Runs - Session 2012¶

ACombSnip.RL1¶

Results | Participants | Input | Summary | Appendix

Run ID: ACombSnip.RL1
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: 431473f4daae9a3f33096e1e21b02366
Run description: Sequential Dependency + Combining Related Queries + Query Expansion using Snippets and Clicked Snippets

ACombSnip.RL2¶

Results | Participants | Input | Summary | Appendix

Run ID: ACombSnip.RL2
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: 710fc8ed5534c73e8c3faaf0eb4f512c
Run description: Sequential Dependency + Combining Related Queries + Query Expansion using Snippets and Clicked Snippets

ACombSnip.RL3¶

Results | Participants | Input | Summary | Appendix

Run ID: ACombSnip.RL3
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: b4a4a86b61a89daec3aef079cccfc6d6
Run description: Sequential Dependency + Combining Related Queries + Query Expansion using Snippets and Clicked Snippets

ACombSnip.RL4¶

Results | Participants | Input | Summary | Appendix

Run ID: ACombSnip.RL4
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 848adae37ad459e434085926580c9b3c
Run description: Sequential Dependency + Combining Related Queries + Query Expansion using Snippets and Clicked Snippets

BDocExpDoc.RL1¶

Results | Participants | Input | Summary | Appendix

Run ID: BDocExpDoc.RL1
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: 9e44de3aa8da3a381f79f95c1d9d9e92
Run description: Sequential Dependency + Query Expansion from Previous Queries Document, Document and Clicked Document

BDocExpDoc.RL2¶

Results | Participants | Input | Summary | Appendix

Run ID: BDocExpDoc.RL2
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: ae6b0611c4ec13cc39cf67cbee8826ff
Run description: Sequential Dependency + Query Expansion from Previous Queries Document, Document and Clicked Document

BDocExpDoc.RL3¶

Results | Participants | Input | Summary | Appendix

Run ID: BDocExpDoc.RL3
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: 8b85596685108900c964028332c22965
Run description: Sequential Dependency + Query Expansion from Previous Queries Document, Document and Clicked Document

BDocExpDoc.RL4¶

Results | Participants | Input | Summary | Appendix

Run ID: BDocExpDoc.RL4
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 5dc8c621de33b168b7b38ec9088a443c
Run description: Sequential Dependency + Query Expansion from Previous Queries Document, Document and Clicked Document

CWIrun1.RL1¶

Run ID: CWIrun1.RL1
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: 36a347932fb21e6d46ca8596aaf10e10
Run description: This run tests a novel logic-probabilistic model: task RL1 uses a uninformative prior

CWIrun1.RL2¶

Run ID: CWIrun1.RL2
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: f85cab553ce7d1bde01b0e55e3b92fe3
Run description: This run tests a novel logic-probabilistic model: task RL2 uses a uninformative prior updated with a query event whose likelihood is estimated by means of a Kolmogorov-Smirnov test

CWIrun1.RL3¶

Run ID: CWIrun1.RL3
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: 20be2926572fbcfa31525bc6163e1468
Run description: This run tests a novel logic-probabilistic model: task RL2 uses a prior extended with result lists and updated with a query event whose likelihood is estimated by means of a Kolmogorov-Smirnov test.

CWIrun1.RL4¶

Run ID: CWIrun1.RL4
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 20be2926572fbcfa31525bc6163e1468
Run description: This run tests a novel logic-probabilistic model: task RL3 uses a prior extended with result lists and updated with a query event whose likelihood is estimated by means of a Kolmogorov-Smirnov test.

CWIrun3.RL1¶

Run ID: CWIrun3.RL1
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: 74697df95fd01c63206efeb29f79d3f6
Run description: logic probabilistic model, uninformative prior based on a vocabulary extracted from the top 3 documents in response to the topic descriptions

CWIrun3.RL2¶

Run ID: CWIrun3.RL2
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: 912b3458f8afe825e784af7119e152e0
Run description: logic probabilistic model, uninformative prior based on a vocabulary extracted from the top 3 documents in response to the topic descriptions. In RL2 the prior is updated with the observed queries + a likelihood function for the query event

CWIrun3.RL3¶

Run ID: CWIrun3.RL3
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: ac018369b015119d322056852b372786
Run description: logic probabilistic model, uninformative prior based on a vocabulary extracted from the top 3 documents in response to the topic descriptions. In RL3 the prior is extended with the subtopics derived from the result lists

CWIrun3.RL4¶

Run ID: CWIrun3.RL4
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: f16d9c5c8bbfdf4886d3fe2b21a794e9
Run description: logic probabilistic model, uninformative prior based on a vocabulary extracted from the top 3 documents in response to the topic descriptions. In RL4 the prior is extended with the subtopics derived from the click lists

essexSAnchor.RL1¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSAnchor.RL1
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL1
MD5: c71f77b8c8a437e1b648bf66ae193070
Run description: RL1 is simply generated by using the current query and the maximum likelihood language model. Waterloo spam rankings were used to filter spam is applied as a post retrieval process.

essexSAnchor.RL2¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSAnchor.RL2
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL2
MD5: 718cee4017ff0e73bbf6ff82b297df7c
Run description: RL2 was generated by expanding the current query with expansions inferred using all the queries in the session. Association rules were applied on query logs (simulated from anchor logs) to derive the expansions.

essexSAnchor.RL3¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSAnchor.RL3
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL3
MD5: 151b770848dba83fba465b755c18c526
Run description: RL3 was generated by expanding the current query with expansions inferred from the anchor text of documents displayed to the user throughout the session.

essexSAnchor.RL4¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSAnchor.RL4
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL4
MD5: 0401f3837d355b1fbe497073d6385991
Run description: RL4 was generated by expanding the current query with expansions inferred from the anchor text of documents clicked by the user throughout the session.

essexSWiki.RL1¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSWiki.RL1
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL1
MD5: 159016359d1b1c06df34ad2a9b9906a9
Run description: RL1 is simply generated by using the current query and the maximum likelihood language model. Waterloo Spam rankings were used to filter spam as a post retrieval process.

essexSWiki.RL2¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSWiki.RL2
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL2
MD5: e222504c5d7db02f262ba1c31e87831c
Run description: RL2 was generated by using the current query and the term 'wikipedia' as an expansion!

essexSWiki.RL3¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSWiki.RL3
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL3
MD5: e222504c5d7db02f262ba1c31e87831c
Run description: RL3 was generated by using the current query and the term 'wikipedia' as an expansion!

essexSWiki.RL4¶

Results | Participants | Input | Summary | Appendix

Run ID: essexSWiki.RL4
Participant: UoE
Track: Session
Year: 2012
Submission: 8/24/2012
Type: automatic
Task: RL4
MD5: e222504c5d7db02f262ba1c31e87831c
Run description: RL4 was generated by using the current query and the term 'wikipedia' as an expansion!

guphrase1.RL1¶

Run ID: guphrase1.RL1
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: bc0bca90daadf71569a285bd00a4db9b
Run description: find phrases in the current query from feedback

guphrase1.RL2¶

Run ID: guphrase1.RL2
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: 383220988e845d923cbc4b0a3ce59267
Run description: find phrases in the previous and current queries from feedback

guphrase1.RL3¶

Run ID: guphrase1.RL3
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: 526909ccdbc89401e13e22da274154a4
Run description: select previous queries by discovering user's intention

guphrase1.RL4¶

Run ID: guphrase1.RL4
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 526909ccdbc89401e13e22da274154a4
Run description: select previous queries by discovering user's intention from click data

guphrase2.RL1¶

Run ID: guphrase2.RL1
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: cd29bc35adf7ee2e8652c98eb505d221
Run description: group the current query into phrases based on the feedback

guphrase2.RL2¶

Run ID: guphrase2.RL2
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: a378fae6d2660872e4d438e619a43a73
Run description: group the previous and current queries into phrases based on the feedback

guphrase2.RL3¶

Run ID: guphrase2.RL3
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: 2bfedfe37056c17757a1f366baacbb6a
Run description: select the previous results by discovering the user's intention

guphrase2.RL4¶

Run ID: guphrase2.RL4
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 2bfedfe37056c17757a1f366baacbb6a
Run description: select the previous results by discovering the user's intention and click data

gurelaxphr.RL1¶

Run ID: gurelaxphr.RL1
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: a354aeff937251c37655a1f34031f720
Run description: group words in the current query into phrases by a relaxed rule from the feedback

gurelaxphr.RL2¶

Run ID: gurelaxphr.RL2
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: 6fd20c84f40a4e20d578f0060ab131e3
Run description: group words in the previous and current queries into phrases by a relaxed rule from the feedback

gurelaxphr.RL3¶

Run ID: gurelaxphr.RL3
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: 8e9f598f0d749f292553effa8aa329c9
Run description: choose the previous queries by discovering user's intention

gurelaxphr.RL4¶

Run ID: gurelaxphr.RL4
Participant: Georgetown
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 57c8fb2013db4af681c64f13ff1f417e
Run description: choose the previous queries by discovering user's intention from the click data

ICTNET12SER1.RL1¶

Run ID: ICTNET12SER1.RL1
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL1
MD5: 53c47b742135503ceda47c213ad588a9
Run description: result from our ad-hoc system

ICTNET12SER1.RL2¶

Run ID: ICTNET12SER1.RL2
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL2
MD5: 376753066933a50fd1cc5603433f4f2f
Run description: rerank ad-hoc results based on query extending

ICTNET12SER1.RL3¶

Run ID: ICTNET12SER1.RL3
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL3
MD5: 07af76f877878863a5379455e1279b47
Run description: construct a virtual document of the query from the ranked list and rerank ad-hoc results based on cos_sim score

ICTNET12SER1.RL4¶

Run ID: ICTNET12SER1.RL4
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL4
MD5: 0bbfb20bbcf76af4f988a8582817a0bb
Run description: using clicked document(both contents and dwell time) to reranking our ad-hoc results

ICTNET12SER2.RL1¶

Run ID: ICTNET12SER2.RL1
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL1
MD5: cc6b8a10be981d9b862986559877ca4a
Run description: using cos-sim with google search reuslts as the description of the query

ICTNET12SER2.RL2¶

Run ID: ICTNET12SER2.RL2
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL2
MD5: 261a6d6d86e6eedd007aaefff3836bd4
Run description: using learning to rank algorithm to rerank our ad-hoc results

ICTNET12SER2.RL3¶

Run ID: ICTNET12SER2.RL3
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL3
MD5: 97935c18a0a10aca6961675f95988322
Run description: using learning to rank algorithm to rerank our ad-hoc results

ICTNET12SER2.RL4¶

Run ID: ICTNET12SER2.RL4
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL4
MD5: 9d54a49ce7dfa5f0000569376b734665
Run description: using learning to rank algorithm to rerank our ad-hoc results

ICTNET12SER3.RL1¶

Run ID: ICTNET12SER3.RL1
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL1
MD5: 41530843d5bfe68b39be55cdc682d9fd
Run description: using learning to rank algorithm to rerank our ad-hoc results, with a feature calculating cos-sim with google search reuslts as the description of the query

ICTNET12SER3.RL2¶

Run ID: ICTNET12SER3.RL2
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL2
MD5: 0a36edd065fefc71d674baa7a11fcd5d
Run description: using learning to rank algorithm to rerank our ad-hoc results, with a feature calculating cos-sim with google search reuslts as the description of the query

ICTNET12SER3.RL3¶

Run ID: ICTNET12SER3.RL3
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL3
MD5: 2882914c0538993af9b376133d6884f1
Run description: using learning to rank algorithm to rerank our ad-hoc results, with a feature calculating cos-sim with google search reuslts as the description of the query

ICTNET12SER3.RL4¶

Run ID: ICTNET12SER3.RL4
Participant: ICTNET
Track: Session
Year: 2012
Submission: 8/26/2012
Type: automatic
Task: RL4
MD5: 49d249e59aa58e23a5587f0d809ebee0
Run description: using learning to rank algorithm to rerank our ad-hoc results, with a feature calculating cos-sim with google search reuslts as the description of the query

PITTSHQM.RL1¶

Run ID: PITTSHQM.RL1
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL1
MD5: b0e41c0543bcaedf0bd93cd6ed906a4d
Run description: Using SHQM method last year (without SDM features).

PITTSHQM.RL2¶

Run ID: PITTSHQM.RL2
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL2
MD5: 286fbd9f01a172b1b77853afe833f2be
Run description: Using SHQM method last year (without SDM features).

PITTSHQM.RL3¶

Run ID: PITTSHQM.RL3
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL3
MD5: 5c3117f518ed237455673c2c7cd91e0c
Run description: Using SHQM method last year (without SDM features).

PITTSHQM.RL4¶

Run ID: PITTSHQM.RL4
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL4
MD5: 83c8961d9a64bbd09dc378d604d235e3
Run description: Using SHQM method last year (without SDM features).

PITTSHQMnov.RL1¶

Run ID: PITTSHQMnov.RL1
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL1
MD5: 603114b0632831cf8e7efe0f735e3803
Run description: Using SHQM method last year (without SDM features) and consider adaptive browsing novelty.

PITTSHQMnov.RL2¶

Run ID: PITTSHQMnov.RL2
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL2
MD5: 3e0550dd4929b2bc26d5bf0e4ec5bd80
Run description: Using SHQM method last year (without SDM features) and consider adaptive browsing novelty.

PITTSHQMnov.RL3¶

Run ID: PITTSHQMnov.RL3
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL3
MD5: 8149c3a8fe49d7e7e377bb8eb6468bc8
Run description: Using SHQM method last year (without SDM features) and consider adaptive browsing novelty.

PITTSHQMnov.RL4¶

Run ID: PITTSHQMnov.RL4
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL4
MD5: f9678863ee6b3d3a6fa78fd7a8a7d2a1
Run description: Using SHQM method last year (without SDM features) and consider adaptive browsing novelty.

PITTSHQMsdm.RL1¶

Run ID: PITTSHQMsdm.RL1
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL1
MD5: a66059d8c2813cb703d15484ca524eb7
Run description: Using SHQM method last year (including SDM features).

PITTSHQMsdm.RL2¶

Run ID: PITTSHQMsdm.RL2
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL2
MD5: ff2a5184279f0deed9dfb04e8a3ae7b9
Run description: Using SHQM method last year (including SDM features).

PITTSHQMsdm.RL3¶

Run ID: PITTSHQMsdm.RL3
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL3
MD5: 58be4fd867edaa04039dfdd9de4911ff
Run description: Using SHQM method last year (including SDM features).

PITTSHQMsdm.RL4¶

Run ID: PITTSHQMsdm.RL4
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL4
MD5: 67a65cd9a74e4af2649fe6054d997fc6
Run description: Using SHQM method last year (including SDM features).

PITTSHQMsnov.RL1¶

Run ID: PITTSHQMsnov.RL1
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL1
MD5: f4d7d6b878790416c94cf8b3815ab985
Run description: Using SHQM method last year (including SDM features) considering adaptive browsing novelty.

PITTSHQMsnov.RL2¶

Run ID: PITTSHQMsnov.RL2
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL2
MD5: 436c7e7ea23589e91847cb6561173ab8
Run description: Using SHQM method last year (including SDM features) considering adaptive browsing novelty.

PITTSHQMsnov.RL3¶

Run ID: PITTSHQMsnov.RL3
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL3
MD5: 56413664309c1bd27eba020a3969a6ef
Run description: Using SHQM method last year (including SDM features) considering adaptive browsing novelty.

PITTSHQMsnov.RL4¶

Run ID: PITTSHQMsnov.RL4
Participant: PITT
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL4
MD5: 72d941c34dbeaed44cdb018fa31ac900
Run description: Using SHQM method last year (including SDM features) considering adaptive browsing novelty.

RutgersHu.RL1¶

Run ID: RutgersHu.RL1
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL1
MD5: f115840d65e2204b47d2f7fcd4cd0c32
Run description: RUN RutgersHu first manually classified the task type according to task description, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

RutgersHu.RL2¶

Run ID: RutgersHu.RL2
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL2
MD5: 675b458e295fc8d156fa967b748d9a79
Run description: RUN RutgersHu first manually classified the task type according to task description, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

RutgersHu.RL3¶

Run ID: RutgersHu.RL3
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL3
MD5: 675b458e295fc8d156fa967b748d9a79
Run description: RUN RutgersHu first manually classified the task type according to task description, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

RutgersHu.RL4¶

Run ID: RutgersHu.RL4
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL4
MD5: 285be65252c79035e769851f3c90dfee
Run description: RUN RutgersHu first manually classified the task type according to task description, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

RutgersM.RL1¶

Run ID: RutgersM.RL1
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL1
MD5: 8d24bda4cd860755263f9d8bbfb09b20
Run description: RUN RutgersM used behavioral measures in each session and applied the predictive model to predict the task type automatically, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

RutgersM.RL2¶

Run ID: RutgersM.RL2
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL2
MD5: f413e405ad1c81b9b07d4e98e2d94f4e
Run description: RUN RutgersM used behavioral measures in each session and applied the predictive model to predict the task type automatically, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

RutgersM.RL3¶

Run ID: RutgersM.RL3
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL3
MD5: f413e405ad1c81b9b07d4e98e2d94f4e
Run description: RUN RutgersM used behavioral measures in each session and applied the predictive model to predict the task type automatically, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

RutgersM.RL4¶

Run ID: RutgersM.RL4
Participant: ruiiltrec2012
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL4
MD5: b2b463cdad520d640b97aaa5c486fd73
Run description: RUN RutgersM used behavioral measures in each session and applied the predictive model to predict the task type automatically, and then applied task-specific model to predict document usefulness. Relevance feedback was conducted on the predicted useful documents.

TUDrun.RL1¶

Run ID: TUDrun.RL1
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: a149a1363b7eeec074820c0ed54efdcd
Run description: Run developed in collaboration with TUD: baseline

TUDrun.RL2¶

Run ID: TUDrun.RL2
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: 36812fb11010fabaf8ec731c650267d9
Run description: Run developed in collaboration with TUD: similarity based on a co-occurrence graph

TUDrun.RL3¶

Run ID: TUDrun.RL3
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: 36812fb11010fabaf8ec731c650267d9
Run description: Run developed in collaboration with TUD: similarity based on a co-occurrence graph

TUDrun.RL4¶

Run ID: TUDrun.RL4
Participant: CWI
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 36812fb11010fabaf8ec731c650267d9
Run description: Run developed in collaboration with TUD: similarity based on a co-occurrence graph

UAlbany.RL1¶

Run ID: UAlbany.RL1
Participant: UAlbanySession
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL1
MD5: 8d05a75d627e7948950493703cb34a76
Run description: used pseudo relevant feedback; limited symbols ("'",".","(",")") in the queries were manually changed to 64 based

UAlbany.RL2¶

Run ID: UAlbany.RL2
Participant: UAlbanySession
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL2
MD5: a45e3d8f2beaac99136ebad2ee10e66f
Run description: compared the adjacent two queries and picked the 'good' query; used the good queries to expand the 'current query'; limited symbols (like "'", "." in a URL, "(", ")") were converted to 64 based format manually; pseudo relevant feedback was used;

UAlbany.RL3¶

Run ID: UAlbany.RL3
Participant: UAlbanySession
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL3
MD5: db08d9d69ce6e4400547bbd28afe7d67
Run description: compared the adjacent two queries and picked the 'good' query; criteria were set to pick 'good' pages of 'good' queries; used the good queries and the titles of the good pages to expand the 'current query'; limited symbols (like "'", "." in a URL, "(", ")") were converted to 64 based format manually; pseudo relevant feedback was used;

UAlbany.RL4¶

Run ID: UAlbany.RL4
Participant: UAlbanySession
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL4
MD5: c5a47cc4cc205efe46f496899dfac152
Run description: compared the adjacent two queries and picked the 'good' query; results in the clicked list of the good queries were judged and good pages were picked; if there was no click for an interaction, good pages were picked using the same criteria as those used in RL3; used the good queries and the titles of the good pages to expand the 'current query'; limited symbols (like "'", "." in a URL, "(", ")") were converted to 64 based format manually; pseudo relevant feedback was used;

webis12cnqe.RL1¶

Run ID: webis12cnqe.RL1
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL1
MD5: f133262ca47ca3bd25be6a8cc3f4db1e
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12indqe is that the ChatNoir search engine is used instead of Indri.

webis12cnqe.RL2¶

Run ID: webis12cnqe.RL2
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL2
MD5: 9eea8c34fe168ec34bd3d9bcabf45f9c
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12indqe is that the ChatNoir search engine is used instead of Indri.

webis12cnqe.RL3¶

Run ID: webis12cnqe.RL3
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL3
MD5: d83ddc26f1d5ae237139a9ac27239c9c
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12indqe is that the ChatNoir search engine is used instead of Indri.

webis12cnqe.RL4¶

Run ID: webis12cnqe.RL4
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL4
MD5: fce7aaf91845d8daf8bfe70db69c1712
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12indqe is that the ChatNoir search engine is used instead of Indri.

webis12cnse.RL1¶

Run ID: webis12cnse.RL1
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL1
MD5: fcbba1b6671cc49d1217c78ada97204c
Run description: In this run we manually defined which sessions in the TREC 2012 session track dataset cover the same topic and expand the given query only with key phrases extracted from these similar sessions (from queries in RL2, from document titles, snippets and full texts in RL3). In RL4 we add results that were already clicked by users in other sessions, but this time we don't restrict this expansion to documents that were clicked at least two times. Thus, much more documents are added to the result list than in webis12cnqe and webis12indqe.

webis12cnse.RL2¶

Run ID: webis12cnse.RL2
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL2
MD5: 3b7274d94efd4bc59307cdc3dcda7fee
Run description: In this run we manually defined which sessions in the TREC 2012 session track dataset cover the same topic and expand the given query only with key phrases extracted from these similar sessions (from queries in RL2, from document titles, snippets and full texts in RL3). In RL4 we add results that were already clicked by users in other sessions, but this time we don't restrict this expansion to documents that were clicked at least two times. Thus, much more documents are added to the result list than in webis12cnqe and webis12indqe.

webis12cnse.RL3¶

Run ID: webis12cnse.RL3
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL3
MD5: 367954bd6f287909bfb2c569ba68c130
Run description: In this run we manually defined which sessions in the TREC 2012 session track dataset cover the same topic and expand the given query only with key phrases extracted from these similar sessions (from queries in RL2, from document titles, snippets and full texts in RL3). In RL4 we add results that were already clicked by users in other sessions, but this time we don't restrict this expansion to documents that were clicked at least two times. Thus, much more documents are added to the result list than in webis12cnqe and webis12indqe.

webis12cnse.RL4¶

Run ID: webis12cnse.RL4
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: manual
Task: RL4
MD5: ec8ceb7567fa6d70f0cd772052bcd151
Run description: In this run we manually defined which sessions in the TREC 2012 session track dataset cover the same topic and expand the given query only with key phrases extracted from these similar sessions (from queries in RL2, from document titles, snippets and full texts in RL3). In RL4 we add results that were already clicked by users in other sessions, but this time we don't restrict this expansion to documents that were clicked at least two times. Thus, much more documents are added to the result list than in webis12cnqe and webis12indqe.

webis12indqe.RL1¶

Run ID: webis12indqe.RL1
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL1
MD5: b26523e98e7b60142c65ba695e054112
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (spam, duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12cnqe is that the Indri search engine is used instead of ChatNoir.

webis12indqe.RL2¶

Run ID: webis12indqe.RL2
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL2
MD5: dafa2a7e54391f998af4e83b01a22d5c
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (spam, duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12cnqe is that the Indri search engine is used instead of ChatNoir.

webis12indqe.RL3¶

Run ID: webis12indqe.RL3
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL3
MD5: 03d70f4467c1286c5193d51b435a418c
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (spam, duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12cnqe is that the Indri search engine is used instead of ChatNoir.

webis12indqe.RL4¶

Run ID: webis12indqe.RL4
Participant: webis
Track: Session
Year: 2012
Submission: 8/29/2012
Type: automatic
Task: RL4
MD5: 1e0da4d51f4575513bd7c76ddd2f432d
Run description: This full automatic run first decides whether query expansion might be useful or not (we identified new queries and acronyms to be useful). Then, we automatically extract key phrases of the previous queries (RL2) and of the titles, snippets and full document texts of the already shown results (RL3). In RL4 we use only these documents for key phrase extraction that were clicked by the user and thus seem to be especially helpful. The original queries get segmented and Wikipedia articles whose titles exactly match one of the segments are added to the result list. Furthermore in RL4 we add documents of automatically detected similar sessions (given both in the TREC 2012 session track data and of an locally existing search log) that were clicked at least two times. In a last step the result list gets filtered (spam, duplicates, results with more than 7000 words and already clicked documents) and Wikipedia articles in the top-100 get ranked to top. The difference to webis12cnqe is that the Indri search engine is used instead of ChatNoir.

wildcat1.RL1¶

Run ID: wildcat1.RL1
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL1
MD5: d91297360999fdc93adfc658c55b25be
Run description: wildcat1.RL1 uses the current query to get the search results. The documents with spam score less than 40 are filtered out.

wildcat1.RL2¶

Run ID: wildcat1.RL2
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL2
MD5: 14b2ed72397882c8e2a941c24235b6f0
Run description: wildcat1.RL2 uses user behavior model to do the query expansion. The terms in current query and in history queries are given different weights.

wildcat1.RL3¶

Run ID: wildcat1.RL3
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL3
MD5: 5e6f60582f2d804845368e58809e4f11
Run description: wildcat1.RL3 uses anchor texts provided by Twente to do the query expansion. This run extracts the anchor texts of the websites in ranked lists and extracts 10 words with highest frequency in the anchor texts. wildcat1.RL3 added the top 10 words to the query to generate a new query. The terms in current query have weight 0.7 and the expanded terms of anchor texts have weight 0.3 .

wildcat1.RL4¶

Run ID: wildcat1.RL4
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL4
MD5: 26acf6b7f161abc9c12201382b46f4e5
Run description: In this run, we use the results of wildcat1.RL3 and re-rank the results according to their similarities with the clicked documents. The attention time on the clicked documents affect the re-rank results.

wildcat2.RL1¶

Run ID: wildcat2.RL1
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL1
MD5: f1541c5963fa97dc96de0c8fefa3ccc4
Run description: wildcat2.RL1 uses the current query to get the search results. The documents are re-ranked by pagerank score.

wildcat2.RL2¶

Run ID: wildcat2.RL2
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL2
MD5: 92d6a1af65d83f4a2528b4d355d3d60b
Run description: wildcat2.RL2 uses the current query to get the search results. The documents are re-ranked by pagerank score and indri score.

wildcat2.RL3¶

Run ID: wildcat2.RL3
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL3
MD5: a4f55645ecba3bd1689930822b2c4780
Run description: In this run, we use the rank information of ranked lists for history queries to re-rank the search results of wildcat1.RL2

wildcat2.RL4¶

Run ID: wildcat2.RL4
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL4
MD5: 2fe49aacb6a3dd99efd6323fd1d2be78
Run description: In this run, we use the titles and snippets of the clicked documents to expand the query. This run uses the expanded query to get the search results and the spam documents are filtered out.

wildcat3.RL1¶

Run ID: wildcat3.RL1
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL1
MD5: a71faf6997552e0fbf4685fe79b7cc6f
Run description: wildcat3.RL1 uses the current query to get the search results. And this run calculates the similarity between the returned search results and the current query, then re-ranks the results according to the similarity score.

wildcat3.RL2¶

Run ID: wildcat3.RL2
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL2
MD5: 26bf4082c2082dd777ff1a35abe91101
Run description: This run uses user behavior model to do the query expansion.After the results are returned, this run calculates the similarity between the returned documents in search results and the expanded query, then re-ranks the results according to the similarity score.

wildcat3.RL3¶

Run ID: wildcat3.RL3
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL3
MD5: d23110edb2ce649121e972e411240794
Run description: wildcat3.RL3 uses the meta tags in the documents of ranked lists for past queries to do the query expansion. The terms in the current query have weight 0.5 and the top 10 terms extracted from meta have weight 0.5 .

wildcat3.RL4¶

Run ID: wildcat3.RL4
Participant: pris411
Track: Session
Year: 2012
Submission: 8/28/2012
Type: automatic
Task: RL4
MD5: cb9dbb1eb1cdd5c70088e7d2c40dd6d4
Run description: wildcat3.RL4 uses the clicked order to do the query expansion. wildcat3.RL4 extracts terms of the clicked titles and gives them with different weights according to the click sequence. Then this run combines the current query with the weighted terms of clicked titles to generate the new query.

WQExpFqDSnip.RL1¶

Results | Participants | Input | Summary | Appendix

Run ID: WQExpFqDSnip.RL1
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL1
MD5: 4d263dd08180d38065e8c7c972ab4751
Run description: Sequential Dependency + Query Expansion from Previous Queries, Snippets and Clicked Snippets

WQExpFqDSnip.RL2¶

Results | Participants | Input | Summary | Appendix

Run ID: WQExpFqDSnip.RL2
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL2
MD5: 5058c66f686319002bb2a8417207564d
Run description: Sequential Dependency + Query Expansion from Previous Queries, Snippets and Clicked Snippets

WQExpFqDSnip.RL3¶

Results | Participants | Input | Summary | Appendix

Run ID: WQExpFqDSnip.RL3
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL3
MD5: b4892b9d375c6b1b5614b741fc64defe
Run description: Sequential Dependency + Query Expansion from Previous Queries, Snippets and Clicked Snippets

WQExpFqDSnip.RL4¶

Results | Participants | Input | Summary | Appendix

Run ID: WQExpFqDSnip.RL4
Participant: udel
Track: Session
Year: 2012
Submission: 8/30/2012
Type: automatic
Task: RL4
MD5: 79c84b4039ce4a349a7978c9cd77b48c
Run description: Sequential Dependency + Query Expansion from Previous Queries, Snippets and Clicked Snippets