVM with linear-kernel to classify blogs in stdbaseline1, followed by re-ranking blogs by topic-facet mixture model.

BIT10std1fd4¶

Run ID: BIT10std1fd4
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 60835a05732179425682d16d46b44049
Run description: BIT10std1fd4 uses SVM with linear-kernel to classify blogs in stdbaseline1, followed by re-ranking blogs by topic-facet mixture model. The parameter settings are different to BIT10std1fd1, BIT10std1fd2 and BIT10std1fd3.

BIT10std2fd1¶

Run ID: BIT10std2fd1
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: e3b9446033f41f99ac3090dd628a770d
Run description: BIT10std2fd1 uses SVM with linear-kernel to classify blogs in stdbaseline2, followed by re-ranking blogs by topic-facet mixture model.

BIT10std3fd1¶

Run ID: BIT10std3fd1
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 5202b5101749856ea92c9a05db21dbd8
Run description: BIT10std3fd1 uses SVM with linear-kernel to classify blogs in stdbaseline3, followed by re-ranking blogs by topic-facet mixture model.

BITblog10bl1¶

Run ID: BITblog10bl1
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: 5fc94216ae9ea6b124337b1ffea1d5d4
Run description: BITblog10bl1 uses language model based approach to solve the blog relevance ranking task. The parameters are determined by the training results of old topics. The indexed Wikipedia corpus is used to obtain feedback documents automatically.

BITblog10bl2¶

Run ID: BITblog10bl2
Participant: BIT
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: f19fa29db1ef141c36488dd3270e8a19
Run description: BITblog10bl2 uses language model based approach to solve the blog relevance ranking task. The parameter settings are different to BITblog10bl1. The indexed Wikipedia corpus is used to obtain feedback documents automatically.

bloggerModel¶

Run ID: bloggerModel
Participant: ULugano
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: d056e61ed4e374af84d8d8b5467d590f
Run description: Using the bloggerModel with Kullback-Leibler for weighting terms

CombMNZ¶

Run ID: CombMNZ
Participant: ULugano
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: 2c86cd74ec8df7fe6b24fa375e61101c
Run description: We cluster the posts of each day and retrieve headlines for each cluster and then combine results by CombMNZ.

ComSumScores¶

Run ID: ComSumScores
Participant: ULugano
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: 406d95e314f3ad0dd67883ab16e87fff
Run description: we Clustered the posts for each date, and generated one query per cluster, run it against the TRC2 collection.

FEUPirlab1¶

Run ID: FEUPirlab1
Participant: feup
Track: Blog
Year: 2010
Submission: 8/3/2010
Type: automatic
Task: blfeed
MD5: 5ac5a78c2264471c3a14b5e669165adc
Run description: In this run, we retrieve documents based on the following score: BM25score + k * log(DOCindegree), where k=4 (optimized for P@10). Then we aggregate documents' scores into a feed score by summing all scores and dividing by the total number of documents in the feed. We use all 50 topics from last year (TREC 2009) and all 50 topics from this year (TREC 2010).

FEUPirlab2¶

Run ID: FEUPirlab2
Participant: feup
Track: Blog
Year: 2010
Submission: 8/3/2010
Type: automatic
Task: blfeed
MD5: feda3f5685e361fe77c422af3acfcada
Run description: In this run, we retrieve documents based on the following score: BM25score + k * log(DOChindex), where k=4 (optimized for R-prec). Then we aggregate documents' scores into a feed score by summing all scores and dividing by the total number of documents in the feed. We use all 50 topics from last year (TREC 2009) and all 50 topics from this year (TREC 2010).

hitFeeds1¶

Run ID: hitFeeds1
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 1b16ac6a739535cf5ee3e1636a0f3a3e
Run description: hitFeed_s1. This run is based on the stdbaseline1 baseline. We use our own Maximum Entropy Model toolkit to predict the facet inclination of every post in a feed, and then calculate the feed's facet inclination by the same method as it used in baseline runs. Finally, we combine the feed's relevant score to calculate the feed's faceted score, and reach the ranking list of the faceted blog distillation. This run uses the SentiWordNet(the latest version) and a regular expression lib(Boost.Regex). We also use our own Maximum Entropy Model toolkit

hitFeeds2¶

Run ID: hitFeeds2
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 3906cceadd4bcf662d9543f1490b1c3e
Run description: hitFeed_s2. This run is based on the stdbaseline2 baseline. We use our own Maximum Entropy Model toolkit to predict the facet inlination of every post in a feed, and then calculate the feed's facet inlination by the same method as it used in baseline runs. Finally, we combine the feed's relevant score to calculate the feed's faceted score, and reach the ranking list of the faceted blog distillation. This run uses the SentiWordNet(the latest version) and a regular expression lib(Boost.Regex). We also use our own Maximum Entropy Model toolkit

hitFeeds3¶

Run ID: hitFeeds3
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 4711da05e643774de8771af5d02e90b2
Run description: hitFeed_s3. This run is based on the stdbaseline3 baseline. We use our own Maximum Entropy Model toolkit to predict the facet inlination of every post in a feed, and then calculate the feed's facet inlination by the same method as it used in baseline runs. Finally, we combine the feed's relevant score to calculate the feed's faceted score, and reach the ranking list of the faceted blog distillation. This run uses the SentiWordNet(the latest version) and a regular expression lib(Boost.Regex). We also use our own Maximum Entropy Model toolkit

hitQFeedbl¶

Run ID: hitQFeedbl
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 400bbc0d3e5fae5e57d98d6dd68008a6
Run description: hitQFeedbl. This run is based on the hitQuerybl baseline. We use our own Maximum Entropy Model toolkit to predict the facet inlination of every post in a feed, and then calculate the feed's facet inlination by the same method as it used in baseline runs. Finally, we combine the feed's relevant score to calculate the feed's faceted score, and reach the ranking list of the faceted blog distillation.

hitQFeedR¶

Run ID: hitQFeedR
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: c2d694740b0e8578d997a6c499a805c8
Run description: hitQFeedR. This run has no corresponding baseline(N/A), it is based on the query-relevant result which is also the base of the hitQuerybl baseline. I fill the BASELINE RUN field with "hitQuerybl" only because of the submission error happend when I used "N/A" for the BASELINE RUN field. We use our own Maximum Entropy Model toolkit to predict the facet inlination of every post in a feed, and then calculate the feed's facet inlination by the same method as it used in baseline runs. Finally, we combine the feed's relevant score to calculate the feed's faceted score, and reach the ranking list of the faceted blog distillation. This run uses the SentiWordNet(the latest version) and a regular expression lib(Boost.Regex). We also use our own Maximum Entropy Model toolkit

hitQuerybl¶

Run ID: hitQuerybl
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: 8c0ad8224b4c92dbf808f1f139beb1c1
Run description: The query is construted automatically only using query field of the topic. We think that a title of blog may supply important topic-relevant information. So the query contains two part: one is query field of the topic, with stop words removed, and the other is title context query, built by the first part. The two parts are combined with a specified weight. The feed relevant score of this run is calculated according to integrating the average score of all posts within a feed and the average score of the most relevant N posts of the feed . The run uses the Indri(the latest version) as a search platform and a stop-word file containing 418 words. The stop-word file is included in the Indri project

hitTDNbl¶

Run ID: hitTDNbl
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: manual
Task: blfeed
MD5: 11122ba77b199440ef47837289470296
Run description: The query is construted manually by using query field,description field and narrative field of the topic. We also combine title context query within the manual query with a specified weight. The feed relevant score of this run is calculated according to integrating the average score of all posts within a feed and the average score of the most relevant N posts of the feed . The run uses the Indri(latest version) as a search platform and a stop-word file containing 418 words. The stop-word file is included in the Indri project

hitTDNfeedbl¶

Run ID: hitTDNfeedbl
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: manual
Task: feed
MD5: 9ae76ca1e683a08b410b9344fd1b2e77
Run description: hitTDNfeedbl. This run is based on the hitTDNblbaseline. We use our own Maximum Entropy Model toolkit to predict the facet inlination of every post in a feed, and then calculate the feed's facet inlination by the same method as it used in baseline runs. Finally, we combine the feed's relevant score to calculate the feed's faceted score, and reach the ranking list of the faceted blog distillation. This run uses the SentiWordNet(the latest version) and a regular expression lib(Boost.Regex). We also use our own Maximum Entropy Model toolkit

hitTDNfeedR¶

Run ID: hitTDNfeedR
Participant: HIT_LTRC
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: manual
Task: feed
MD5: c76949d82802bcd2461524e83a0ca132
Run description: hitTDNfeedR. This run has no corresponding baseline(N/A), it is based on the query-relevant result which is also the base of the hitTDNbl baseline. I fill the BASELINE RUN field with "hitTDNbl" only because of the submission error happend when I used "N/A" for the BASELINE RUN field. We use our own Maximum Entropy Model toolkit to predict the facet inclination of every post in a feed, and then calculate the feed's facet inclination by the same method as it used in baseline runs. Finally, we combine the feed's relevant score to calculate the feed's faceted score, and reach the ranking list of the faceted blog distillation. This run uses the SentiWordNet(the latest version) and a regular expression lib(Boost.Regex). We also use our own Maximum Entropy Model toolkit

ICTNETBD1¶

Run ID: ICTNETBD1
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: fefcb67a78df01084f7a0543e58e238a
Run description: It's the baseline "ICTNETBDRun1" itself.

ICTNETBD2¶

Run ID: ICTNETBD2
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: a873f7912eab28ca9c5851a62b117211
Run description: Its just the baseline ICTNETBDRun2 itself without considering any faceted feature

ICTNETBD3¶

Run ID: ICTNETBD3
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: fd2cf349e731a828e24b0237b8b6208a
Run description: Its an unsubmitted baseline without considering any faceted feature (we denote it as "ICTNETBD3"). This baseline adopts the idea of ensemble ranking and combines various rankings by different models to the final run. please note that when we selected "N/A" as baseline, the submitting system just reported error, so we selected "ICTNETBDRun2" temporarily here.

ICTNETBD4¶

Run ID: ICTNETBD4
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 8e852067effa9b7aa1b7a8b3810b7372
Run description: It's a baseline run without considering any facet feature! It's base on topic language model which is learnt using Google blog search engine without considering any facet feature. It's for the comparison with ICTNETFBD4 to see how a facet inclination language mode affect the performance of faceted blog distillation task. Note that we mean to select "N/A" as baseline, however the system just reported errors. So, we selected "ICTNETBDRun1" as baseline temporarily

ICTNETBDRun1¶

Run ID: ICTNETBDRun1
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: 3877a99e24545235f05f09eaced0ca84
Run description: We combined many rankings by different models to the final run. We used Wiki for query expansion.

ICTNETBDRun2¶

Run ID: ICTNETBDRun2
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: e0d7c26011b57de5fdac2cdfe9f924a9
Run description: We combined rankings generated by different models to a final ranking We used Wiki for query expansion

ICTNETFBD1¶

Run ID: ICTNETFBD1
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 83e50b2efbc0cb31cebccbb5d838cdd0
Run description: We use Google blog search engine and annotated data by Know-center to learn language model for each facet inclination, and then the facet inclination language model was used to improve baseline ICTNETBDRun1.

ICTNETFBD2¶

Run ID: ICTNETFBD2
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 842998ae6e573516820b03e6430f9b7e
Run description: We use Google blog search engine and annotated data by Know-center to learn language model for each facet inclination.

ICTNETFBD3¶

Run ID: ICTNETFBD3
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 9c2e484ce82af1752c8102fadd72d927
Run description: Its a run based on an unsubmitted baseline "ICTNETBD3". When I selected "N/A" as baseline, it just reported errors. So I selected the submitted baseline "ICTNETBDRun2" temporarily here. We use Google blog search engine and annotated data by Know-center to learn language model for each facet inclination.

ICTNETFBD4¶

Run ID: ICTNETFBD4
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: c999ee94504b07e3e62ce1e36f63799b
Run description: We use Google blog search engine and annotated data by Know-center to learn language model for each facet inclination.the ranking of feed is simply based on negative KL-divergence between feed language model and facet inclination language model! Note that we actually wanted to select "N/A" as baseline, however when we selected "N/A" as baseline, the system just reported errors. So we selected "ICTNETBDRun1" temporarily.

ICTNETFBDs2¶

Run ID: ICTNETFBDs2
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 0a5cb0de038991b025d7265c5fa80add
Run description: It's a faceted run based on "stdbaseline2".

ICTNETFBDs3¶

Run ID: ICTNETFBDs3
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: ae46572aeeee7c73c2193e2d2784072e
Run description: based on stdbaseline3

ICTNETPRRun1¶

Results | Participants | Proceedings | Input | Summary

Run ID: ICTNETPRRun1
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 10/1/2010
Type: automatic
Task: newsblogpost
MD5: 5cd4d6662b3e28446554e9e75f5827fc
Run description: we adopted a "ensemble ranking" strategy

ICTNETPRRun2¶

Results | Participants | Proceedings | Input | Summary

Run ID: ICTNETPRRun2
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 10/1/2010
Type: automatic
Task: newsblogpost
MD5: 4645d5e3a023515fa4bbd06f50c67697
Run description: we adopted a "ensemble ranking" strategy

ICTNETPRRun3¶

Results | Participants | Proceedings | Input | Summary

Run ID: ICTNETPRRun3
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 10/4/2010
Type: automatic
Task: newsblogpost
MD5: 9a1815c12bcc3e309ed1057ca7016708
Run description: "ensemble ranking" with diversity features

ICTNETTSRun1¶

Run ID: ICTNETTSRun1
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 9/2/2010
Type: automatic
Task: topstories
MD5: 56899c12f894fb2c63dc1624042d4b7d
Run description: A simple approach training data crawled from Reuters webiste to learn a classifier to categorize news stories into 5 categories.

ICTNETTSRun2¶

Run ID: ICTNETTSRun2
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: 5ed380fbe3e7bb31bd178147eac64fb7
Run description: A simple approach training data crawled from Reuters webiste to learn a classifier to categorize news stories into 5 categories.

ICTNETTSRun3¶

Run ID: ICTNETTSRun3
Participant: ICTNET
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: 0ccb09dc89951ec54887f2d1b4dba0b9
Run description: A simple approach training data crawled from Reuters webiste to learn a classifier to categorize news stories into 5 categories.

ikm100bindog¶

Run ID: ikm100bindog
Participant: ikm100
Track: Blog
Year: 2010
Submission: 9/4/2010
Type: automatic
Task: topstories
MD5: 404294286de2b79fadceaed127d4a16f
Run description: We use the results of last year and NYTimes corpus as training set

ikm100jing¶

Run ID: ikm100jing
Participant: ikm100
Track: Blog
Year: 2010
Submission: 9/4/2010
Type: automatic
Task: topstories
MD5: 9247b31b23c9820ed807cf68ee3fb43b
Run description: We use the results of last year and NYTimes corpus as training set

ikm100ufan¶

Run ID: ikm100ufan
Participant: ikm100
Track: Blog
Year: 2010
Submission: 9/4/2010
Type: automatic
Task: topstories
MD5: 148515b0e17d853797898e02a83bc7bf
Run description: We use the results of last year and NYTimes corpus as training set

KLE1¶

Results | Participants | Proceedings | Input | Summary

Run ID: KLE1
Participant: POSTECH_KLE
Track: Blog
Year: 2010
Submission: 10/4/2010
Type: automatic
Task: newsblogpost
MD5: ef68e031d9f88eff9e0f7e64ed5d33cb
Run description: relevance and similarity

KLE2¶

Results | Participants | Proceedings | Input | Summary

Run ID: KLE2
Participant: POSTECH_KLE
Track: Blog
Year: 2010
Submission: 10/4/2010
Type: automatic
Task: newsblogpost
MD5: 9b9fb4515ca112ec211d9fa029554983
Run description: relevance and similarity

KLERUN1¶

Run ID: KLERUN1
Participant: POSTECH_KLE
Track: Blog
Year: 2010
Submission: 9/5/2010
Type: automatic
Task: topstories
MD5: 36d91dd2f2b9a1b401f3f45dd04a5d0e
Run description: KL-divergence between blog language model and news language model Using NYTimes news articles for news stories classification

KLERUN2¶

Run ID: KLERUN2
Participant: POSTECH_KLE
Track: Blog
Year: 2010
Submission: 9/5/2010
Type: automatic
Task: topstories
MD5: 10b067df8531331b574ead320d4da0b0
Run description: KL-divergence between blog language model and news language model + event extraction & ranking Using NYTimes news articles for news stories classification

KLERUN3¶

Run ID: KLERUN3
Participant: POSTECH_KLE
Track: Blog
Year: 2010
Submission: 9/5/2010
Type: automatic
Task: topstories
MD5: 8f95f2ee2dfbfb650cb201a8a278b86c
Run description: KL-divergence between blog language model and news language model + event extraction & ranking + temporal profile + term importance Using NYTimes news articles for news stories classification

LexMIRuns1¶

Run ID: LexMIRuns1
Participant: ULugano
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: cfbd41aab1f5d4305cb6f11de9d9452e
Run description: This run uses a general lexicon for opinion facet and an opinion weighting method for personal facet based on a dictionary which is built from BLOG06 dataset using Mutual Information for weighting terms. For Indepth facet it uses the cross entropy.

LexMIRuns2¶

Run ID: LexMIRuns2
Participant: ULugano
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 6d0b8f5961965b676ec44311fa561e56
Run description: This run uses a general lexicon for opinion facet and an opinion weighting method for personal facet based on a dictionary which is built from BLOG06 dataset using Mutual Information for weighting terms. For Indepth facet it uses the cross entropy.

LexMIRuns3¶

Run ID: LexMIRuns3
Participant: ULugano
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 853e5c662743cad67dc52ee9382b7432
Run description: his run uses a general lexicon for opinion facet and an opinion weighting method for personal facet based on a dictionary which is built from BLOG06 dataset using Mutual Information for weighting terms. For Indepth facet it uses the cross entropy.

LMPI¶

Run ID: LMPI
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 062ad01e5c0fa4df21ad9e790ab26149
Run description: This submission is based on the baseline of PULM, using the Indri language model to calculate the relevance score. The faceted scored by some features, and each feature was treated equally. Finally, rerank the product of the relevance score and faceted score.

LMPII¶

Run ID: LMPII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 548170f2061f13e29c360b2e73916e68
Run description: This submission is based on the baseline of PULM, using the Indri language model to calculate the relevance score. The faceted scored by some features, and features were given different weights. Finally, rerank the product of the relevance score and faceted score.

LMPIII¶

Run ID: LMPIII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: c5d0ad528d0f1acf87c0ed339fde02d8
Run description: This submission is based on the baseline of PULM, using the Indri language model to calculate the relevance score. The faceted scored by some features, and experimental rules. Finally, rerank the product of the relevance score and faceted score.

MILexRuns1¶

Run ID: MILexRuns1
Participant: ULugano
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 6351ee90efb5c21fd88b77eb8fa1e29d
Run description: his run uses an opinion weighting method for opinion facet based on a dictionary which is built from BLOG06 dataset using Mutual Information for weighting terms and a general lexicon for personal facet. For Indepth facet it uses the cross entropy.

MILexRuns2¶

Run ID: MILexRuns2
Participant: ULugano
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 582b9693cb90200ad511d7df1a812761
Run description: his run uses an opinion weighting method for opinion facet based on a dictionary which is built from BLOG06 dataset using Mutual Information for weighting terms and a general lexicon for personal facet. For Indepth facet it uses the cross entropy.

MILexRuns3¶

Run ID: MILexRuns3
Participant: ULugano
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 4ebaa835134e18c4402239537e7cafdc
Run description: his run uses an opinion weighting method for opinion facet based on a dictionary which is built from BLOG06 dataset using Mutual Information for weighting terms and a general lexicon for personal facet. For Indepth facet it uses the cross entropy.

OkapiPI¶

Run ID: OkapiPI
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: b4f65e5c8e423dda435e9eb7a44fc1e9
Run description: This submission is based on the baseline of PUBM, using Okapi BM25 to calculate the relevance score. The faceted scored by some features, and each feature was treated equally.

OkapiPII¶

Run ID: OkapiPII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 7f0740a00d4fd0e3262f241301c4de03
Run description: This submission is based on the baseline of PUBM, using Okapi BM25 to calculate the relevance score. The faceted scored by some features, and features were given different priorities.

OkapiPIII¶

Run ID: OkapiPIII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 3a689a1366e69833b7f9b58e1acc46fc
Run description: This submission is based on the baseline of PUBM, using Okapi BM25 to calculate the relevance score. The faceted scored by some features, and experimental rules.

OWAScores¶

Run ID: OWAScores
Participant: ULugano
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: 1d0a39c2005d98c7cb692a5dd5add47c
Run description: We cluster the posts of each day and retrieve headlines for each cluster and then combine results by OWA operators.

PKUTM111onB1¶

Run ID: PKUTM111onB1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 4a6a5632c504045edd6bc840611e6f65
Run description: PKUTM facet run1 on baseline of PKUTMB1

PKUTM111onB2¶

Run ID: PKUTM111onB2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 5162689a43f36e3b839f03a37589a003
Run description: PKUTM facet run1 on baseline of PKUTMB2

PKUTM111STD1¶

Run ID: PKUTM111STD1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 24cb05feaee6ceadcf2d149d8a26ae59
Run description: PKUTM facet run1 on baseline of stdbaseline1

PKUTM111STD2¶

Run ID: PKUTM111STD2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 5dabbbe556fd7413188f7c2f1aa6b5a0
Run description: PKUTM facet run1 on baseline of stdbaseline1

PKUTM111STD3¶

Run ID: PKUTM111STD3
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 2c086906bb6af2eba352a61d834ec624
Run description: PKUTM facet run1 on baseline of stdbaseline3

PKUTM121onB1¶

Run ID: PKUTM121onB1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 382eac3d75de7d2ee0d56c59d85aa671
Run description: PKUTM facet run2 on baseline of PKUTMB1

PKUTM121STD2¶

Run ID: PKUTM121STD2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 299425a33ff5fa2e909496dc5d1e00d0
Run description: PKUTM facet run2 on baseline of stdbaseline1

PKUTM121STD3¶

Run ID: PKUTM121STD3
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 6d796eda3c8fbd5819ef1c6c56e38ac9
Run description: PKUTM facet run2 on baseline of stdbaseline3

PKUTM123STD1¶

Run ID: PKUTM123STD1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 1d6431ad77528383f5fc972de025e18a
Run description: PKUTM facet run2 on baseline of stdbaseline1

PKUTM211onB1¶

Run ID: PKUTM211onB1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 80c9db6e6da51f19a04e4a517ff3d243
Run description: PKUTM facet run3 on baseline of PKUTMB1

PKUTM211onB2¶

Run ID: PKUTM211onB2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: b3371cdf003778c5eb4e6db98cb9beaf
Run description: PKUTM facet run2 on baseline of PKUTMB2

PKUTM211STD1¶

Run ID: PKUTM211STD1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 57be00a2ff75d0a30f9a342eae5e43a3
Run description: PKUTM facet run3 on baseline of stdbaseline1

PKUTM211STD2¶

Run ID: PKUTM211STD2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 8227fa13ea92ff300eb2a1fbfb4238f9
Run description: PKUTM facet run4 on baseline of stdbaseline1

PKUTM211STD3¶

Run ID: PKUTM211STD3
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: b21cd29afcf456064ac1dcf155023d9c
Run description: PKUTM facet run3 on baseline of stdbaseline3

PKUTM221onB2¶

Run ID: PKUTM221onB2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 8bff90db971787e8eeb1de16b0f6ba8e
Run description: PKUTM facet run3 on baseline of PKUTMB2

PKUTM221STD2¶

Run ID: PKUTM221STD2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: ab03f13b016504ee48cb60d4cf96b8de
Run description: PKUTM facet run3 on baseline of stdbaseline1

PKUTM221STD3¶

Run ID: PKUTM221STD3
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 70c77d60c1f3f665d400fc4074737d05
Run description: PKUTM facet run4 on baseline of stdbaseline3

PKUTM222STD1¶

Run ID: PKUTM222STD1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 45a1dfe30b9e75d29811425381e22ac1
Run description: PKUTM facet run4 on baseline of stdbaseline1

PKUTM321onB1¶

Run ID: PKUTM321onB1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: cf70ed12467ca7968c0abafd1f4738d2
Run description: PKUTM facet run4 on baseline of PKUTMB1

PKUTM321onB2¶

Run ID: PKUTM321onB2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: d1ab623f63e06d0eb9cf6b9e2714ff13
Run description: PKUTM facet run3 on baseline of PKUTMB2

PKUTMB1¶

Run ID: PKUTMB1
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: c7b4983b18babb0a0b96fc97263303d7
Run description: This is our "query-only" baseline run.

PKUTMB2¶

Run ID: PKUTMB2
Participant: PKUTM
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: 9084d323686928f31ef55589bfaddc01
Run description: This is our another baseline which use "query" ,"description" and "narrative" of the topics.

pris¶

Run ID: pris
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/6/2010
Type: automatic
Task: blfeed
MD5: 0225e8dc8ab2ff1dc93b0ec9b4df38a5
Run description: Indri is used to retrieve single post first. Posts Average algorithm is used to determine the ranking of feeds.

prisb¶

Run ID: prisb
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/4/2010
Type: automatic
Task: blfeed
MD5: 78037fa84962dc8d331278bc2a1e6bac
Run description: Posts Average algorithm is used for determining the ranking of feeds based on Indri. A Learning Query Expansion algorithm is designed to improve the pricision of topic-relevance retrieval.

PrisQ01¶

Run ID: PrisQ01
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: d5c421152c948fd5b182866920bf4a61
Run description: This version is based on our own query-only baseline whose runtag is "pris". We use a sentimental lexicon as the external resource.

PrisQ02¶

Run ID: PrisQ02
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 767ab5a19ba1ebfe7305ccb2fdf268e1
Run description: This version is based on our own query-only baseline whose runtag is "pris". We use a sentimental lexicon as the external resource.

PrisQ03¶

Run ID: PrisQ03
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: b0f8d24028e8fdec56c72d6c63c503b9
Run description: This version is based on our own query-only baseline whose runtag is "pris". We use a sentimental lexicon as the external resource.

PrisQ04¶

Run ID: PrisQ04
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 67f6cd6a387f6c4545a8fa6251afcdc0
Run description: This version is based on our own query-only baseline whose runtag is "pris". We use a sentimental lexicon as the external resource.

PrisQE1¶

Run ID: PrisQE1
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 437c7c6543725c9d40f53921be724006
Run description: This version is based on our own query-expansion baseline whose runtag is "prisb". We use a sentimental lexicon as the external resource.

PrisQE2¶

Run ID: PrisQE2
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 52a10d70189c31a1eb44c75a63eef735
Run description: This version is based on our own query-expansion baseline whose runtag is "prisb". We use a sentimental lexicon as the external resource.

PrisStdQ0¶

Run ID: PrisStdQ0
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 16cd00478ca52e1b91a413b8e078de38
Run description: This version is based on the standard baseline 1, and is our group's best results of this sub-task. We use a sentimental lexicon as the external resource.

PrisStdQ02¶

Run ID: PrisStdQ02
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 7a7771669f06f7a734b189a2bece7242
Run description: This version is based on the standard baseline 1. For the distillation of the "opinionated vs fatual" facet, we use the Informaion Gain to generate a "factual" lexicon.

PrisStdQ03¶

Run ID: PrisStdQ03
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: fef65fae9f0114e434a91ac56ba071cd
Run description: This version is based on the standard baseline 1. We use a sentimental lexicon as the external resource.

PrisStdQ04¶

Run ID: PrisStdQ04
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 278410f60c7e9fb1bb7d1c01ceb4a368
Run description: This version is based on the standard baseline 1. We use a sentimental lexicon as the external resource.

PrisStdQE1¶

Run ID: PrisStdQE1
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 58c495f03cc1ae32b40ec92f4c83db9b
Run description: This version is based on the standard baseline 1, and we run query expansion further. We use a sentimental lexicon as the external resource.

PrisStdQE2¶

Run ID: PrisStdQE2
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 2264aadf29de4083c682154b25314dc5
Run description: This version is based on the standard baseline 1, and we run query expansion further. We use a sentimental lexicon as the external resource.

PrisStdQE3¶

Run ID: PrisStdQE3
Participant: PRIS
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 2b5119e9e73e53e4526cf456e014f882
Run description: This version is based on the standard baseline 1, and we run query expansion further. We use a sentimental lexicon as the external resource.

PUBM¶

Run ID: PUBM
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: 8335ab6ba90fb7b6a2f2d8a0efa9302c
Run description: First, indri established the index across the blog08 collection corpus. And, only use the query section of the topic field; then, query the relevance permalinks using Okapi BM25. Finally, find out all the relevance feeds.

PULM¶

Run ID: PULM
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/9/2010
Type: automatic
Task: blfeed
MD5: 2175a78e81c0fd7c834184e960f56924
Run description: First, indri established the index across the blog08 collection corpus. And, only use the query section of the topic field; then, query the relevance permalinks using indri language model. Finally, find out all the relevance feeds.

rmit2step¶

Run ID: rmit2step
Participant: RMIT
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: af2c38caf7c943196d245e19cf81141d
Run description: A simple approach simulating the human judgment process, involving two steps. Step 1: Use the highest post score in a feed as feed score to generate a ranking, keeping top 100 feeds for each topic Step 2: Re-rank the feeds with the sum of post scores in each feed to generate the final ranking

rmitfaceted¶

Run ID: rmitfaceted
Participant: RMIT
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 4d3bc4b77231fc4b33087109fc79f354
Run description: Used cross entropy for the indepth/shallow facets, and machine learning methods for the opinionated/factual facets. Assumption has been made that factual blogs are official, and opinionated blogs are personal.

rmitprob¶

Run ID: rmitprob
Participant: RMIT
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: 1f3e6d26df6eb173db300f48368a4a28
Run description: A probablistic approach. Post scores are first scaled to probabilities, then the probability of a feed being relevant to a topic can then be calculated.

Run1¶

Run ID: Run1
Participant: UniNE
Track: Blog
Year: 2010
Submission: 8/13/2010
Type: automatic
Task: blfeed
MD5: 979e16d7cf4c2ee472cba9f67ca0ba61
Run description: We will use our Lucene-based search engine. This search engine is based on a data fusion operator (Z-score) of various IR models (Okapi BM25, Divergence from Randomness, Language Model). We then apply a language model to detect the polarity of the retrieved sentences and classify them according to their content (word-based).

run1¶

Results | Participants | Proceedings | Input | Summary

Run ID: run1
Participant: ikm100
Track: Blog
Year: 2010
Submission: 10/5/2010
Type: automatic
Task: newsblogpost
MD5: 63a66525d82eaca5768f8f37a7dd9f47
Run description: We combined news-post cosine similarity and post's comment count to calculate score and selected post which category is same as TRC2-News' category.

run2¶

Results | Participants | Proceedings | Input | Summary

Run ID: run2
Participant: ikm100
Track: Blog
Year: 2010
Submission: 10/5/2010
Type: automatic
Task: newsblogpost
MD5: ba5d6776d6778ebb7e3dc36e6695a89a
Run description: We combined news-post cosine similarity and post's comment count to calculate score and selected post which category is same as TRC2-News' category.

run3¶

Results | Participants | Proceedings | Input | Summary

Run ID: run3
Participant: ikm100
Track: Blog
Year: 2010
Submission: 10/5/2010
Type: automatic
Task: newsblogpost
MD5: 182f22729df91f7bf35828424848c0cb
Run description: We combined news-post cosine similarity and post's comment count to calculate score and selected post which category is same as TRC2-News' category.

Run3¶

Run ID: Run3
Participant: UniNE
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 8e5d8138749b073e6488d4a464b96b99
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on Z Score previously trained on the opinionated corpora and natural language vocabulary lists.

run3swnpn10¶

Run ID: run3swnpn10
Participant: UniNE
Track: Blog
Year: 2010
Submission: 9/16/2010
Type: automatic
Task: feed
MD5: 1ac7a52c53211a57edccb465686711dc
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on topic relevance statistics and natural language vocabulary lists (SentiWordNet).

run3swnpn20¶

Run ID: run3swnpn20
Participant: UniNE
Track: Blog
Year: 2010
Submission: 9/16/2010
Type: automatic
Task: feed
MD5: ea0b0d0504186467db32a4f4bb36a20d
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on topic relevance statistics and natural language vocabulary lists (SentiWordNet).

run3swnpn30¶

Run ID: run3swnpn30
Participant: UniNE
Track: Blog
Year: 2010
Submission: 9/16/2010
Type: automatic
Task: feed
MD5: 8cf9763caa0c49f6dee1883f6ce3df7b
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on topic relevance statistics and natural language vocabulary lists (SentiWordNet).

run3zscore1¶

Run ID: run3zscore1
Participant: UniNE
Track: Blog
Year: 2010
Submission: 9/16/2010
Type: automatic
Task: feed
MD5: e3c2dbd7592bd9d55a97a7cbf92fd49b
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on Z Score model previously trained on the opinionated and not opinionated corpora.

run3zscore2¶

Run ID: run3zscore2
Participant: UniNE
Track: Blog
Year: 2010
Submission: 9/16/2010
Type: automatic
Task: feed
MD5: d736eea6c12bdf78cc143b0d131e4f3e
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on Z Score model previously trained on the opinionated and not opinionated corpora.

run3zscore3¶

Run ID: run3zscore3
Participant: UniNE
Track: Blog
Year: 2010
Submission: 9/16/2010
Type: automatic
Task: feed
MD5: 509fc76ac3d20f3843f9636c7de2e563
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on Z Score model previously trained on the opinionated and not opinionated corpora.

Run4¶

Run ID: Run4
Participant: UniNE
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 291afcf3933c6ba8213de85b42789635
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on Z Score previously trained on the opinionated corpora and natural language vocabulary lists.

Run5¶

Run ID: Run5
Participant: UniNE
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 8118ba9d9f37f01ea47c40a42fc70436
Run description: We use the topic relevance analysis and retrieval statistics on the first step. In the second step, in order to get improvement classification we use statistical methods based on Z Score previously trained on the opinionated corpora and natural language vocabulary lists.

stanford1¶

Run ID: stanford1
Participant: StanfordNLP
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: 54c16150f74f6a48cf2c1cac2f6eca3c
Run description: Passage Ranking Over Simple Single Indri Retrieval Baseline

stanford2¶

Run ID: stanford2
Participant: StanfordNLP
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: b52e30d39274469d9656a6ffe1b9d7fc
Run description: Passage Ranking Over Three Simple Single Indri Retrieval Baselines

Std1stPI¶

Run ID: Std1stPI
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: c856315f50bfa3c0c6a9aa2d13ffd5dd
Run description: This submission is based on the baseline of stdbaseline1. The faceted scored by some features, and each feature was treated equally. Finally, rerank the product of the relevance score and faceted score.

Std1stPII¶

Run ID: Std1stPII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 53b9edf8226a92f4cc88752301a8abf7
Run description: This submission is based on the baseline of stdbaseline1. The faceted scored by some features, and features were given different weights. Finally, rerank the product of the relevance score and faceted score.

Std1stPIII¶

Run ID: Std1stPIII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: fdec2b980b11ce2ffd6ac9b9c4f49f3e
Run description: This submission is based on the baseline of stdbaseline1. The faceted scored by some features, and experimental rules. Finally, rerank the product of the relevance score and faceted score.

Std2ndPI¶

Run ID: Std2ndPI
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: d91a6dc5735edda6c1575607eaf33047
Run description: This submission is based on the baseline of stdbaseline2. The faceted scored by some features, and each feature was treated equally. Finally, rerank the product of the relevance score and faceted score.

Std2ndPII¶

Run ID: Std2ndPII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: 2d2af775fe1c6e643d4193804ed09df7
Run description: This submission is based on the baseline of stdbaseline2. The faceted scored by some features, and features were given different weights. Finally, rerank the product of the relevance score and faceted score.

Std2ndPIII¶

Run ID: Std2ndPIII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: f95b23e21d7057bee4ae5fd2a9d3d25a
Run description: This submission is based on the baseline of stdbaseline2. The faceted scored by some features, and experimental rules. Finally, rerank the product of the relevance score and faceted score.

Std3rdPI¶

Run ID: Std3rdPI
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/28/2010
Type: automatic
Task: feed
MD5: f6302dfae1e35d565ee771f3a59f1e16
Run description: This submission is based on the baseline of stdbaseline3. The faceted scored by some features, and each feature was treated equally. Finally, rerank the product of the relevance score and faceted score.

Std3rdPII¶

Run ID: Std3rdPII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: 664d493c6d57349c0591de2815e09cfd
Run description: This submission is based on the baseline of stdbaseline3. The faceted scored by some features, and features were given different weights. Finally, rerank the product of the relevance score and faceted score.

Std3rdPIII¶

Run ID: Std3rdPIII
Participant: PCUHK
Track: Blog
Year: 2010
Submission: 8/29/2010
Type: automatic
Task: feed
MD5: b24dd244143712f0ef9fbd930246641a
Run description: This submission is based on the baseline of stdbaseline3. The faceted scored by some features, and experimental rules. Finally, rerank the product of the relevance score and faceted score.

strath1¶

Run ID: strath1
Participant: UoS
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: a660c93034f85e7394b35239c4ef7bda
Run description: The words frequencies of use of which in the blog corpus increased substantially on the day of the query were used as queries for the headlines corpus.

strath2¶

Run ID: strath2
Participant: UoS
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: manual
Task: topstories
MD5: 7f4ab4d9f731cb1fca9d7fd484486d82
Run description: Event descriptions were taken from the Current Event article of Wikipedia Portal on the day of the query (e.g. 22 January 2008 for the first topic). The description of each event was sent to Bing search engine. The words from the returned snippets more frequently occurring than on the Web in average were used to query the headlines corpus on the given day.

strath3¶

Run ID: strath3
Participant: UoS
Track: Blog
Year: 2010
Submission: 9/7/2010
Type: automatic
Task: topstories
MD5: a94dbb09faaac2c8570fa7bed75b6c09
Run description: The words frequencies of use of which in the blog corpus increased substantially on the day of the query were used as queries for the headlines corpus. The weights were applied to the query words based on the amount of increase.

uicfbdrun2¶

Run ID: uicfbdrun2
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 70a0f35fd5f006f7b14c215a96eb087c
Run description: This facet blog distillation run take our own baseline as input, and for each topic we rerank the baseline according to the extent of two interested facet values respectively.

uicfbdstd1a¶

Run ID: uicfbdstd1a
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 2c51eeb1915a7c6a0504c0227aa631b0
Run description: This facet blog distillation run take stdbaseline1 as input. Due to the lack of document-level IR results, we managed to create a document-level ranking for each topic according to the scores assigned by our own retrieval system. Then, for each topic, we re-rank stdbaseline1 based on the artificial-created document-level ranking with respect to each interested facet of that topic.

uicfbdstd1b¶

Run ID: uicfbdstd1b
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: bcb464ff1a3c06220f10b6b8ef0c4408
Run description: This facet blog distillation run take stdbaseline1 as input. Due to the lack of document-level IR results, with respect to each interested facet of each topic, we re-rank stdbaseline1 according to the ranking of our own facet blog distillation and put those feeds which appear in stdbaseline1 but absent from our own baseline uicfeedir2 at the bottom of the ranking.

uicfbdstd1c¶

Run ID: uicfbdstd1c
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: db8113fd9fea0781c5081bfd552abe5a
Run description: This facet blog distillation run take stdbaseline1 as input. Due to the lack of document-level IR results, with respect to each interested facet of each topic, we re-rank stdbaseline1 according to the feed ranking which is calculated on the basis of our own feed baseline, uicfeedir2.

uicfbdstd2a¶

Run ID: uicfbdstd2a
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 7592bbc720a2353718664624ab516425
Run description: This facet blog distillation run take stdbaseline2 as input. Due to the lack of document-level IR results, we managed to create a document-level ranking for each topic according to the scores assigned by our own retrieval system. Then, for each topic, we re-rank stdbaseline2 based on the artificial-created document-level ranking with respect to each interested facet of that topic.

uicfbdstd2b¶

Run ID: uicfbdstd2b
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 6d3edfdac61ecca427e685313d8d2ee9
Run description: This facet blog distillation run take stdbaseline2 as input. Due to the lack of document-level IR results, with respect to each interested facet of each topic, we re-rank stdbaseline2 according to the feed ranking which is calculated on the basis of our own feed baseline: uicfeedir2.

uicfbdstd3a¶

Run ID: uicfbdstd3a
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: 4fb3f20cf2fd74bd18cb2bbf667d1b1b
Run description: This facet blog distillation run take stdbaseline3 as input. Due to the lack of document-level IR results, we managed to create a document-level ranking for each topic according to the scores assigned by our own retrieval system. Then, for each topic, we re-rank stdbaseline3 based on the artificial-created document-level ranking with respect to each interested facet of that topic.

uicfbdstd3b¶

Run ID: uicfbdstd3b
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/30/2010
Type: automatic
Task: feed
MD5: ae207ed0bbfb869ba60a540f6151ecab
Run description: This facet blog distillation run take stdbaseline3 as input. Due to the lack of document-level IR results, with respect to each interested facet of each topic, we re-rank stdbaseline3 according to the feed ranking which is calculated on the basis of our own feed baseline, uicfeedir2.

uicfeedir1¶

Run ID: uicfeedir1
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: 7b66d9b087cc31fa802201d936255829
Run description: This run is similar to uicfeedir2, however, we use a slightly different method to calculate the similarity of a feed and a query.

uicfeedir2¶

Run ID: uicfeedir2
Participant: UICIR
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: 31de9d14ff2b31f810719d9a9a63faf4
Run description: This run is generated on basis of improved concept-based retrieval system. We only consider Top 5k retrieved document when calculating the similarity of a feed with respect to a query.

uogTrapeMN5k¶

Run ID: uogTrapeMN5k
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: b46e28cd0acd811474ab4e23e1b513bd
Run description: Advanced voting technique on top of Divergence from Randomness term and proximity weighting models.

uogTrCh¶

Run ID: uogTrCh
Participant: uogTr
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: 8c71f1dfbf6be88abfcb1f98f9001188
Run description: A principled Voting approach using blog posts retrieved using the headline field. Voting is done using COMSUM in conjunction with a variant of CRCS. The overall ranking is subjected to non-English and near duplicate removal and then classified into the named categories using a binary decision based on a normalized probability (0.05).

uogTrdxE¶

Results | Participants | Proceedings | Input | Summary

Run ID: uogTrdxE
Participant: uogTr
Track: Blog
Year: 2010
Submission: 10/4/2010
Type: automatic
Task: newsblogpost
MD5: 7f37124088c19d13c63de16c0655eaf2
Run description: Explicit diversification based on related entities.

uogTrdxF¶

Results | Participants | Proceedings | Input | Summary

Run ID: uogTrdxF
Participant: uogTr
Track: Blog
Year: 2010
Submission: 10/4/2010
Type: automatic
Task: newsblogpost
MD5: e541f40387eff0923c278c81ed7bc8a0
Run description: Explicit diversification based on facet inclinations.

uogTrfC728¶

Run ID: uogTrfC728
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: fcbdd9a9b1d949af2a5ba57583b1fc42
Run description: Classification based facet re-ranker, using many features.

uogTrfC728s1¶

Run ID: uogTrfC728s1
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 2275d26e3457c3796e0c8251b983e9a6
Run description: Classification based facet re-ranker, using many features (cf uogTrfC728), applied on stdbaseline1

uogTrfC728s2¶

Run ID: uogTrfC728s2
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 9a4a4559599b3b3060247c390076d72f
Run description: Classification based facet re-ranker, using many features (cf uogTrfC728), applied on stdbaseline2

uogTrfC728s3¶

Run ID: uogTrfC728s3
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 91084698800440a35531f8d22d4ccc19
Run description: Classification based facet re-ranker, using many features (cf uogTrfC728), applied on stdbaseline3

uogTrfC919¶

Run ID: uogTrfC919
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 65120401f0c52808288b1e98db1a30d1
Run description: Classification based facet re-ranker, using even more features

uogTrfC919s1¶

Run ID: uogTrfC919s1
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 57022033d1e3ca62dcf137503cb68182
Run description: Classification based facet re-ranker, using even more features (cf approach of uogTrfC919), applied to stdbaseline1

uogTrfC919s2¶

Run ID: uogTrfC919s2
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 36ebb131383d127de90c776b8c70bc82
Run description: Classification based facet re-ranker, using even more features (cf approach of uogTrfC919), applied to stdbaseline2

uogTrfC919s3¶

Run ID: uogTrfC919s3
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 9305dacb2feea44028eca893606514b2
Run description: Classification based facet re-ranker, using even more features (cf approach of uogTrfC919), applied to stdbaseline3

uogTrfL728¶

Run ID: uogTrfL728
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: e733eb5825d21b0e2bba37aa32f16bdd
Run description: Learned facet re-ranker, based on many features.

uogTrfL728s1¶

Run ID: uogTrfL728s1
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 32df6b0a0aab169197922a6e6b066007
Run description: Learned facet re-ranker, based on many features (approach cf uogTrfL728), applied to stdbaseline1

uogTrfL728s2¶

Run ID: uogTrfL728s2
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 94e94f05db6481bcbc37c70e49b0a5f6
Run description: Learned facet re-ranker, based on many features (approach cf uogTrfL728), applied to stdbaseline2

uogTrfL728s3¶

Run ID: uogTrfL728s3
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 2ec7f13968d603dc25fad95bf93eaa62
Run description: Learned facet re-ranker, based on many features (approach cf uogTrfL728), applied to stdbaseline3

uogTrfL919¶

Run ID: uogTrfL919
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 45c7ab16122934f4bc27d5c3b05177c0
Run description: Learned facet re-ranker, based on even more features.

uogTrfL919s1¶

Run ID: uogTrfL919s1
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 6c509b8c977c4eff95fde15d0b0231d2
Run description: Learned facet re-ranker, based on even more features (cf uogTrfL919), applied to stdbaseline1.

uogTrfL919s2¶

Run ID: uogTrfL919s2
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: f68df11b8617c8c09bcb7b50001dc692
Run description: Learned facet re-ranker, based on even more features (cf uogTrfL919), applied to stdbaseline2.

uogTrfL919s3¶

Run ID: uogTrfL919s3
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/31/2010
Type: automatic
Task: feed
MD5: 800a7c8fb8fed46892992593bae15c9c
Run description: Learned facet re-ranker, based on even more features (cf uogTrfL919), applied to stdbaseline3.

uogTrL81¶

Results | Participants | Proceedings | Input | Summary

Run ID: uogTrL81
Participant: uogTr
Track: Blog
Year: 2010
Submission: 10/4/2010
Type: automatic
Task: newsblogpost
MD5: aaf2d7b10b8dce27c21234cac2640221
Run description: Learned model including various story ranking and blog post features.

uogTrLC151¶

Run ID: uogTrLC151
Participant: uogTr
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: d21b642484559ba3263d5a572a89cfdc
Run description: A machine learned approach based upon the 09 topics. 151 features derived from our CRCS Voting approach, using both the headline and content under a restricted parameter setting were combined under a learning to rank approach. The overall ranking is subjected to non-English and near duplicate removal and then classified into the named categories using a filtering classification approach.

uogTrLV1076¶

Run ID: uogTrLV1076
Participant: uogTr
Track: Blog
Year: 2010
Submission: 9/6/2010
Type: automatic
Task: topstories
MD5: bbe241c782b5a7d6220132ef3640934d
Run description: A machine learned approach based upon the 09 topics. 1076 features derived from our Votes and CRCS approaches, multiple story representations using both the headline and content as well as varying amounts of historical evidence were combined under a learning to rank approach.

uogTrLv450¶

Run ID: uogTrLv450
Participant: uogTr
Track: Blog
Year: 2010
Submission: 8/10/2010
Type: automatic
Task: blfeed
MD5: 83f3acdfae6514e7bf17446a1b180c52
Run description: Learned voting technique model.