Skip to content

Runs - Legal 2008

adhocpool08

Results | Participants | Proceedings | Appendix

  • Run ID: adhocpool08
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 9/10/2008
  • Task: interactive
  • MD5: 5ad7473cf5f101923ab38a46b6c525c9
  • Run description: This run is the full depth-100000 pool of all submitted Ad Hoc runs for topics 102, 103 and 104, including the reference Boolean runs (but not including the randomL08 run).

Clearwell08

Results | Participants | Appendix

  • Run ID: Clearwell08
  • Participant: Clearwell
  • Track: Legal
  • Year: 2008
  • Submission: 9/11/2008
  • Task: interactive
  • MD5: 6b6dcf739d4d1481f5652beac8c3b423
  • Run description: We used several techniques for mining the data. The CDIP collection was processed using Perl and other Unix scripts to cleanse the content. We then used Clearwell E-Discovery Platform and other third-party eDiscovery packages to extract documents based on keywords, boolean search, fuzzy search and concept search. We then used iterative searching techniques driven by human review. Quality control was based on a sample (95% confidence level) with review of both the non-responsive collection and responsive collection, followed by native export. The final stage was to consolidate and clean the submission using another set of Unix scripts and Perl scripts.

CTFgge10kBr0

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFgge10kBr0
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: adhoc
  • MD5: 16b1ee4db78f30ec1fb70dc128da93a2
  • Run description: We submit automatically constructed weighted queries obtained indexing the following topic fields - request text - initial proposal by defendant - rejoinder by plaintiff - final negotiated Boolean query ... and appling a semi-automatically built stop word list. K constant (k=100.000).

CTFgge4kBr0

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFgge4kBr0
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: adhoc
  • MD5: 8180c03927feb11c68b4339b707241b0
  • Run description: We submit automatically constructed weighted queries obtained indexing the following topic fields - request text - initial proposal by defendant - rejoinder by plaintiff - final negotiated Boolean query ... and appling a semi-automatically built stop word list. K constant (k=40.000) with boolean re-rank.

CTFggeBkBr0

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFggeBkBr0
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: adhoc
  • MD5: 4854936a6becba1daf089a7de7ae173b
  • Run description: We submit automatically constructed weighted queries obtained indexing the following topic fields - request text - initial proposal by defendant - rejoinder by plaintiff - final negotiated Boolean query ... and appling a semi-automatically built stop word list. K based on B values, with boolean re-rank.

CTFggeBkBr1

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFggeBkBr1
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: adhoc
  • MD5: 135f1ade73826b73a2f00baa61186f1f
  • Run description: We submit automatically constructed weighted queries obtained indexing the following topic fields - request text - initial proposal by defendant - rejoinder by plaintiff - final negotiated Boolean query ... and appling a semi-automatically built stop word list. K based on B values, with a different factor for boolean re-rank.

CTFggeRkBr0

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFggeRkBr0
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: adhoc
  • MD5: e784cc7478ecad9fc12bc5a8d0d82ce9
  • Run description: We submit automatically constructed weighted queries obtained indexing the following topic fields - request text - initial proposal by defendant - rejoinder by plaintiff - final negotiated Boolean query ... and appling a semi-automatically built stop word list. K=B with boolean re-rank.

CTFggeSkBr0

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFggeSkBr0
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: adhoc
  • MD5: 2982dd22a6c364f1abd83c6c8a3de22c
  • Run description: We submit automatically constructed weighted queries obtained indexing the following topic fields - request text - initial proposal by defendant - rejoinder by plaintiff - final negotiated Boolean query ... and appling a semi-automatically built stop word list. K computed also considering best values of K at Legal track 2007, with boolean re-rank.

CTFrtSk

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFrtSk
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 1693884962779f6b4878b6180483ce25
  • Run description: Baseline using the Terrier Framework with a parameter free IR model.

CTFrtSkBr0

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: CTFrtSkBr0
  • Participant: WIR
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 542bcf729352839f320a83c47af49834
  • Run description: We re-rank the baseline using the boolean reference run.

H52008

Results | Participants | Proceedings | Appendix

  • Run ID: H52008
  • Participant: H5_2008
  • Track: Legal
  • Year: 2008
  • Submission: 9/12/2008
  • Task: interactive
  • MD5: 39f00907cc9c65eb283872d27fe5d8bc
  • Run description: H5 has submitted an assessment of documents our system has identified as responsive to topic 103 of TREC's Legal Track Interactive Task a total of 608807 documents. The H5 system combined human expertise and advanced technologies to assess the totality of the corpus under investigation.

IowaSL0804

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL0804
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: fda369ec9dd1917f3bf6efd91d71701f
  • Run description: The source fields are converted to a term vector and submitted as a query. Five terms are extracted from the top three documents returned from an initial query. These terms are then added to the term vector for the final query. Documents are given a score and ranked as per IowaSL08Ref.

IowaSL0804b

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL0804b
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 97d33ba8e222dfe684c4791284e34dfd
  • Run description: Documents are retrieved and ranked in the same manner as IowaSL0804. Additionally, documents that were also found by the Boolean Reference run provided by the track organizers are given a boost to their score, and the set is then reranked. The boost, a multiplier of 1.8, was found through training on 2007 results. K-values are taken from B, and Kh is B/2.

IowaSL0805

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL0805
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 79f97ce510a5aadb16b54d47314622cb
  • Run description: Documents are retrieved and ranked as in IowaSL0804 except that, prior to being used in initial or final queries, the Request Text is automatically modified to remove terms that appear in too many instances of this field across all the topics.

IowaSL0805b

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL0805b
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 448e2f145c3fcbbdd69f8462757c662a
  • Run description: Documents are retrieved and ranked in the same manner as IowaSL0804. Additionally, documents that were also found by the Boolean Reference run provided by the track organizers are given a boost to their score, and the set is then reranked. The boost, a multiplier of 1.5, was found through training on 2007 results. K-values are taken from B, and Kh is B/2.

IowaSL0808b

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL0808b
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: b7185230ba699b7065fbfb2b69a1ab16
  • Run description: Documents are retrieved and ranked in the same manner as IowaSL0804b, except that the entirety of the negotiation history is also included as a source of terms for the query term vectors.

IowaSL0808m2

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL0808m2
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 0ed1eaa5904016a3d687f58d1c1a585d
  • Run description: The documents submitted for this run were found by fusing the results from IowaSL0804b and IowaSL0805b. The runs were weighted 1.0 and 0.755 respectively, in accordance to the relative performance of similar runs submitted to the TREC 2007 Legal Track ad hoc task, using the F1@R measure.

IowaSL0808m3

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL0808m3
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 032c60e02a592349e5962929d2d09480
  • Run description: The documents submitted for this run were found by fusing the results from IowaSL0804b, IowaSL0805b, and IowaSL0808b. The runs were weighted 2.0, 1.0, and 3.0 respectively. The rationale for the weighting is to grant more weight to runs with more information from the topic source fields contributing to the run queries.

IowaSL08Ref

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: IowaSL08Ref
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 59f71590f2b685ed6abad7a330cc4ae5
  • Run description: This is the requested standard run. The source field is converted to a term vector and submitted as a query. Documents are given a score according to the Okapi-BM25 function with parameters as described in our paper for the TREC Legal Track 2007; documents are ranked based on this score. K-values are taken from B, and Kh is B/2.

IowaSL08RF1A

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RF1A
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 132813034fe23d7e70cc41129da59b56
  • Run description: A collection of features is calculated for all documents submitted in the source runs for IowaSL08RF1B, judged and unjudged. The features describe the position of the document in the contributing runs, as well as the relevance characteristics of its "neighborhood". We train a classifier using the calculated features of the judged documents. The classifier, implemented using Wekas SMO package, fits a logit regression to generate probabilities for relevance and nonrelevance for each of the test documents. It is then applied to the unjudged, which are then ranked by estimated probability of relevance.

IowaSL08RF1B

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RF1B
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 17bac6e99ccf633548ddec35d724ca84
  • Run description: Each submitted run for the previous ad hoc tasks were pooled to build a collection of judged and unjudged documents. The unjudged documents are merged, ranked using a score reflecting a summation of the ranks across the runs and a multiplier reflecting the number of runs to select the document (CombMNZ). For the 2007 topics, the 101000 documents with the best scores are selected.

IowaSL08RF2A

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RF2A
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: affe1e85ef4bf2c04b1ad257fb857d45
  • Run description: This run is processed in the same manner IowaSL08RF1A. The documents pooled for IowaSL08RF2B are split into training and evaluation pools in the same manner as described for IowaSL08RF1A.

IowaSL08RF2B

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RF2B
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 3c8419f735a35bd3fb48e4be64dc4a9a
  • Run description: This run is processed in the same manner IowaSL08RF1B, the difference laying in the runs contributing to the pool of judged and unjudged documents. The seven runs submitted by the IowaS group for the TREC Legal 2007 Ad Hoc task are used for the subset of 2007 topics. Results generated Post TREC 2007 by applying 2007 strategies against the 2006 topics are used as sources for the 2006 topics.

IowaSL08RF2C

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RF2C
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: c97ebf13229b91be3f25f5d0dc184c39
  • Run description: This run is also processed in the same manner IowaSL08RF2A. The documents pooled for IowaSL08RF2B are split into training and evaluation pools in the same manner as described for IowaSL08RF1A. For this run, the training pool is discarded, and the classifiers built using all submitted runs, for run IowaSL08RF1A, are used for classifying and ranking the evaluation pool.

IowaSL08RF3A

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RF3A
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 785155108a9e6ace99b0361f5890d7bf
  • Run description: This run is also processed in the same manner IowaSL08RF1A. The documents pooled for IowaSL08RF3B are split into training and evaluation pools in the same manner as described for IowaSL08RF1A.

IowaSL08RF3B

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RF3B
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: d20054ef308dbabb2a6d549e014b1b32
  • Run description: This run is processed in the same manner IowaSL08RF1B, the difference laying in the runs contributing to the pool of judged and unjudged documents. For 2006 topics, four runs with leading performance on the R-Prec measure across all topics were selected, and four runs with poorer performance were selected. A similar technique was used to identify contributing runs for the 2007 topics, using the Est_Recall@B measure.

IowaSL08RFTr

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: IowaSL08RFTr
  • Participant: UIowa-Srinivasan
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: automatic
  • Task: feedback
  • MD5: 386d6789b26d3a85addf0d2637229e63
  • Run description: This run is run in the same manner as our ad hoc run IowaSL0804 with the following modifications to take advantage of relevance feedback. Documents that have not been judged to be relevant are deemed ineligible for providing terms for query expansion. Training has allowed us to expand the number of query expansion terms to 12 from up to 10 relevant documents. The boolean reference run boost has not been applied. K and Kh values are from R and R/3 respectively.

otL08db

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08db
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: adhoc
  • MD5: c9b60e4a20bf8a7c5998a4858fc2c6dd
  • Run description: a ranked Boolean run based on the initial proposal by defendant (same as otL08fb except that the defendant Boolean query was used instead of the final query)

otL08fb

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08fb
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: manual
  • Task: adhoc
  • MD5: 6ef71be6eacef02259aa4b77b8afc5e4
  • Run description: Boolean run based on final negotiated query (same as reference Boolean run refL08B except re-ordered by relevance)

otL08fbe

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08fbe
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/2/2008
  • Type: manual
  • Task: adhoc
  • MD5: a21ab560ea0652a647bdedaf3e577228
  • Run description: blind feedback expansion of the final Boolean query, ranked-based fusion, weight 2 on original otL08fb query and weight 1 on each of first 2 retrieved rows of otL08fb as query; K includes weight sum of 2 or more, Kh includes weight sum of 3 or more

otL08frw

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08frw
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/2/2008
  • Type: manual
  • Task: adhoc
  • MD5: d98d678cd56d844d0605589510f1df1e
  • Run description: weighted rank-based fusion of Boolean and vector runs, weight 3 on otL08fb, weight 2 on otL08rvl, weight 1 on otL08fv; K includes weight sum of 3 or more, Kh includes weight sum of 4 or more

otL08fv

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08fv
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/2/2008
  • Type: manual
  • Task: adhoc
  • MD5: 74d4406d7ea8ea506851e244955be185
  • Run description: vector run based on final negotiated query (same as otL08fb except that all Boolean, phrase and proximity operators were changed to the OR operator); K includes rsv of 200 or more, Kh includes rsv of 225 or more

otL08rv

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08rv
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/3/2008
  • Type: manual
  • Task: adhoc
  • MD5: fee085b46516f66c234b53b1d600b7a9
  • Run description: vector run based on request text terms (same as otL08rvl except no stemming, i.e. English inflections were not matched)

otL08rvl

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08rvl
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/2/2008
  • Type: manual
  • Task: adhoc
  • MD5: 68d7173eaff1232ac3a5b3c333a64ac3
  • Run description: vector run based on request text terms (same as otL08fv except (1) the terms were taken from the request text field instead of the final Boolean field, (2) English inflections were matched, (3) common instruction words (e.g. "please", "produce", "documents") were manually removed); K includes rsv of 200 or more, Kh includes rsv of 225 or more

otL08rvlq

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: otL08rvlq
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/3/2008
  • Type: manual
  • Task: adhoc
  • MD5: ebf92d2a38be57347de074bdd10feb2a
  • Run description: vector run based on request text terms (same as otL08rvl except that common instruction words (e.g. "please", "produce", "documents") were not removed)

otRF08F

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08F
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/5/2008
  • Type: manual
  • Task: feedback
  • MD5: 65a95358d3f60713b7e1909da0d67b0b
  • Run description: pure relevance feedback run based on forming a query from the known relevant documents (excluding those larger than 1 million bytes); no topic fields were used; K includes rsv of 200 or more, Kh includes rsv of 225 or more

otRF08fb

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08fb
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/4/2008
  • Type: manual
  • Task: feedback
  • MD5: 8eb6ecc299ff67700754dcda997bd176
  • Run description: Boolean run based on final negotiated query (same as reference Boolean run refRF08B except re-ordered by relevance)

otRF08fbF

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08fbF
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/5/2008
  • Type: manual
  • Task: feedback
  • MD5: 3f1f5049db57bf467db3259551e635a5
  • Run description: ranked-based fusion of Boolean (otRF08fb) and feedback (otRF08F) runs, equal weights; K and Kh are the same as for otRF08F

otRF08fbFR

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08fbFR
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: manual
  • Task: feedback
  • MD5: d52e41bbede5c3b6442b99cd72e75297
  • Run description: same documents as otRF08fbF and same K and Kh as otRF08FR

otRF08FR

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08FR
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: feedback
  • MD5: f8dbe07323d710a6c62b5b9e5c77e1b5
  • Run description: same as otRF08F except that K is based on estimated R from humL06tvz sample for first 15 topics and L07 est_rel for latter 25 topics; Kh same as K

otRF08frw

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08frw
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: feedback
  • MD5: 192ea4ffe69b61778acf3497228de674
  • Run description: weighted rank-based fusion of Boolean and vector runs (same approach as otL08frw, i.e. weight 3 on otRF08fb, weight 2 on otRF08rvl, weight 1 on otRF08fv; K includes weight sum of 3 or more, Kh includes weight sum of 4 or more)

otRF08fv

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08fv
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: feedback
  • MD5: 475d6eb7fc07bd1039f11b29d06e1943
  • Run description: vector run based on final negotiated query terms (same approach as otL08fv)

otRF08rvl

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: otRF08rvl
  • Participant: ot
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: feedback
  • MD5: 812cf06ecac5a30b79e4067030817e0b
  • Run description: vector run based on request text terms (same approach as otL08rvl)

PittSIST1

Results | Participants | Proceedings | Appendix

  • Run ID: PittSIST1
  • Participant: Pitt_IR_Team
  • Track: Legal
  • Year: 2008
  • Submission: 9/11/2008
  • Task: interactive
  • MD5: 4433a59396c1580b202aa5ede7f3d3be
  • Run description: We have around 5,000 documents for T102 and around 32,000 documents for T103.

RMITbp1

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: RMITbp1
  • Participant: rmit
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: c5ba3f029132b3f3d72716fb7d18d13b
  • Run description: Boolean ranked search on OCR fields.

RMITbp2

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: RMITbp2
  • Participant: rmit
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: a82f2e85e6eaf5a36ac813f34c9ada57
  • Run description: Boolean ranked search on the processed OCR field (removed HTML entities and soft hyphens).

RMITbp3

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: RMITbp3
  • Participant: rmit
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: f3cc77c26d8305f9000a30677e0f753b
  • Run description: Boolean ranked search on processed OCR fields (removed HTML entities, soft hyphens, and low frequency terms).

RMITrp1

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: RMITrp1
  • Participant: rmit
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: cb115377c82cc319d67fb966649e2310
  • Run description: Request text search on OCR field.

RMITrp2

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: RMITrp2
  • Participant: rmit
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 1bcf6cd0544a01a539e763d23c4e5824
  • Run description: Request text search on the processed OCR field.(removed HTML entities and soft hyphens).

RMITrp3

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: RMITrp3
  • Participant: rmit
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: c41b18dace5e713e0b984417da1f61d3
  • Run description: Request text search on the processed OCR field (removed HTML entities, soft hyphens, and low frequency terms).

SabL08ab1

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: SabL08ab1
  • Participant: sabir.buckley
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: automatic
  • Task: adhoc
  • MD5: f906359b43f3777a7de1d70b57e1c28b
  • Run description: Basic blind feedback, top 30 docs, add 20terms. SMART ltu_Lnu vector run. default constant K. Used all boolean query negotiation terms

SabL08ar2

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: SabL08ar2
  • Participant: sabir.buckley
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: e8df102b8938e5bc5d67ff383714c282
  • Run description: Very basic SMART ltu_Lnu vector run. default constant K.

SabL08arbn

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: SabL08arbn
  • Participant: sabir.buckley
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 624bd96419d0fc229ec42ce9fc969b30
  • Run description: Very basic SMART ltu_Lnu vector run. default constant K. Used all boolean query negotiation terms

SabL08rf1

Results | Participants | Summary (resid) | Summary (residH) | Appendix

  • Run ID: SabL08rf1
  • Participant: sabir.buckley
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: automatic
  • Task: feedback
  • MD5: d1151356dfa14eaa255ad3930301584c
  • Run description: Basic rocchio feedback, all judged docs, add 40terms. SMART ltu_Lnu vector run. default constant K. Used all boolean query negotiation terms

SabL08rfbase

Results | Participants | Summary (resid) | Summary (residH) | Appendix

  • Run ID: SabL08rfbase
  • Participant: sabir.buckley
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: automatic
  • Task: feedback
  • MD5: 9aa0d5dccc9ecf5920c2e81d47e790f2
  • Run description: Baseline for Rocchiofeedback, SMART ltu_Lnu vector run. default constant K. Used all boolean query negotiation terms

ublegal08

Results | Participants | Proceedings | Appendix

  • Run ID: ublegal08
  • Participant: SUNY_Buffalo
  • Track: Legal
  • Year: 2008
  • Submission: 9/12/2008
  • Task: interactive
  • MD5: 7b821d8b3fb3c57f534b171f9dc9570d
  • Run description: We used the online search system of UCSF Legacy Tobacco Documents Library. Our main effort was on query formulation. We tried many queries, read many retrieved documents, and had quite a few clarification emails with the topic authority. Our submitted results contain documents retrieved using four separate Boolean queries. Each of the four queries aims at certain groups of responsive documents, and we feel together they give better search results in terms of both recall and precision. No degree of relevance is implied in the order of the submitted documents because the system is a pure Boolean system.

UCBM25T10Th5

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UCBM25T10Th5
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 622ff82f25f18fadacfcc21e19e22013
  • Run description: BM25 with 10 terms from judged rel docs added to each query (as long as the term weight was at least 5)

UCBM25T5Th5

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UCBM25T5Th5
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: a5a91244ab1114f284c5a2d7ad552fcb
  • Run description: BM25 with automatic rel feedback using top 5 terms from all relevant docs (with term weight > 5)

UCEDLSIa

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UCEDLSIa
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 3166f03fd9bd164fef86dc75c5f35ac1
  • Run description: Distributed EDLSI with lsi_k=41 and x=.2. no folding up.

UCEDLSIb

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UCEDLSIb
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 29ff5174e6bb91aaeb14f737bc948c37
  • Run description: EDLSI run with LSI_k =10 and x=.2

UCPwrT10Th5

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UCPwrT10Th5
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: automatic
  • Task: feedback
  • MD5: bd5210e6a209fcfa33f4c1b8baeae751
  • Run description: Power Norm with auto rel feedback, the top 10 terms from any document judged relevant.

UCPwrT5Th5

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UCPwrT5Th5
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 0f3531e6136b31fb24087fcb766cc37c
  • Run description: Power Norm with 5 terms added to each query. Top 5 weighted terms from any relevant document.

UCRFBM25BL

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UCRFBM25BL
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 17937895390fbc9b268f688f28a6b2b1
  • Run description: BM25 baseline for comparison with BM25T5Th5 and BM25T10Th5

UCRFPwrBL

Results | Participants | Proceedings | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UCRFPwrBL
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: feedback
  • MD5: 7faf4e2dbd3d8876f8e1ded0f2d0c23e
  • Run description: Power Norm baseline for UCPwrT5Th5 and UCPwrT10Th5

UIowa08Leg3

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UIowa08Leg3
  • Participant: UI-ICTS
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: e87a2693e999fa4f42f423914df3778d
  • Run description: Boolean query processed with standard query analyzer. This is then as pseudo-relevance feedback (top 10 documents) to a modified query that includes authors, mentions, recipients and prodbox.

UIowa08Lega

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UIowa08Lega
  • Participant: UI-ICTS
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: f84666013e1c37dbd15fa79f4cc6700c
  • Run description: boolean query processed with standard query analyzer - NOTE - this is actually a corrected version of UIowa08Leg1, which was accidently run with the UIowa08Leg2 configuraiton.

UIowa08LegE0

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UIowa08LegE0
  • Participant: UI-ICTS
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: e0526aac47789a295ebef6d5a31762f2
  • Run description: required request run

UIowa08LegE1

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UIowa08LegE1
  • Participant: UI-ICTS
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 865a4c0a9ddae60c65f34204cbb7d1e5
  • Run description: boolean query processed with standard query analyzer

UIowa08LegE2

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UIowa08LegE2
  • Participant: UI-ICTS
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: f552b7f8998b0dbbeeabdcf61911b714
  • Run description: boolean query processed with custom query analyzer handling complex proximity expressions

UIowa08LegE4

Results | Participants | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UIowa08LegE4
  • Participant: UI-ICTS
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: a8554cec31dee9ad6a5d66568a31d8c8
  • Run description: Boolean query processed with custom query analyzer handling complex proximity expressions. This is then as pseudo-relevance feedback (top 10 documents) to a modified query that includes authors, mentions, recipients and prodbox.

UMDAURCC40

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UMDAURCC40
  • Participant: UMCP
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 70c74fc65c013df438173e7b195760f9
  • Run description: Query construction uses request text, initial proposal by defendent, rejoinder by plaintiff, and final negotiated Boolean query. We did one iteration of blind relevance feedback using metadata expansion. We substituted a portion of the boolean run using a portion of our run.

UMDAURCP3

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UMDAURCP3
  • Participant: UMCP
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 7e33ec12084467061d572c0c89d9a80b
  • Run description: Query construction uses request text, initial proposal by defendent, rejoinder by plaintiff, and final negotiated Boolean query. We did one iteration of blind relevance feedback using metadata expansion. Then combined with boolean results in a hill climbing manner.

UMDCRC40

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UMDCRC40
  • Participant: UMCP
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 9b40afcfda932132a7f3591b4dc0ec9a
  • Run description: Query construction uses request text, initial proposal by defendent, rejoinder by plaintiff, and final negotiated Boolean query. We substituted a portion of the boolean run using a portion of our run. No blind feedback is done.

UMDCRP3

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UMDCRP3
  • Participant: UMCP
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 639326ff1afc7e04cbf604ee2d3cff64
  • Run description: Query construction uses request text, initial proposal by defendent, rejoinder by plaintiff, and final negotiated Boolean query. Combined with boolean results in a hill climbing manner.

UMDSTD

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UMDSTD
  • Participant: UMCP
  • Track: Legal
  • Year: 2008
  • Submission: 8/4/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 995739c28bc39e5af5984a26911606ad
  • Run description: This is the required standard condition run.

UMKCTL08RF1

Results | Participants | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UMKCTL08RF1
  • Participant: UMKCSCE
  • Track: Legal
  • Year: 2008
  • Submission: 8/3/2008
  • Type: manual
  • Task: feedback
  • MD5: 2c21ed265925318161d84cd005b115a5
  • Run description: TL08RF VSM Expand15 Ratio1 SPDL719-719_0.8 101k

UMKCTL08RF2

Results | Participants | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UMKCTL08RF2
  • Participant: UMKCSCE
  • Track: Legal
  • Year: 2008
  • Submission: 8/3/2008
  • Type: manual
  • Task: feedback
  • MD5: fd9d2a008b7e7f6110411b92e6fa61d2
  • Run description: TL08RF BM25 Expand15 Ratio1 s0.9k3 101k K1000

UMKCTL08RF3

Results | Participants | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UMKCTL08RF3
  • Participant: UMKCSCE
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: manual
  • Task: feedback
  • MD5: a18be2eb096c7829d3d2d00655e1d9eb
  • Run description: UMKC TL08RF LM_0.2-1.0 Expand15 Ratio1 101k K1000

UMKCTL08RF5

Results | Participants | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UMKCTL08RF5
  • Participant: UMKCSCE
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: manual
  • Task: feedback
  • MD5: 990669981769c01993081bbb7906ac22
  • Run description: CombMNZ Rank Fusion Normalized Score refRF08B.append

UMKCTL08RF6

Results | Participants | Summary (resid) | Summary (residH) | Appendix

  • Run ID: UMKCTL08RF6
  • Participant: UMKCSCE
  • Track: Legal
  • Year: 2008
  • Submission: 8/7/2008
  • Type: manual
  • Task: feedback
  • MD5: a163ad4e565a143d3d1a0224420f765d
  • Run description: CombMNZ Rank Fusion Normalized Score with refRF08B refRF08B.append

UrsinusBM25a

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UrsinusBM25a
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/2/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 7d7385a29594e795af72b0bc23f79afd
  • Run description: BM25 baseline optimized using TREC 06 and 07 queries for training

UrsinusBM25b

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UrsinusBM25b
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: f0265e47cc86f73eac914d6c42610d8a
  • Run description: Rel feedback BM25 D1 T5 Thres 200

UrsinusPwrA

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UrsinusPwrA
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/4/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 692a6ca61802433bb0c306ea67ac8fb8
  • Run description: Power norm with 2006 optimal values

UrsinusPwrB

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UrsinusPwrB
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/5/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 732f0665c44cc3f1640a07f002f79580
  • Run description: Automated relevance feedback 1 Doc 5 Terms, threshold 1.0. Power Normalization.

UrsinusPwrC

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UrsinusPwrC
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: e487e57d0bd1356f4b98f78bb62d508d
  • Run description: rel feedback power norm top 3 docs, top 5 terms, thres 1.4

UrsinusVa

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: UrsinusVa
  • Participant: Ursinus_College
  • Track: Legal
  • Year: 2008
  • Submission: 8/6/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 89bb1cacfdee5f23e5a19d265af9f889
  • Run description: Vector baseline for comparison with EDLSI runs

uvabase

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: uvabase
  • Participant: UAmsterdam
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: e420b08355ca7d63d70dd9e5307ae4bd
  • Run description: Base Run - Took just the text of RequestText, with a regular stoplist. K and Kh are by our algorithm.

wat1fuse

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat1fuse
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: eb6b975a8bd57b25af340e448bde181f
  • Run description: A fusion of the following 8 runs - boolean - relaxed boolean - okapi on request text field - okapi on request text field, stemming on - okapi using the text in the boolean - 4-gram okapi on request text field - 4-gram okapi using the text in the boolean - 4-gram okapi using words in request text field K was produce using an average of a log and non-log linear regression on scores. Kh is half of K

wat2text

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat2text
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 790b5b911477b31a9fbde30196d7bde3
  • Run description: A fusion of the following 4 runs - okapi on request text field - okapi on request text field, stemming on - 4-gram okapi on request text field - 4-gram okapi using words in request text field K =25000 Kh is half of K

wat3nobool

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat3nobool
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: f362e10df6a492ce2fc5becce5b66395
  • Run description: wat3nobool A fusion of the following 8 runs excluding documents retrieved by the reference boolean - boolean - relaxed boolean - okapi on request text field - okapi on request text field, stemming on - okapi using the text in the boolean - 4-gram okapi on request text field - 4-gram okapi using the text in the boolean - 4-gram okapi using words in request text field K =100000 Kh is half of K

wat4fuse

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat4fuse
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: c8606ff08d205f2abbb4f2335849aa90
  • Run description: A fusion of the following 8 runs - boolean - relaxed boolean - okapi on request text field - okapi on request text field, stemming on - okapi using the text in the boolean - 4-gram okapi on request text field - 4-gram okapi using the text in the boolean - 4-gram okapi using words in request text field K was produce using a linear regression on scores. Kh is half of K

wat5fuse

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat5fuse
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 068560d9f7247403a364ff466b3f1d18
  • Run description: A fusion of the following 8 runs - boolean - relaxed boolean - okapi on request text field - okapi on request text field, stemming on - okapi using the text in the boolean - 4-gram okapi on request text field - 4-gram okapi using the text in the boolean - 4-gram okapi using words in request text field K was produce using a log linear regression on scores. Kh is half of K

wat6fuse

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat6fuse
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: e44065330eced3c7f20449a35428dd37
  • Run description: A fusion of the following 8 runs - boolean - relaxed boolean - okapi on request text field - okapi on request text field, stemming on - okapi using the text in the boolean - 4-gram okapi on request text field - 4-gram okapi using the text in the boolean - 4-gram okapi using words in request text field K = 25000 Kh is half of K

wat7fuse

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat7fuse
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: bfe00d86efdc057f29cc0cf000ad0962
  • Run description: A fusion of the following 8 runs - boolean - relaxed boolean - okapi on request text field - okapi on request text field, stemming on - okapi using the text in the boolean - 4-gram okapi on request text field - 4-gram okapi using the text in the boolean - 4-gram okapi using words in request text field K = 100000 Kh is half of K

wat8fuse

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: wat8fuse
  • Participant: UWIR
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: a819b9044028bc77cf64eb87505ba6a8
  • Run description: A fusion of the following 8 runs - boolean - relaxed boolean - okapi on request text field - okapi on request text field, stemming on - okapi using the text in the boolean - 4-gram okapi on request text field - 4-gram okapi using the text in the boolean - 4-gram okapi using words in request text field K = B Kh is half of K

xb

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: xb
  • Participant: UAmsterdam
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 2d5c9bab8d86c8c2a304b57f2b155673
  • Run description: Extended stoplist - with K constant as B and Kh as B/2

xconst

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: xconst
  • Participant: UAmsterdam
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: e5dde18fec2d3ff2567dba4b15f13c35
  • Run description: Extended stoplist - with K constant as 16904 and Kh constant as k*0.5

xk

Results | Participants | Proceedings | Summary (adhoc) | Summary (adhocH) | Appendix

  • Run ID: xk
  • Participant: UAmsterdam
  • Track: Legal
  • Year: 2008
  • Submission: 7/31/2008
  • Type: automatic
  • Task: adhoc
  • MD5: 2b0bc686faae6cfbdfd4d303ac0bb0ca
  • Run description: We took the request text, parsed by an extended stop list, based on the topics of 2006 and 2007. We used our algorithm for K and Kh.