Runs - Genomics 2006¶

asubaral¶

Run ID: asubaral
Participant: arizona-stateu.gonzalez
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: First complete run after question variants in.

asubaral2¶

Run ID: asubaral2
Participant: arizona-stateu.gonzalez
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Using subject-verb-object as part of ranking together with keyword frequency, distance between keywords.

asubaral3¶

Run ID: asubaral3
Participant: arizona-stateu.gonzalez
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Similar to first run, but less restrictive in filtering. Only require the subject to be in the passage.

Baseline1M¶

Results | Participants | Input | Summary | Appendix

Run ID: Baseline1M
Participant: queenslandu.geva
Track: Genomics
Year: 2006
Submission: 7/17/2006
Type: automatic
Run description: Baseline run, Identify paragraphs

BioKI1¶

Run ID: BioKI1
Participant: concordiau.bergler
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Weighted keyphrases interactively optimized over 2005 data for each query. Output limited to sentence boundaries.

BioKI2¶

Run ID: BioKI2
Participant: concordiau.bergler
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Weighted keyphrases interactively optimized over 2005 data for each query. Output limited to paragraph boundaries.

BioKI3¶

Run ID: BioKI3
Participant: concordiau.bergler
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Weighted keyphrases (weight fixed at 25) interactively optimized over 2005 data for all queries. Output limited to paragraph boundaries.

biotext1¶

Run ID: biotext1
Participant: ucal-berkeley.larso
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Basic run. Returns complete legal spans. Ranking based on Lucene score.

biotext3¶

Run ID: biotext3
Participant: ucal-berkeley.larso
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Ranked

biotextweb¶

Run ID: biotextweb
Participant: ucal-berkeley.larso
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Reranking of the first submission run, using n-grams from the Web.

DUTgen1¶

Run ID: DUTgen1
Participant: dalianu.yang
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Rocchio feedback based on 2005's gold standard, Two levels of indexes, BM25, Paragraph-first reranking

DUTgen2¶

Run ID: DUTgen2
Participant: dalianu.yang
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Rocchio feedback based on 2005's gold standard, Two levels of indexes, BM25, Combining reranking

DUTgen3¶

Run ID: DUTgen3
Participant: dalianu.yang
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Rocchio feedback and SVM based on 2005's gold standard, Two levels of indexes, BM25, Paragraph-first reranking

EMCUT1¶

Run ID: EMCUT1
Participant: erasmus.schuemie
Track: Genomics
Year: 2006
Submission: 7/19/2006
Type: automatic
Run description: Document retrieval is performed using a language-modelling approach. Passage selection is based on identification of concepts from the UMLS metathesaurus and a gene thesaurus in both the query and the documents.

EMCUT2¶

Run ID: EMCUT2
Participant: erasmus.schuemie
Track: Genomics
Year: 2006
Submission: 7/19/2006
Type: manual
Run description: Document retrieval is performed using a language-modelling approach. Passage selection is based on identification of concepts from the UMLS metathesaurus and a gene thesaurus in both the query and the documents. The concepts identified in the query were manually checked and corrected.

fdugen1¶

Run ID: fdugen1
Participant: fudanu.niu
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: manual
Run description: passage retrieval, svm classification.

fdugen2¶

Run ID: fdugen2
Participant: fudanu.niu
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: manual
Run description: passage retireval , svm classification, less positive files

fdugen3¶

Run ID: fdugen3
Participant: fudanu.niu
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: manual
Run description: sentence retrieval, pattern matching.

i2rg061¶

Run ID: i2rg061
Participant: inst-infocomm-res.yu
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: document retrieval

i2rg062¶

Run ID: i2rg062
Participant: inst-infocomm-res.yu
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: document reranking

i2rg063¶

Run ID: i2rg063
Participant: inst-infocomm-res.yu
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Passage Retrieval

icb1¶

Run ID: icb1
Participant: weill-med-cornellu
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Run 1 was performed with queries at the full article level only. Slider position 200. In this run, we used the MG4J Vigna scorer as baseline. The Vigna scorer favors matches where search terms appear in short text intervals. All runs are performed with the Twease slider at position 200. At this position, the slider expands the query with all the morphological word variants, abbreviations, and MeSH synonyms that match the query words. Morphological word variants are discovered at runtime, with a statistical model trained on Medline 2006 (Campagne, F. unpublished, 2006). Passages are assigned as the minimal intervals where the query match the documents.

icb2¶

Run ID: icb2
Participant: weill-med-cornellu
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Run 2 was performed with parts of the queries at the sentence-level, when appropriate, other terms matching the rest of the article, and ranking by context. Slider at position 200. Context ranking is a new ranking strategy implemented in our textractor framework for the 2006 TREC genomics track. Context queries are expressed as (query)/(context). Briefly, context ranking allows to rank documents matching query by a context, specified as a query expression (e.g., "colon cancer" as a phrase or keywords with boolean clauses). The words in the context do not necessarily occur in the document being ranked. The documents matching the context part of the query are used to infer words that are associated with the context in the corpus. These words are then used to rank the specific set of documents. All runs are performed with the Twease slider at position 200. At this position, the slider expands the query with all the morphological word variants, abbreviations, and MeSH synonyms that match the query words. Morphological word variants are discovered at runtime, with a statistical model trained on Medline 2006 (Campagne, F. unpublished, 2006). Passages are assigned as the minimal intervals where the query match the documents.

icb3¶

Run ID: icb3
Participant: weill-med-cornellu
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Run 3 was performed with queries at the full article level, ranked by context as in Run 2. The context of queries in Run 2 were added to queries from Run 1 to form queries for this run. Slider at position 200. For each topic, queries have the form (query run 1) / (context run 2). All runs are performed with the Twease slider at position 200. At this position, the slider expands the query with all the morphological word variants, abbreviations, and MeSH synonyms that match the query words. Morphological word variants are discovered at runtime, with a statistical model trained on Medline 2006 (Campagne, F. unpublished, 2006). Passages are assigned as the minimal intervals where the query match the documents.

iitx1¶

Run ID: iitx1
Participant: iit.urbain
Track: Genomics
Year: 2006
Submission: 7/19/2006
Type: automatic
Run description: sentMatchRatioNormSC + passMatchRatioNormSC

iitx2¶

Run ID: iitx2
Participant: iit.urbain
Track: Genomics
Year: 2006
Submission: 7/19/2006
Type: automatic
Run description: sentmatchrationormsc+sentnormsc+passmatchrationormsc+passnormsc)/4

iitx3¶

Run ID: iitx3
Participant: iit.urbain
Track: Genomics
Year: 2006
Submission: 7/19/2006
Type: automatic
Run description: (1sentmatchrationormsc+0.1passmatchrationormsc+0.01sentnormsc+0.001passnormsc)

kyoto1¶

Run ID: kyoto1
Participant: kyotou.wan
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Paragraph-level IR with impact-based retrieval and a probabilistic model for term co-occurrence with their scores merged. Queries expanded automatically with synonymns.

kyoto2¶

Run ID: kyoto2
Participant: kyotou.wan
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: a combination of IR impact-based retrieval at document level with a probabilistic model of term coocurance at paragraph level; for the first phase, queries are automatically expanded using synonyms.

kyoto20¶

Run ID: kyoto20
Participant: kyotou.wan
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: a combination of IR impact-based retrieval at document level with a probabilistic model of term coocurance at paragraph level; for the first phase, original queries are employed.

LARAg06pe0¶

Results | Participants | Input | Summary | Appendix

Run ID: LARAg06pe0
Participant: technion.gabrilovich
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: In the preprocessing phase, documents are indexed with BOW and with an additional set of knowledge-rich features based on Wikipedia concepts. First, a simple BOW query is generated from the topic (no expansion or other enhancements). Then, the top 10 returned documents are mapped into most relevant Wikipedia concepts. The resulting concepts are used to query the second index of documents. No explicit domain-specific knowledge is used. Due to lack of time, retrieval is of entire paragraphs, not passages.

LARAg06pe5¶

Results | Participants | Input | Summary | Appendix

Run ID: LARAg06pe5
Participant: technion.gabrilovich
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Note this run is identical to LARAg06pe0 except the use of query expansion. In the preprocessing phase, documents are indexed with BOW and with an additional set of knowledge-rich features based on Wikipedia concepts. First, a simple BOW query is generated from the topic, with blind feedback query expansion. Then, the top 10 returned documents are mapped into most relevant Wikipedia concepts. The resulting concepts are used to query the second index of documents. No explicit domain-specific knowledge is used. Due to lack of time, retrieval is of entire paragraphs, not passages.

LARAg06t¶

Results | Participants | Input | Summary | Appendix

Run ID: LARAg06t
Participant: technion.gabrilovich
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Document and query are represented using features generated by an auxiliary classifier that was built using world knowledge extracted from Wikipedia. No other domain-specific or general information is used. Due to lack of time, retrieval is of entire paragraphs, not passages.

NLMfusion¶

Run ID: NLMfusion
Participant: nlm.aronson
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: This run is the equally-weighted fusion of the results of four automatic methods (1) Essie, a search engine developed specifically for biomedical text supporting flexible query expansion; (2) NCBI, a method that performs selective query expansion based on theme analysis; (3) UniGe, a method based on the EasyIR search engine using term and document weightings as well as pivoted normalization; and (4) Smart, a method based on the Smart search engine. Automatic query expansion based on MetaMap and Theme was available to each of the basic methods. Each method produced paragraphs which were then merged into a final list.

NLMinter¶

Run ID: NLMinter
Participant: nlm.aronson
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: interactive
Run description: This run consists of manually constructed queries generally consisting of a conjunction of topic terms each of which is a disjunction of synonyms. The synonyms were obtained both by introspection and by consulting databases such as Entrez Gene, GeneCards and MeSH. Query development sometimes also involved examination of PubMed and Essie results of preliminary query formulations. The queries were processed by Essie, and the results were automatically trimmed of text unrelated to the topics.

NLMmanual¶

Run ID: NLMmanual
Participant: nlm.aronson
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: manual
Run description: This is similar to the automatic Essie method which is part of our automatic fusion run but with some manually modified queries and with results automatically trimmed of text unrelated to the topics.

NTUadh1¶

Run ID: NTUadh1
Participant: ntu.chen
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: The underlying retrieval model is KL-divergence. Synonyms for query expansion are selected by checking that the synonyms co-occur with the original query terms in Pubmed's Medline abstracts.

NTUadh2¶

Run ID: NTUadh2
Participant: ntu.chen
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: A baseline run using KL-divergence retrieval model.

NTUadh3¶

Run ID: NTUadh3
Participant: ntu.chen
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: manual
Run description: Same as NTUadh1, except that Nur-77 is manually added to queries containing Nurr-77.

OHSUBigclu¶

Run ID: OHSUBigclu
Participant: ohsu.hersh
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Same as cluster run. Reranking by clustering of similar returns. Parameters for clustering were modified so that cluster were looser.

OHSUCluster¶

Run ID: OHSUCluster
Participant: ohsu.hersh
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Same as noclu. The returned passages were further processed by clustering with CLUTO. Features for clustering are text words from the passage with stopwords filtered out and stemming.

OHSUNoclu¶

Run ID: OHSUNoclu
Participant: ohsu.hersh
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Automatically generated queries with concept expansion. Documents indexed at legal span granularity with Lucene. Retrieved passages scored by tfidf.

PCPsgAspect¶

Run ID: PCPsgAspect
Participant: purdueu.si
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Combine multiple types of resources for constructing queries; Hierarchical language model smoothing; Post result filter; Aspect retrieval based on vector representation of MMR

PCPsgClean¶

Run ID: PCPsgClean
Participant: purdueu.si
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Combine multiple types of resources for constructing queries; Hierarchical language model smoothing; Post result filter

PCPsgRescore¶

Run ID: PCPsgRescore
Participant: purdueu.si
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Combine multiple types of resources for constructing queries; Hierarchical language model smoothing; Post result filter; Combine multiple types of evidence

THU1¶

Results | Participants | Input | Summary | Appendix

Run ID: THU1
Participant: tsinghuau.zhang
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Our best result.

THU2¶

Results | Participants | Input | Summary | Appendix

Run ID: THU2
Participant: tsinghuau.zhang
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Shorter Passages to return.

THU3¶

Results | Participants | Input | Summary | Appendix

Run ID: THU3
Participant: tsinghuau.zhang
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: Longer Passages to return.

Tlab6r2GT2¶

Results | Participants | Input | Summary | Appendix

Run ID: Tlab6r2GT2
Participant: utokyo.ishii
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Automaticcally calculating abstract level of biomedical concepts and disambiguation of them. Another condition.

Tlab6r2GT3¶

Results | Participants | Input | Summary | Appendix

Run ID: Tlab6r2GT3
Participant: utokyo.ishii
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Automaticcally calculating abstract level of biomedical concepts and disambiguation of them. Yet another condition.

Tlab6rGT1¶

Results | Participants | Input | Summary | Appendix

Run ID: Tlab6rGT1
Participant: utokyo.ishii
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Automaticcally calculating abstract level of biomedical concepts and disambiguation of them.

UAmsBaseLine¶

Run ID: UAmsBaseLine
Participant: uamsterdam.meij
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Baseline. Just some naive index-specific acronym axpansion on identified (and extracted) NP's

UAmsExp¶

Run ID: UAmsExp
Participant: uamsterdam.meij
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Massive query expansion, using online resources and iteratively gathered acronyms

UAmsExpSel¶

Run ID: UAmsExpSel
Participant: uamsterdam.meij
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Automatically identified obligatory terms (and expansions)

UBexp1¶

Run ID: UBexp1
Participant: suny-buffalo.ruiz
Track: Genomics
Year: 2006
Submission: 7/17/2006
Type: automatic
Run description: This run uses a pre-retrieval query expansion method that adds gene names and synonyms. Retrieval is performed using SMART Lnu.ltu and returning full paragrpahs.

UBexp1M¶

Run ID: UBexp1M
Participant: suny-buffalo.ruiz
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: The run has been generated with SMART using pivoted normalization.

UBexp2¶

Run ID: UBexp2
Participant: suny-buffalo.ruiz
Track: Genomics
Year: 2006
Submission: 7/17/2006
Type: automatic
Run description: This run uses automatic pre-retrieval query that adds gene names and synonyms. Retrieval is performed using SMART with atn ann weighting scheme. Retrieval step returns full paragraphs.

UBexp2M¶

Run ID: UBexp2M
Participant: suny-buffalo.ruiz
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: The run has been generated with SMART using pivoted normalization (2nd run from Miguel Ruiz).

uchsc1¶

Run ID: uchsc1
Participant: ucolorado.cohen
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Expanded queries are sent to the search engine Lemur. Results undergo zone filtering, and top remaining Lemur results are sent to a singular value decomposition algorithm to expand the results pool by selecting similar paragraphs based on a latent semantic Dirichlet similarity score. Results of the SVD are filtered using Naive Bayes with lexical and conceptual features with training data dervied from manual evaluation of Lemer output.

uchsc2¶

Run ID: uchsc2
Participant: ucolorado.cohen
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Expanded queries are sent to the search engine Lemur. Results undergo zone filtering. A second, less strict, set of queries is sent to the Lemur search engine and results are filtered using zone filtering and Naive Bayes with lexical and conceptual features with training data dervied from manual evaluation of Lemer output.

uchsc3¶

Run ID: uchsc3
Participant: ucolorado.cohen
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: manual
Run description: Expanded queries are sent to the search engine Lemur. Results undergo zone filtering.

UICGenRun1¶

Run ID: UICGenRun1
Participant: uillinois.chicago.zhou
Track: Genomics
Year: 2006
Submission: 7/17/2006
Type: automatic
Run description: two-dimensional ranking

UICGenRun2¶

Run ID: UICGenRun2
Participant: uillinois.chicago.zhou
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: two-dimensional ranking query expansion

UICGenRun3¶

Run ID: UICGenRun3
Participant: uillinois.chicago.zhou
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: 2-dimensional ranking; query expansion; passage retrieval

UIowa06Geno1¶

Run ID: UIowa06Geno1
Participant: uiowa.eichmann
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: NLP processing of question, entire paragraph returned as result

UIowa06Geno2¶

Run ID: UIowa06Geno2
Participant: uiowa.eichmann
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: NLP processing of question, paragraphs contracted to only those sentences mentioning query terms.

UIowa06Geno3¶

Run ID: UIowa06Geno3
Participant: uiowa.eichmann
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: automatic
Run description: NLP processing of question, entire paragraphs returned, but only those at least 300 characters long (as an ad hoc citation exclusion mechanism).

UIUCauto¶

Results | Participants | Input | Summary | Appendix

Run ID: UIUCauto
Participant: uiuc.zhai
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Automatic run.

UIUCinter¶

Results | Participants | Input | Summary | Appendix

Run ID: UIUCinter
Participant: uiuc.zhai
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: interactive
Run description: Interactive run.

UIUCinter2¶

Results | Participants | Input | Summary | Appendix

Run ID: UIUCinter2
Participant: uiuc.zhai
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: interactive
Run description: Interactive run 2.

UMassCIIR1¶

Run ID: UMassCIIR1
Participant: umass.allan
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: interactive
Run description: Query-biased pseudo relevance feedback. 250 word passages with overlap removed.

UMassCIIR1L¶

Run ID: UMassCIIR1L
Participant: umass.allan
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: interactive
Run description: Query-biased pseudo relevance feedback. The UMassCIIR1 run was "legalized" to only be spans from the legalspans file. Legal spans less than 750 chars were excluded.

UMassCIIR2¶

Run ID: UMassCIIR2
Participant: umass.allan
Track: Genomics
Year: 2006
Submission: 7/21/2006
Type: interactive
Run description: Query-biased pseudo relevance feedback. 500 word passages with overlap removed.

UniGe¶

Run ID: UniGe
Participant: uhosp-geneva.ruch
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: Use the easyIR engine a vector-space with tf.idf weightings and a modified version of pivoted normalization. Basic run.

UnigeGO¶

Run ID: UnigeGO
Participant: uhosp-geneva.ruch
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: Use the easyIR engine a vector-space with tf.idf weightings and a modified version of pivoted normalization. GO specific reranking.

UnigeMesh¶

Run ID: UnigeMesh
Participant: uhosp-geneva.ruch
Track: Genomics
Year: 2006
Submission: 7/18/2006
Type: automatic
Run description: Use the easyIR engine a vector-space with tf.idf weightings and a modified version of pivoted normalization. Template-specific semantic filtering and expansion.

UniNE1¶

Run ID: UniNE1
Participant: uneuchatel.savoy
Track: Genomics
Year: 2006
Submission: 7/24/2006
Type: automatic
Run description: Data fusion of two IR systems (based on normalized RSV values using Z-score) IR system 1 Divergence from randomness, word-based indexing, spelling correction & word variant generation IR system 2 Divergence from randomness, 5-gram indexing

UniNE2¶

Run ID: UniNE2
Participant: uneuchatel.savoy
Track: Genomics
Year: 2006
Submission: 7/24/2006
Type: automatic
Run description: Data fusion of two IR systems (based on normalized RSV values (max)) IR system 1 Divergence from randomness, word-based indexing, spelling correction & word variant generation the document title is included to all passages generated from the article IR system 2 Divergence from randomness, 5-gram indexing

UniNE3¶

Run ID: UniNE3
Participant: uneuchatel.savoy
Track: Genomics
Year: 2006
Submission: 7/24/2006
Type: automatic
Run description: Data fusion of two IR systems (based on normalized RSV values (Z-score), baserun for comparisons) IR system 1 Divergence from randomness, word-based indexing IR system 2 Divergence from randomness, 5-gram indexing

UofG0¶

Run ID: UofG0
Participant: uguelph.song
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Retrieval based on the language modeling approach.

UofG1¶

Run ID: UofG1
Participant: uguelph.song
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Retrieval based on the language modeling approach. The results are further filtered based on document coverage.

UofG2¶

Run ID: UofG2
Participant: uguelph.song
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Retrieval based on language modeling approach. The results are further filtered based on document and aspect coverage.

WiscRun1¶

Run ID: WiscRun1
Participant: uwisconsin.craven
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: - Performs POS chunking on topic questions to identify significant noun phrases - Automatically generates expansion term lists for each NP using the MeSH database - Uses Lemur/Indri toolkit to execute queries that require one item in each term list to be found in a paragraph - Ranks results using likelihood of paragraphs given all the expansion term lists concatenated together - Adjusts passage boundaries to include only sentences between the first and last occurrence of key terms

WiscRun2¶

Run ID: WiscRun2
Participant: uwisconsin.craven
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: - Begins with the same baseline results as our WiscRun1 run - Re-ranks these results by performing hierarchical clustering on passage bag-of-words vectors - Interleaves results from clusters to promote aspect diversity (Note that clusters are repeatedly considered in order of their average initial rank)

WiscRun3¶

Run ID: WiscRun3
Participant: uwisconsin.craven
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: - Same baseline results as our WiscRun1 run - Re-ranks using GRASSHOPPER, a graph theoretical algorithm that performs random walk with absorbing states on the results, to automatically balance the representativeness and diversity of the final rank

york06ga1¶

Run ID: york06ga1
Participant: yorku.huang
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: 1. Use Okapi BM25 for concept-based structured query 2. Use the blind feedback with term selection technique 3. Use a dual index model for passage retrieval 4. No asspect-level retrieval

york06ga3¶

Run ID: york06ga3
Participant: yorku.huang
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: Split the top 500 retrieved passages into 5 groups with 100 passages in each group and then use the EM clustering algorithm to re-rank the 100 passages in each group for aspect-level retrieval

york06ga4¶

Run ID: york06ga4
Participant: yorku.huang
Track: Genomics
Year: 2006
Submission: 7/20/2006
Type: automatic
Run description: This run is for document-level retrieval. That is documents will appear in the front of list for only once and those retrieved by different passage previously will be put at the end of list. No aspect-level retrieval.