Skip to content

Runs - Web 2010

1

Results | Participants | Input | Summary | Appendix

  • Run ID: 1
  • Participant: budapest_acad
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: spam
  • Run description: LogitBoost on basic content features.

2

Results | Participants | Input | Summary | Appendix

  • Run ID: 2
  • Participant: budapest_acad
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: spam
  • Run description: LogitBoost on basic content features. Only host level spam filtering.

blv79y00prob

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: blv79y00prob
  • Participant: blv1979
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 4290727cb6dc0c4259d12710f6155341
  • Run description: The ranking method models the probability of a document to be relevant based on the following predictor variables: BM25DOC, BM25PAIR, SPAMRANK, an indicator of word pair presence at distance less than 10 ("close pair"), and 75th percentile of the distances between words constituting a close pair. A linear combination of these predictor variables is converted to the probability measure using the logit function. Coefficients reflecting the impact of each predictor were found using direct maximization of the MAP. We use spam ranks for ClueWeb09 collection provided by Cormack et al 2010.

blv79y00shnk

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: blv79y00shnk
  • Participant: blv1979
  • Track: Web
  • Year: 2010
  • Submission: 8/6/2010
  • Task: adhoc
  • MD5: 8690e4adf20a238b50a6f6fc31db8134
  • Run description: This run uses the modified Buetcher proximity model as described in Schenkel et al 2007. Additionally, we use spam ranks for ClueWeb09 collection provided by Cormack et al 2010. Proximity pairs for most often occurring words are precomputed during indexing to improve retrieval time.

cmuBase10

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: cmuBase10
  • Participant: CMU_LIRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 1c0383dbe56a80edcbb6c89c3166301d
  • Run description: Baseline.

cmuComb10

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: cmuComb10
  • Participant: CMU_LIRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: e32bcb450093a63934e01fae6188964a
  • Run description: Combination of web & wiki expansion

cmuFuTop10

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: cmuFuTop10
  • Participant: CMU_LIRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 33599e19e968ee29104a09460c25eb09
  • Run description: Combines web & wiki expansion

cmuFuTop10D

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: cmuFuTop10D
  • Participant: CMU_LIRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: e908b4957abbf76260639056d2826823
  • Run description: Combines wiki & web expansion

cmuWi10D

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: cmuWi10D
  • Participant: CMU_LIRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 1038841dcee3308c47fca00b1b2d1235
  • Run description: Query expansion over Wikipedia

cmuWiki10

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: cmuWiki10
  • Participant: CMU_LIRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 8858acc1dd8ac466bb44d3a973e21edc
  • Run description: Makes use of query expansion over Wikipedia

DFalah2010

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: DFalah2010
  • Participant: uottawa
  • Track: Web
  • Year: 2010
  • Submission: 8/11/2010
  • Task: adhoc
  • MD5: 978e768bc1b28d897db933ee8137fa6c
  • Run description: selective indexing of text and HTML format

ICTNETAD10R1

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: ICTNETAD10R1
  • Participant: ICTNET
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: 26079a032fd7ca22219350af9199c14b
  • Run description: using BM25 sliding-window model + spamming for spamming, we used Waterloo Spam Rankings for the ClueWeb09 dataset.

ICTNETAD10R2

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: ICTNETAD10R2
  • Participant: ICTNET
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: 457e2c259b831ac0a26b5d1b0b2beb74
  • Run description: using BM25 sliding-window model + title + spamming for spamming, we used Waterloo Spam Rankings for the ClueWeb09 dataset.

ICTNETAD10R3

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: ICTNETAD10R3
  • Participant: ICTNET
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: 5249539b3e351e217a09fe996724e072
  • Run description: using BM25 sliding-window model + title + spamming + pagerank for spamming, we used Waterloo Spam Rankings for the ClueWeb09 dataset.

ICTNETDV10R1

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: ICTNETDV10R1
  • Participant: ICTNET
  • Track: Web
  • Year: 2010
  • Submission: 8/8/2010
  • Task: diversity
  • MD5: e871a42adeed6f95e3c0f9aada966dec
  • Run description: This run is based on clustering ad-hoc results with spamming and query expansion from search engines. For ad-hoc results spamming, we used the Waterloo Spam Rankings for the ClueWeb09 Dataset. For query expansion, we used results from www.google.com, www.cuil.com and search.yippy.com.

ICTNETDV10R2

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: ICTNETDV10R2
  • Participant: ICTNET
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 459585b60b468b83702b35741d35e7ba
  • Run description: In this run, we performed clustering algorithm 120 times and took the average. For ad-hoc results spamming, we used the Waterloo Spam Rankings for the ClueWeb09 Dataset. For query expansion, we used results from www.google.com, www.cuil.com and search.yippy.com.

ICTNETDV10R3

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: ICTNETDV10R3
  • Participant: ICTNET
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: e53a6e795b3fd6ff1c332f6c1d24f831
  • Run description: In this run, we used in-link anchors to diversify the search results. For ad-hoc results spamming, we used the Waterloo Spam Rankings for the ClueWeb09 Dataset. For query expansion, we used results from www.google.com, www.cuil.com and search.yippy.com.

ICTNETSP10R1

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: ICTNETSP10R1
  • Participant: ICTNET
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: spam
  • Run description: Firstly, we count the balance of the page-rank value. It means we count page-rank twice. One is ordinary and the other is at the base of web link structure re-build. We cut off some web link while its recognized as a spamming link. Secondly, we get a spamming score from a content-based classifier. Then, we subtract from this content-based spamming score using the balance of page-rank value. The result is used as the final indicator of one pages spamming level.

irra10b

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: irra10b
  • Participant: IRRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 81476d67a963d41fbb3e12de69540df1
  • Run description: This is the base run.

irra10hp

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: irra10hp
  • Participant: IRRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 2d4d58de756509d12954f01f1df2fe35
  • Run description: This is a high precision run

irra10rob

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: irra10rob
  • Participant: IRRA
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 9a56570f4deb7139cb1d3cc6fd7f4488
  • Run description: This is a robust run

IVORY.70.30

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: IVORY.70.30
  • Participant: UMD
  • Track: Web
  • Year: 2010
  • Submission: 9/15/2010
  • Task: spam
  • MD5: e69fc83fe74537b8cf2af4030965505f
  • Run description: A linear combination of TrustRank and Anti-TrustRank scores, seeded with Waterloo Spamminess scores thresholded at 70 and 30 respectively.

IVORY.90.10

Results | Participants | Proceedings | Input | Summary | Appendix

  • Run ID: IVORY.90.10
  • Participant: UMD
  • Track: Web
  • Year: 2010
  • Submission: 9/15/2010
  • Task: spam
  • MD5: d50b7889fc90d7b5736d6f5254910e6b
  • Run description: A linear combination of TrustRank and Anti-TrustRank scores, seeded with Waterloo Spamminess scores thresholded at 90 and 10 respectively.

IvoryBM25a

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: IvoryBM25a
  • Participant: UMD
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: f563713943c90b11929bc63c0a26daff
  • Run description: An ad-hoc by Ivory, a Hadoop toolkit for web-scale information retrieval research, that ranks documents using Okapi BM25, and leverages Waterloo spaminess scores as document priors.

IvoryL2Rb

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: IvoryL2Rb
  • Participant: isi
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: 75cacbb726d050ce021dcb1a06e60783
  • Run description: Category B -- linear machine learned ranking function (features = bm25 and language modeling term + proximity scores, spam score, anti-trust score, and pagerank)

IvoryLCEb

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: IvoryLCEb
  • Participant: isi
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: 838ab38ca6043686f6bf8e7355c6f305
  • Run description: Category B -- weighted sequential dependence model + spam score + latent concept expansion

IvorySDa

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: IvorySDa
  • Participant: UMD
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: fc99db34d0a7f3a01185891b4f2803f7
  • Run description: An ad-hoc by Ivory, a Hadoop toolkit for web-scale information retrieval research, that ranks documents using the Markov Random Field (MRF) Sequentional Dependence (SD) model with BM25 features, and leverages Waterloo spaminess scores as document priors.

IvoryWSDa

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: IvoryWSDa
  • Participant: UMD
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: 2f8b9ea124409476f8490461be06571f
  • Run description: An ad-hoc by Ivory, a Hadoop toolkit for web-scale information retrieval research, that ranks documents using the Markov Random Field (MRF) Weighted Sequential Dependence (WSD) model with BM25 features, and leverages Waterloo spaminess scores as document priors. The model is learned using TREC Web09 ad-hoc queries and their relevance judgments.

IvoryWSDb

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: IvoryWSDb
  • Participant: isi
  • Track: Web
  • Year: 2010
  • Submission: 8/7/2010
  • Task: adhoc
  • MD5: f7c26774791ae32f3e87a3bf2cb6ab03
  • Run description: Category B -- weighted sequential dependence model + spam score.

MF1

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: MF1
  • Participant: Mediafutures
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 3a6410b89ed192b9a5e76377dcb00cec
  • Run description: Run based on BM25.

MF2

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: MF2
  • Participant: Mediafutures
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 259ffd0ca199a179a9ab1bbef42d9268
  • Run description: Run based on Language modeling.

MMCIl410m1

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: MMCIl410m1
  • Participant: MMCI
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: d9cb69d2e048c6123114d67b49feb0f9
  • Run description: This run aims at providing the retrieval quality of BM25 scores (i.e. similar early precision quality) and at the same time efficient query processing, still considering index size. This run uses two kinds of lists: a) text index lists (short: TLs) where each list stores, for a single stemmed term t, an entry of the form (d.docid; scoreBM25(d; t)) for each document d where this term occurs (d.docid is a unique numerical id for document d), ordered by descending scoreBM25 (, i.e. BM25 score with b=0.5, k1=1.2). b) combined index lists (short: CLs) where each list contains, for a single stemmed term pair (t1; t2), an entry of the form (d.docid; proxprime(d; t1; t2); scoreBM25(d; t1); scoreBM25(d; t2)) for each document where this term pair occurs within a window of 10 terms in the document. The resulting proximity contribution of (t1;t2) for d is stored in proxprime(d; t1; t2). Each index list is ordered by descending proxprime contribution. Pruning proceeds in two dimensions: While both TLs and CLs materialize the first l entries (pruning by list length), CLs in addition impose a minimum score requirement for proxprime (pruning by minimum score). The indexes are tuned for efficient evaluation but still aim at providing BM25 result quality with a maximum index size of 1 TB. For tuning we used the 50 Web Track 2009 topics and considered the average overlap of the top10-results at many pruning levels with the top10-results of the unpruned TL+CL evaluation. As we aim at efficient evaluation, we select the run with the lowest list length providing at least 75% overlap. A list length of 410 with a minimum score requirement of 1.0 for the proximity contribution proxprime qualified. To allow for efficient query processing, this run uses pruned TL and CL indexes that are reordered by docid then. The pruned indexes are the input to a merge join implementation. The input lists consist of pruned and reordered TLs for all stemmed query terms as well as pruned and reordered CLs for all pairs of stemmed query terms. Stopwords are removed. We exclude the 50% spammiest documents according to the Waterloo Fusion spam score.

MMCITLCLl20M

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: MMCITLCLl20M
  • Participant: MMCI
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: ac8ab8f057f00b5c271a75781e19cb87
  • Run description: This run uses two kinds of lists: a) text index lists (short: TLs) where each list stores, for a single stemmed term t, an entry of the form (d.docid; scoreBM25(d; t)) for each document d where this term occurs (d.docid is a unique numerical id for document d), ordered by descending scoreBM25 (, i.e. BM25 score with b=0.5, k1=1.2). b) combined index lists (short: CLs) where each list contains, for a single stemmed term pair (t1; t2), an entry of the form (d.docid; proxprime(d; t1; t2); scoreBM25(d; t1); scoreBM25(d; t2)) for each document where this term pair occurs within a text window of 10 terms in the document. The resulting proximity contribution of (t1;t2) for d is stored in proxprime(d; t1; t2). Each index list is ordered by descending proxprime contribution. This run uses index lists pruned to the first 20M entries as input to an NRA implementation. The input lists consist of TLs for all stemmed query terms and CLs for all pairs of stemmed query terms. Stopwords are removed. We exclude the 50% spammiest documents according to the Waterloo Fusion spam score.

MMCITLl20M

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: MMCITLl20M
  • Participant: MMCI
  • Track: Web
  • Year: 2010
  • Submission: 8/6/2010
  • Task: adhoc
  • MD5: ae2fba758e6f9e301f7ebb87f61f24fa
  • Run description: This run uses text index lists (short: TLs) where each list stores, for a single stemmed term t, an entry of the form (d.docid; scoreBM25(d; t)) for each document d where this term occurs (d.docid is a unique numerical id for document d), ordered by descending scoreBM25 (, i.e. BM25 score with b=0.5, k1=1.2). This run uses index lists pruned to the first 20M entries as input to an NRA implementation. The input lists consist of TLs for all stemmed query terms. Stopwords are removed. We exclude the 50% spammiest documents according to the Waterloo Fusion spam score.

msrsv1

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: msrsv1
  • Participant: msrsv
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 84e0c10469a6265756cb04ded7779eca
  • Run description: MS_1xBM25F_15xMAC-Fixed_5xlogSpamScore-Fusion_100xlogsalsa-setr-5-6-6200-2900-aut_16xspanscore Additional resource: IP address info used to collapse anchors.

msrsv1div

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: msrsv1div
  • Participant: msrsv
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 84e0c10469a6265756cb04ded7779eca
  • Run description: MS_1xBM25F_15xMAC-Fixed_5xlogSpamScore-Fusion_100xlogsalsa-setr-5-6-6200-2900-aut_16xspanscore Additional resource: IP address info used to collapse anchors.

msrsv2

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: msrsv2
  • Participant: msrsv
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 541be77ae9cc50a2fe85fd59ca61a7d3
  • Run description: MS_1xBM25F_15xMAC-Fixed-URI_5xlogSpamScore-Fusion_100xlogsalsa-setr-5-6-6200-2900-aut_16xspanscore Additional resource: IP address info used to collapse anchors.

msrsv2div

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: msrsv2div
  • Participant: msrsv
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 541be77ae9cc50a2fe85fd59ca61a7d3
  • Run description: MS_1xBM25F_15xMAC-Fixed-URI_5xlogSpamScore-Fusion_100xlogsalsa-setr-5-6-6200-2900-aut_16xspanscore Additional resource: IP address info used to collapse anchors.

msrsv3

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: msrsv3
  • Participant: msrsv
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 561d06653f52652a66fdd2c7324f53cc
  • Run description: MS_1xBM25F_15xMAC-Fixed_5xlogSpamScore-Fusion_100xlogsalsa-setr-5-6-6200-2900-aut Additional resource: IP address info used to collapse anchors.

msrsv3div

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: msrsv3div
  • Participant: msrsv
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 561d06653f52652a66fdd2c7324f53cc
  • Run description: MS_1xBM25F_15xMAC-Fixed_5xlogSpamScore-Fusion_100xlogsalsa-setr-5-6-6200-2900-aut Additional resource: IP address info used to collapse anchors.

pkusewm1

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: pkusewm1
  • Participant: PKUSEWM
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: e99a754c2c7f183f031f9570a733c434
  • Run description: No additional resource used.This is extension of bm25.

qirdcsuog1

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: qirdcsuog1
  • Participant: qirdcsuog
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: d44e2e771c19d678404e9d3e68fbcabd
  • Run description: Reranking over dirichlet language models results of a clueweb A run. No document prior is used. Spam filtering is performed before indexing.

qirdcsuog2

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: qirdcsuog2
  • Participant: qirdcsuog
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 660d4de50ddc6747391f55af0e73ae6d
  • Run description: Reranking over dirichlet language models results of a clueweb A run. No document prior is used. Spam filtering is performed before indexing.

qirdcsuog3

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: qirdcsuog3
  • Participant: qirdcsuog
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 296b481e22deeef47d93dd4d13084c39
  • Run description: Reranking over standard bm25 results of a clueweb B run. Spam filtering is performed before indexing.

sztaki1

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: sztaki1
  • Participant: budapest_acad
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: a747673f398719dc2651b3c3481e35ae
  • Run description: First run. We did not use any additional resource.

THUIR10DvNov

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: THUIR10DvNov
  • Participant: THUIR
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 987bf88ca9602313af857aa4683875b0
  • Run description: Diversifying results with document novelty detection. Novelty is calculated based on harmonic-mean of similarities with previous results. With duplicated page panelty. Baseline is THUIR10Qa for ad hoc task.

THUIR10DvQaH

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: THUIR10DvQaH
  • Participant: THUIR
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 595c2bc05aba27c901b9009b13f9fd86
  • Run description: Use ad hoc result THUIR10QaHt (but with a different option on exact matching between site-level anchor text and query). with Duplicated page panelty.

THUIR10DvQEW

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: THUIR10DvQEW
  • Participant: THUIR
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: d2474460257376f801a2e78c5f3e89cb
  • Run description: Subtopic detection by Wikipedia. Query expansion based on subtopics. Full text retrieval on page content combined with site-level anchors, with different weights on both title and anchor fields. Possible spam pages are reduced using the spam list provided by UWaterloo. PageRank is embedded. With duplicated page panelty.

THUIR10Qa

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: THUIR10Qa
  • Participant: THUIR
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 9cdce407cd2a256feb4baf798b4478d5
  • Run description: Full text retrieval on page content combined with site-level anchors, with different weights on both title and anchor fields. Possible spam pages are reduced using the spam list provided by UWaterloo. Result list is also reranked with both PageRank and the exact matching between page-level anchor text and query content.

THUIR10QaHt

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: THUIR10QaHt
  • Participant: THUIR
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: bea0a1470ad176b99f73d2316b747b7b
  • Run description: Selected result re-ranking with HITS by top hub or authority page set based on THUIR10Qa.

THUIR10Str

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: THUIR10Str
  • Participant: THUIR
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 567157679df60ba3903182f1f385ae7f
  • Run description: Full text retrieval on page content combined with site-level anchors, with different weights on both title and anchor fields. Possible spam pages are reduced using the spam list provided by UWaterloo. PageRank is embedded. Result re-ranking with full query string match on the full text and page-level anchor text.

UAMSA10d2a8

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UAMSA10d2a8
  • Participant: UAmsterdam
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: ded71521df0f5803f8d5d988c7023134
  • Run description: Mixture of document and anchor-text runs with a linear length prior probability for both document and anchor-text representations. Scores are combined 0.2 document score + 0.8 anchor-text score.

UAMSA10mSF30

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UAMSA10mSF30
  • Participant: UAmsterdam
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: d21dd76a978ad4165b8971d920d33e1c
  • Run description: Combination of document and anchor-text runs with linear length priors for document and anchor-text representations. Scores combined as 0.20 document score + 0.80 anchor-text score. Results are post-filtered on spam using the Waterloo spam rankings (tresholded at 30% spammiest pages).

UAMSA10mSFPR

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UAMSA10mSFPR
  • Participant: UAmsterdam
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 662112e8de38720cd4d7104c401e19e5
  • Run description: Mixture of document and anchor-text runs with linear length priors on document and anchor-text representation. The mixture run is re-ranked on PageRank scores provided by CMU and spam-filtered using the Waterloo spam rankings with a threshold on the 30th percentile.

UAMSD10ancB

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UAMSD10ancB
  • Participant: UAmsterdam
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 42cfdf57d305110bc0ad852bb045d748
  • Run description: Anchor-text run with linear length prior on anchor-text representation using category B.

UAMSD10ancPR

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UAMSD10ancPR
  • Participant: UAmsterdam
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 43be2f894286d6bf6c395041efe601ad
  • Run description: Anchor-text run with linear length prior on the anchor-text representation. Results are re-ranked using PageRank scores provided by CMU.

UAMSD10aSRfu

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UAMSD10aSRfu
  • Participant: UAmsterdam
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 24f4e2a7cc20d9e37cb66b5b7c8a3b22
  • Run description: Anchor-text run with linear length prior on the anchor-text representation. Results are re-ranked using the spam percentiles provided by Waterloo (Fusion method).

UCDSIFTDiv

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UCDSIFTDiv
  • Participant: UCDSIFT
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 085a36fd7fa00a49537fde5132a47312
  • Run description: Data Fusion using SlideFuse using runs from Terrier and Indri as inputs.

UCDSIFTMAP

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UCDSIFTMAP
  • Participant: UCDSIFT
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 977ae3ef86f66e69a2c1b9ea71bc4227
  • Run description: Data Fusion using MAPFuse using runs from Terrier and Indri as inputs.

UCDSIFTProb

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UCDSIFTProb
  • Participant: UCDSIFT
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 9057fb957cbbac9d4500e6beebab6b90
  • Run description: Data Fusion using ProbFuse using runs from Terrier and Indri as inputs.

UCDSIFTSlide

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UCDSIFTSlide
  • Participant: UCDSIFT
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 01158eabb5188a589fd831c67cc8525c
  • Run description: Data Fusion using SlideFuse using runs from Terrier and Indri as inputs.

udelCluster

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: udelCluster
  • Participant: udel
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: b1f8b1624f4d81684898e0a4fc11cd45
  • Run description: clustering for diversity

udelFMRM

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: udelFMRM
  • Participant: udel
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 6e6a2f5fd7c331a079863941ed013c4b
  • Run description: facet model

udelIndriB

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: udelIndriB
  • Participant: udel
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: 0ba97afa0da05fefc3b3b4100bb1f244
  • Run description: Metzler+Croft dependence model run

udelIndriB2

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: udelIndriB2
  • Participant: udel
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: ad7c1179b1cceb1cb283dc5105a897c4
  • Run description: more DM

udelIndriWP

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: udelIndriWP
  • Participant: udel
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: b255e7059eeec87c588a95f7af03ac88
  • Run description: Metzler+Croft dependence model run, wikipedia docs only

udelSimPrune

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: udelSimPrune
  • Participant: udel
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 5bc8b1693a0cac2d4ca2d2f57618f7a4
  • Run description: document similarity pruning for diversity

UDWebLOG

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UDWebLOG
  • Participant: Udel_Fang
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: d301c207bc2082d7b641b1f113faf2b7
  • Run description: We use the diversification retrieval function LOG derived from our diversification framework that formally defines the diversification problem as an optimization problem. The subtopics are extracted by the proposed frequent pattern mining algorithm.

UDWebPCOV

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UDWebPCOV
  • Participant: Udel_Fang
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: 3dc60290a82b85ddd00691ed59e7a13c
  • Run description: We use the diversification retrieval function PCOV derived from our diversification framework that formally defines the diversification problem as an optimization problem. The subtopics are extracted by the proposed frequent pattern mining algorithm.

UDWebSQR

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UDWebSQR
  • Participant: Udel_Fang
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: diversity
  • MD5: a4b98c3095f6ae85f3ea0921c9ce673c
  • Run description: We use the diversification retrieval function SQR derived from our diversification framework that formally defines the diversification problem as an optimization problem. The subtopics are extracted by the proposed frequent pattern mining algorithm.

UMa10BASF

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UMa10BASF
  • Participant: unimelb
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 2d0f221970dd823d1172bd7064ab02e9
  • Run description: BM25 on impact, with anchor, with spam filtering

UMa10BSF

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UMa10BSF
  • Participant: unimelb
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: b64be5d1c6d3f80ef9b1193b144b9ac5
  • Run description: BM25 on impact, no anchor, with spam filtering

UMa10IASF

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UMa10IASF
  • Participant: unimelb
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 8de62f510d8c3a26c2157039767852ce
  • Run description: Original impact model, with anchor, with spam filtering

umassSDM

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: umassSDM
  • Participant: umass
  • Track: Web
  • Year: 2010
  • Submission: 8/5/2010
  • Task: adhoc
  • MD5: 18306182231e9fe3edbf195768d45d79
  • Run description: This run is performed using the Indri search engine. This run employs the standard sequential dependence model (Metzler & Croft, 2005).

umassSDMW

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: umassSDMW
  • Participant: umass
  • Track: Web
  • Year: 2010
  • Submission: 8/5/2010
  • Task: adhoc
  • MD5: 8fc2a6194940b797f605112aa5396abf
  • Run description: This run is performed using the Indri search engine. This run promotes Wikipedia pages.

umasswfb520

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: umasswfb520
  • Participant: umass
  • Track: Web
  • Year: 2010
  • Submission: 8/5/2010
  • Task: adhoc
  • MD5: 8a27cbb134c50b73f2cd0be025b7586a
  • Run description: This run is performed using the Indri search engine. This is run employs pseudo-relevance feedback.

UMd10ASF

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UMd10ASF
  • Participant: unimelb
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: bd6abb98ece003e1a6c65b8a23efe423
  • Run description: anchor-only run

UMd10BASF

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UMd10BASF
  • Participant: unimelb
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 8a5450af8d9140915182eda59a3c5227
  • Run description: BM25 on impact, with anchor

UMd10IASF

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: UMd10IASF
  • Participant: unimelb
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 53965cd1f4f6fb4fd6a1702ae2eeafb6
  • Run description: Original impact, with anchor

uogTrA40n

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uogTrA40n
  • Participant: uogTr
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: fc8c7ac966cf2d72d4000871063e7ab9
  • Run description: Learned model including various Divergence from Randomness and other features on navigational queries.

uogTrA42

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uogTrA42
  • Participant: uogTr
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: f4b19c8961fc51d095c58d7b02c53aaf
  • Run description: Learned model including various Divergence from Randomness and other features.

uogTrA42x

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uogTrA42x
  • Participant: uogTr
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 27d9f97884c9dc0553641b3e9454de78
  • Run description: Uniform diversification with learning support.

uogTrB67

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uogTrB67
  • Participant: uogTr
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: d45687a6b5cad1992350bbd506767818
  • Run description: Learned model including various Divergence from Randomness and other features.

uogTrB67LTS

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uogTrB67LTS
  • Participant: uogTr
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: adhoc
  • MD5: 9951b2f9246390322775602c210523cd
  • Run description: A novel, automatic ranking function selection, combining several learned models that each have many features, such as Divergence from Randomness models.

uogTrB67xS

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uogTrB67xS
  • Participant: uogTr
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 378c045b9365bdefd0726d792f2106a3
  • Run description: Selective diversification with learning support.

uogTrBdphxS

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uogTrBdphxS
  • Participant: uogTr
  • Track: Web
  • Year: 2010
  • Submission: 8/10/2010
  • Task: diversity
  • MD5: 0529e02ed69cd8de124f56952e401c9b
  • Run description: Selective diversification without learning support.

utwente3

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: utwente3
  • Participant: utwente
  • Track: Web
  • Year: 2010
  • Submission: 8/5/2010
  • Task: adhoc
  • MD5: 10143fc839ae60c1417ad71105e45764
  • Run description: Retrieval based on anchor text; language modeling approach with minimal smoothing.

utwente4

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: utwente4
  • Participant: utwente
  • Track: Web
  • Year: 2010
  • Submission: 8/5/2010
  • Task: adhoc
  • MD5: 208a6f4f08fe08fbbd4e78bc444168d3
  • Run description: Retrieval based on anchor text and titles, language modeling approach with minimal smoothing.

utwente4SF

Results | Participants | Proceedings | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: utwente4SF
  • Participant: utwente
  • Track: Web
  • Year: 2010
  • Submission: 8/5/2010
  • Task: adhoc
  • MD5: 9c09359a6f25d9afd8a3cd69e29ef04e
  • Run description: Retrieval based on anchor text and titles; language modeling approach with minimal smoothing. Spam filtering (University of Waterloo's spaminess scores) as a postprocessing step.

uwgym

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: uwgym
  • Participant: uwaterlooclarke
  • Track: Web
  • Year: 2010
  • Submission: 8/4/2010
  • Task: diversity
  • MD5: 748c9314043d5d7750bffede28d72e09
  • Run description: This run has been generated based on commercial search engine results.

york10wA1

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: york10wA1
  • Participant: york
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: e082b6ead7b26194bddf50d1f8f6b68a
  • Run description: This run is based on the classic Okapi BM25 model. The parameter b is set to default value 0.75, and the parameter k1 is 1.2. 10000 documents are retrieved for each topic. The whole html part in each document is used for the retrieval process. No query expansion is utilized.

york10wA2

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: york10wA2
  • Participant: york
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: cd6c04ac7c69f5c83e4d6ea3f6ef9ea9
  • Run description: This run is based on the Dirichlet language model which uses Bayesian smoothing with Dirichlet Prior. This has one parameter, mu. It is set to 2500 in this run. 10000 documents are retrieved for each topic. The whole html part in each document is used for the retrieval process. No query expansion is utilized.

york10wA3

Results | Participants | Input | Summary (trec_eval) | Summary (ndeval) | Summary (gdeval) | Appendix

  • Run ID: york10wA3
  • Participant: york
  • Track: Web
  • Year: 2010
  • Submission: 8/9/2010
  • Task: adhoc
  • MD5: dcd4a5d783e2a530ef3209d4656c2212
  • Run description: This run is based on a variant of TFIDF model, Lemur TFIDF weighting model. It applies the Okapi TF formula from the BM25 model. Its b parameter is set to 0.75, and k1 parameter is 1.2. 10000 documents are retrieved for each topic. The whole html part in each document is used for the retrieval process. No query expansion is utilized.