Skip to content

Runs - Entity 2011

CARDHTML

Results | Participants | Proceedings

  • Run ID: CARDHTML
  • Participant: CARD.UALR
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: automatic
  • Task: reflod
  • MD5: d287442d93e99aee2743d326e77a405a
  • Run description: We used dbpedia for entity verification.

CARDHTMLA

Results | Participants | Proceedings

  • Run ID: CARDHTMLA
  • Participant: CARD.UALR
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: manual
  • Task: reflod
  • MD5: f5f1721c7d85f21044520dba892af670
  • Run description: We used dbpedia for entity verification.

CARDHTMLB

Results | Participants | Proceedings

  • Run ID: CARDHTMLB
  • Participant: CARD.UALR
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: automatic
  • Task: reflod
  • MD5: 42e807c85022a396032325967a27f725
  • Run description: We obtained relevant documents from Bing API. We used dbpedia for entity verification.

CARDHTMLM

Results | Participants | Proceedings

  • Run ID: CARDHTMLM
  • Participant: CARD.UALR
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: manual
  • Task: reflod
  • MD5: a71497742858cbb816c13ef9af780245
  • Run description: We used dbpedia for entity verification.

ICSTaveAll

Results | Participants | Input | Summary | Appendix

  • Run ID: ICSTaveAll
  • Participant: ICSTPKU
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: automatic
  • Task: ref
  • MD5: 9bbd10b56ed6ea8b76b279b17125f797
  • Run description: We first recognize homepages from ClueWeb09. Then we get a collection of homepages. And the answer must be in this collection if we have a high confidence of this homepage collection. Then we use this entity list to extract target entities from the related web pages.

ICSTaveSni

Results | Participants | Input | Summary | Appendix

  • Run ID: ICSTaveSni
  • Participant: ICSTPKU
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: automatic
  • Task: ref
  • MD5: 17790d424a62af541930cafedda47f37
  • Run description: We first recognize homepages from ClueWeb09. Then we get a collection of homepages. And the answer must be in this collection if we have a high confidence of this homepage collection. Then we use this entity list to extract target entities from the related web pages.

ICSTmaxAll

Results | Participants | Input | Summary | Appendix

  • Run ID: ICSTmaxAll
  • Participant: ICSTPKU
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: automatic
  • Task: ref
  • MD5: 055853d68e2f4130586ef24528b1d28a
  • Run description: We first recognize homepages from ClueWeb09. Then we get a collection of homepages. And the answer must be in this collection if we have a high confidence of this homepage collection. Then we use this entity list to extract target entities from the related web pages.

ICSTmaxSni

Results | Participants | Input | Summary | Appendix

  • Run ID: ICSTmaxSni
  • Participant: ICSTPKU
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: automatic
  • Task: ref
  • MD5: bb449d8e788395a69fd118d1c79e9b5b
  • Run description: We first recognize homepages from ClueWeb09. Then we get a collection of homepages. And the answer must be in this collection if we have a high confidence of this homepage collection. Then we use this entity list to extract target entities from the related web pages.

ICTNET11ENR1

Results | Participants | Proceedings | Summary

  • Run ID: ICTNET11ENR1
  • Participant: ICTNET
  • Track: Entity
  • Year: 2011
  • Submission: 8/10/2011
  • Type: automatic
  • Task: elc
  • MD5: 3493f9dd6ace047cf7584abad7cfeee8
  • Run description: A simple search on Sindice-2011.

ICTNET11ENR2

Results | Participants | Proceedings | Summary

  • Run ID: ICTNET11ENR2
  • Participant: ICTNET
  • Track: Entity
  • Year: 2011
  • Submission: 8/15/2011
  • Type: automatic
  • Task: elc
  • MD5: 14e566c45099c7d7bc1b3f79bc6183f2
  • Run description: Search on Sindice2011 based on some rules.

ilpslinkcand

Results | Participants | Proceedings | Summary

  • Run ID: ilpslinkcand
  • Participant: COMMIT
  • Track: Entity
  • Year: 2011
  • Submission: 8/14/2011
  • Type: automatic
  • Task: elc
  • MD5: 311759e24924814a73f38148b1b932a8
  • Run description: This method uses the object that examples link to, to retrieve candidates. The candidates are then ranked based on the link overlap with the example entities.

ilpslinkOL

Results | Participants | Proceedings | Summary

  • Run ID: ilpslinkOL
  • Participant: COMMIT
  • Track: Entity
  • Year: 2011
  • Submission: 8/14/2011
  • Type: automatic
  • Task: elc
  • MD5: ace4b04b2811e9bfebf5c63264263dd7
  • Run description: Entities are ranked based on the link overlap between the example entities and candidate entities. Candidate entities are harvested from 3 sources: homepage domain of the source entity, pages linking to the source entity homepage domain, and dbpedia pages relevant to the narrative and source entity.

ilpsPMIcMNZ

Results | Participants | Proceedings | Summary

  • Run ID: ilpsPMIcMNZ
  • Participant: COMMIT
  • Track: Entity
  • Year: 2011
  • Submission: 8/15/2011
  • Type: automatic
  • Task: elc
  • MD5: cd2c9f44176d10488f5a19ac1308e52c
  • Run description: Method combines ranked lists from text, and link based runs. Mehtod weights candidates found by each of the methods by co-ocurrence in the homepage ; inlinks to homepage and wikipedia pages relevant to the narrative.

ilpsTextFilt

Results | Participants | Proceedings | Summary

  • Run ID: ilpsTextFilt
  • Participant: COMMIT
  • Track: Entity
  • Year: 2011
  • Submission: 8/14/2011
  • Type: automatic
  • Task: elc
  • MD5: 5c908369eb6843bcfeee0eebdc33ee90
  • Run description: Method finds candidates by querying a LOD index (ED entity centered) with the relation and source entity. Candidates are filtered based on the "type" field (created by the trec entity tool) and the given dbpedia type.

LIAcwb

Results | Participants | Proceedings | Summary

  • Run ID: LIAcwb
  • Participant: LIA
  • Track: Entity
  • Year: 2011
  • Submission: 8/15/2011
  • Type: automatic
  • Task: elc
  • MD5: ed3dddd092b9c34bbf38bdaf172967ad
  • Run description: Our baseline, with a QA-like framework, using as named entities relevance measures the compacity and confidence in the selected uri. The documents used are retrieved in the clueweb09 with a local indri index.

LIAwb

Results | Participants | Proceedings | Summary

  • Run ID: LIAwb
  • Participant: LIA
  • Track: Entity
  • Year: 2011
  • Submission: 8/15/2011
  • Type: automatic
  • Task: elc
  • MD5: 0d6d7572c17a02706414e4229d3b3b1a
  • Run description: Our second baseline, the same as the LIAcwb run but using Bing api to retrieve document instead of clueweb09.

LIAwc

Results | Participants | Proceedings | Summary

  • Run ID: LIAwc
  • Participant: LIA
  • Track: Entity
  • Year: 2011
  • Submission: 8/15/2011
  • Type: automatic
  • Task: elc
  • MD5: b75085594d9603f7c3dd8d9331a0edab
  • Run description: This run is an extension of LIA-wb. The homepage of the source entities was crawled and added to the document collection retrieved with Bing Api in order to found passages and candidate named entities. Two new scores were introduced here. The first one is how many examples are with a given candidate in lists or tables. The second is computed by downloading snippets about each candidate and example which were then clustered in two groups. A candidate earned more if its associated snippets are with snippets associated to the examples.

LIAwd

Results | Participants | Proceedings | Summary

  • Run ID: LIAwd
  • Participant: LIA
  • Track: Entity
  • Year: 2011
  • Submission: 8/15/2011
  • Type: automatic
  • Task: elc
  • MD5: 0e5748812dd0fd2b69b657e3573c02ca
  • Run description: This run is an extension of LIA-wb. The homepage of the source entities was crawled and added to the document collection retrieved with Bing Api in order to found passages and candidate named entities. Two new scores were introduced here. The first one is how many examples are with a given candidate in lists or tables. The second is computed by downloading snippets about each candidate and example, then for each candidate the similarity between word probabilites in its snippets and the word probabilities in examples's snippets is computed and used to ranked it.

PRISELC1

Results | Participants | Proceedings | Summary

  • Run ID: PRISELC1
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 8/11/2011
  • Type: manual
  • Task: elc
  • MD5: 1f015299354d418027bb046ca814c0b4
  • Run description: DCM model, Entity Activation Forces model. Resources: Lemur ClueWeb Online Query Service, Stanford NER, Wikipedia

PRISELC2

Results | Participants | Proceedings | Summary

  • Run ID: PRISELC2
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 8/11/2011
  • Type: manual
  • Task: elc
  • MD5: ab4e59c2b0edbda9052c2cd50ec6d688
  • Run description: DCM model, distance between keywords and entities. Resources: Google, Stanford NER, Wikipedia

PRISELC3

Results | Participants | Proceedings | Summary

  • Run ID: PRISELC3
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 8/11/2011
  • Type: manual
  • Task: elc
  • MD5: d36f3d2c82fd214ff1882feddbd46c85
  • Run description: DCM model, Entity Activation Forces model. Resources: Google, Stanford NER, Wikipedia

PRISELC4

Results | Participants | Proceedings | Summary

  • Run ID: PRISELC4
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 8/11/2011
  • Type: manual
  • Task: elc
  • MD5: b2fc7ac806c03b9e388b9a556c047462
  • Run description: DCM model, distance between keywords and entities, Entity Activation Forces model. Resources: Google, Stanford NER, Wikipedia

PRISREF1

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: PRISREF1
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: manual
  • Task: ref
  • MD5: b3d562711c4a638bec0d892bc2a17dde
  • Run description: 2010 DCM model, Lemur ClueWeb Online Query Service, Stanford NER, Google

PRISREF2

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: PRISREF2
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: manual
  • Task: ref
  • MD5: 13bdd73835a20f84119dce5b25c678bf
  • Run description: snippet, Lemur ClueWeb Online Query Service, Stanford NER, Google

PRISREF3

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: PRISREF3
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: manual
  • Task: ref
  • MD5: d93bbb900ed68e0048d37d85dc5513a6
  • Run description: 2010 DCM model, distance between keywords and entities, Lemur ClueWeb Online Query Service, Stanford NER, Google

PRISREF4

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: PRISREF4
  • Participant: PRIS
  • Track: Entity
  • Year: 2011
  • Submission: 7/4/2011
  • Type: manual
  • Task: ref
  • MD5: b9d3a1744e69668e4314dca6a9dd74fd
  • Run description: 2011 DCM model, distance between keywords and entities, Lemur ClueWeb Online Query Service, Stanford NER, Google

stecam

Results | Participants | Summary

  • Run ID: stecam
  • Participant: deri
  • Track: Entity
  • Year: 2011
  • Submission: 8/16/2011
  • Type: automatic
  • Task: elc
  • MD5: c1530c6e89a0afeeafd667ac0c0e7e89
  • Run description: - the ranking algorithm extends bm25f by adding an additional structure within a field.

stecam2

Results | Participants | Summary

  • Run ID: stecam2
  • Participant: deri
  • Track: Entity
  • Year: 2011
  • Submission: 8/16/2011
  • Type: automatic
  • Task: elc
  • MD5: 850986123e7a3a9a7d7924e59025b19e
  • Run description: - the ranking algorithm extends bm25f by adding an additional structure within a field; - query dependent weights are used.

stecam3

Results | Participants | Summary

  • Run ID: stecam3
  • Participant: deri
  • Track: Entity
  • Year: 2011
  • Submission: 8/16/2011
  • Type: automatic
  • Task: elc
  • MD5: adaf9adf5184fd68e2c2ea327cda97dc
  • Run description: - the sindice ranking algorithm which is a basic tf-idf, i.e., the one that comes with the Lucene search engine; - query dependent weights.

TongKeyEN1

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: TongKeyEN1
  • Participant: TongKey
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: automatic
  • Task: ref
  • MD5: a6e6be8d0cdbc97cfd70d6846edb69aa
  • Run description: Multiple keywords used for entity ranking Strict filtering rules Pattern recognition Authority page mining Feature-based method for homepage detection

TongKeyEN2

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: TongKeyEN2
  • Participant: TongKey
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: manual
  • Task: ref
  • MD5: d53215c35ac3155d73d4ea26c7270406
  • Run description: Multiple keywords used for entity ranking Pattern recognition Authority page mining Feature-based method for homepage detection

uiRunAll01

Results | Participants | Proceedings | Summary

  • Run ID: uiRunAll01
  • Participant: FASILKOMUI
  • Track: Entity
  • Year: 2011
  • Submission: 8/16/2011
  • Type: automatic
  • Task: elc
  • MD5: c919e26906d51f89e4fd09b9e37ec8af
  • Run description: We use Dbpedia Spotlight (http://spotlight.dbpedia.org) to identify the entities based on the dbpedia target from ClueWeb snippets, by which at the same time we also count the occurences of each entity as a parameter to rank the result list. Beyond that, Dbpedia Spotlight also gives relevant URI for each entities, which could be used to validate the URI from Sindice search results.

uiRunAll02

Results | Participants | Proceedings | Summary

  • Run ID: uiRunAll02
  • Participant: FASILKOMUI
  • Track: Entity
  • Year: 2011
  • Submission: 8/16/2011
  • Type: automatic
  • Task: elc
  • MD5: e77364c700626a843e82f25f23a346e6
  • Run description: The main strategy is the same as Run1. The difference is that in this run, we also included the entity examples from each query in the result list.

uiRunAll03

Results | Participants | Proceedings | Summary

  • Run ID: uiRunAll03
  • Participant: FASILKOMUI
  • Track: Entity
  • Year: 2011
  • Submission: 8/16/2011
  • Type: automatic
  • Task: elc
  • MD5: 95589232267c1ce8bbb7d1a629733522
  • Run description: The main strategy is the same as Run2. The difference is that we adjust the rank function by penalized the score of the example candidates which were included in the result list. For all run, we use the Sindice-ED-2011 Dataset.

WhuRun

Results | Participants | Summary

  • Run ID: WhuRun
  • Participant: irlabwhu2011
  • Track: Entity
  • Year: 2011
  • Submission: 8/16/2011
  • Type: automatic
  • Task: elc
  • MD5: 5ae02ee4a71ab47100859b522ba4fba9
  • Run description: We mainly used the query syntax based SIREN to complete this task

WhuRun1

Results | Participants | Input | Summary | Appendix

  • Run ID: WhuRun1
  • Participant: irlabwhu2011
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: manual
  • Task: ref
  • MD5: f90e16cea466c0bf32c8923fe698c82a
  • Run description: We use natural language processing method to construct an efficient query regarding to the topic description. We analysis the top 200 documents returned by Lemur ClueWeb Online Query Service. Then we use lucene to index the context around candidate entities. At last, narrative is used as query, the rank of the context is the rank of the candidate entity.

WhuRun2

Results | Participants | Input | Summary | Appendix

  • Run ID: WhuRun2
  • Participant: irlabwhu2011
  • Track: Entity
  • Year: 2011
  • Submission: 7/5/2011
  • Type: manual
  • Task: ref
  • MD5: 5f71bf1675787acf9b8d7ed9755167ba
  • Run description: We use natural language processing method to construct an efficient query regarding to the topic description. We analysis the top 200 documents returned by Lemur ClueWeb Online Query Service. Besides the co-occurrance of candidate entity with the given entity, we compute the contribution of each co-occurrance using the context around the candidate entity. Then we sum the score for candidates ranking.