Runs - Temporal Summarization 2013¶

Baseline¶

Participants | Proceedings | Input | Appendix

Run ID: Baseline
Participant: hltcoe
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: 71c54d3bc7fc5a3f20f6e955e0de593a
Run description: Baseline: Filtering document by time interval, containing all query keywords, cosine similarity with event query and title larger than 0.2. Select sentence by considering: relevance to the query event, novelty compare to previously selected sentences, coverage of collected name entities from previously selected sentences, and check whether containing numbers.

BasePred¶

Participants | Proceedings | Input | Appendix

Run ID: BasePred
Participant: hltcoe
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: 133d468921456909f2d542b4e3d340a2
Run description: Baseline+Predicate coverage check. Upone baseline, add predicate check for sentence selector, where predicates are collected from previously selected sentences. Predicates are annotated by using Stanford NLP toolkit POS tagger, which uses a MaxEnt model trained with external resources.

cluster1¶

Participants | Proceedings | Input | Appendix

Run ID: cluster1
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: c361549a24f0b77b1c6548be1c88ac37
Run description: Result based on KBA data

cluster2¶

Participants | Proceedings | Input | Appendix

Run ID: cluster2
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 2967e9e60e9f9840bcb1d2a2661d4eb5
Run description: Result based on KBA data

cluster3¶

Participants | Proceedings | Input | Appendix

Run ID: cluster3
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 7453e9d10b908e62ad14e31378c97d44
Run description: Result based on KBA data

cluster4¶

Participants | Proceedings | Input | Appendix

Run ID: cluster4
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 2aa11bb53fb83144980c659f2f05254b
Run description: Result based on KBA data

cluster5¶

Participants | Proceedings | Input | Appendix

Run ID: cluster5
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: 98019adf2f01808cb6f9ed360c3fd636
Run description: only KBA corpus

CosineEgrep¶

Participants | Proceedings | Input | Appendix

Run ID: CosineEgrep
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 8c3358f31af73c1a43ca0cb6261e9134
Run description: Language Modelling with Dirichlet Priors to generate initial documents off of given queries. Run egrep over returned sentences of event type and synonyms. Take sentences returned from that and feed them into a cosine similarity metric based off of upper/lower case character counts and digit counts. Thus, similar sentences are similar to previously seen sentences. Performed on an hour by hour basis.

EXTERNAL¶

Participants | Proceedings | Input | Appendix

Run ID: EXTERNAL
Participant: hltcoe
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: 65ee48ead23f506a81d21c8d7830b097
Run description: Baseline+Predicate checking+Wikipedia query expansion. Given a query event, find relevant wikipedia pages, and collect predicates as the initial predicates to describe events

NormEgrep¶

Participants | Proceedings | Input | Appendix

Run ID: NormEgrep
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 4edf3619bea8fd2b3d05dbcb7545b274
Run description: Language Modelling with Dirichlet Priors to generate initial documents off of given queries. Run egrep over returned sentences of event type and synonyms. Take sentences returned from that and feed them into a Euclidean norm based similarity metric using upper/lower case character counts and digit counts. Thus, similar sentences are similar to previously seen sentences. Performed on an hour by hour basis.

PRISTS1¶

Participants | Proceedings | Input | Appendix

Run ID: PRISTS1
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: vt
MD5: e7187ffb0e2238f9d1864acc838bd2f3
Run description: event 1,6and10 data source is KBA,no external training data.

PRISTS2¶

Participants | Proceedings | Input | Appendix

Run ID: PRISTS2
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: vt
MD5: f6ede41e53765d5dbfe511cd3eecb856
Run description: event 1,2,3,4,5,6,8,9,10 data source is KBA,no external training data.

PRISTS3¶

Participants | Proceedings | Input | Appendix

Run ID: PRISTS3
Participant: PRIS
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: vt
MD5: f59dc7b6cbcd307a1c212f4f73e04cf3
Run description: total events except for 7

Q0¶

Participants | Proceedings | Input | Appendix

Run ID: Q0
Participant: BJUT
Track: Temporal Summarization
Year: 2013
Submission: 8/17/2013
Task: sus
MD5: 5a25b5760e52fba1e738f6ee5826b70a
Run description: In this run,we use KBA corpus

Q1¶

Participants | Proceedings | Input | Appendix

Run ID: Q1
Participant: BJUT
Track: Temporal Summarization
Year: 2013
Submission: 8/17/2013
Task: vt
MD5: b9dcbcfc5d3112771268fb4f9fed1d95
Run description: In this run,we use KBA corpus

Q2¶

Participants | Proceedings | Input | Appendix

Run ID: Q2
Participant: BJUT
Track: Temporal Summarization
Year: 2013
Submission: 8/21/2013
Task: sus
MD5: e49b177a11ead786c668d9367a0afe68
Run description: We use KBA corpus in this run

rg1¶

Participants | Proceedings | Input | Appendix

Run ID: rg1
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 7e6073adb49bd9cd82939463a50c7dce
Run description: NA

rg2¶

Participants | Proceedings | Input | Appendix

Run ID: rg2
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 519c5bf302a8fec4254a2efd397f467c
Run description: NA

rg3¶

Participants | Proceedings | Input | Appendix

Run ID: rg3
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 6d6721908b23797abb9ae0f7db7ae833
Run description: NA

rg4¶

Participants | Proceedings | Input | Appendix

Run ID: rg4
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 97374d245dc519b334106205dd74cba8
Run description: NA

run1¶

Participants | Proceedings | Input | Appendix

Run ID: run1
Participant: ICTNET
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: cd7993a5fd764d0e4f2f050842564b4e
Run description: we used the "KBA 2013 english-and-unknown-language streamcorpus", which is from October 2011 to February 2013.

run2¶

Participants | Proceedings | Input | Appendix

Run ID: run2
Participant: ICTNET
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: 0df5c63047523f1ec566a934f61c46a9
Run description: we used the "KBA 2013 english-and-unknown-language streamcorpus", which is from October 2011 to February 2013.

SUS1¶

Participants | Proceedings | Input | Appendix

Run ID: SUS1
Participant: wim_GY_2013
Track: Temporal Summarization
Year: 2013
Submission: 9/1/2013
Task: sus
MD5: 44ebc9fbc28fd5038c23acf0d0b8711e
Run description: We come from the Zhengzhou Information Science and Technology Institute. This is the first run we provide for the Sequential Update Summarization in TS2013. The Run selects summary sentences according to the content of the documents.

TuneBasePred2¶

Participants | Proceedings | Input | Appendix

Run ID: TuneBasePred2
Participant: hltcoe
Track: Temporal Summarization
Year: 2013
Submission: 9/5/2013
Task: sus
MD5: 31ab80fef0c00c76beece3fc2f801dd5
Run description: Filtering document by time interval and cosine similarity to the query topic Select sentence by considering relevance to the topic, novelty with previously seen sentences, coverage of previously seen event major name entities and predicates, and whether containing numbers. Predicates are annotated by using Stanford NLP toolkit, which the POS tagger is using a model trained with external resource.

TuneExternal2¶

Participants | Proceedings | Input | Appendix

Run ID: TuneExternal2
Participant: hltcoe
Track: Temporal Summarization
Year: 2013
Submission: 9/5/2013
Task: sus
MD5: 0ba27b7a98be529b3d13ce70c4b16c94
Run description: Based on TuneBasePred run, use top relevant Wikipedia pages to initialize predicates for a query event. When calculate cosine similarity, use tf-idf as term weight, where Google idfs data is used.

uogTrEMMQ2¶

Participants | Proceedings | Input | Appendix

Run ID: uogTrEMMQ2
Participant: uogTr
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: c5245341a428fd4df37ef8bec17e47bc
Run description: Summary by hour, query focussed summary. Topic query expanded via external time-aligned corpora (wordnet & wikipedia). Ranking sentences by similarity to expanded query, summary selection considering redundancy and diversity.

uogTrNMM¶

Participants | Proceedings | Input | Appendix

Run ID: uogTrNMM
Participant: uogTr
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 216e630eeef562e09ee4fa40ada3df56
Run description: By hour summarisation from the English KBA corpus only. Selects one or more query-focued sentences per document, considering novelty and redundancy.

uogTrNMTm1MM3¶

Participants | Proceedings | Input | Appendix

Run ID: uogTrNMTm1MM3
Participant: uogTr
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: d7c5cf9bd19f05c7667fe999939a7a99
Run description: Summary by hour, ranking sentences by query, selection considering redundancy and diversity. Summary length over time adapted via topic modelling.

uogTrNMTm3FMM4¶

Participants | Proceedings | Input | Appendix

Run ID: uogTrNMTm3FMM4
Participant: uogTr
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: sus
MD5: 653fd89abcd8bbf1764e654737414518
Run description: Summary by hour, ranking sentences by query, summary selection considering redundancy and diversity. Sentences are adaptively selected using topic modelling and then filtered by their confidence value.

uogTrNSQ1¶

Participants | Proceedings | Input | Appendix

Run ID: uogTrNSQ1
Participant: uogTr
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: b92f4e4b6a9fcc0f38b7283b908f1442
Run description: By hour summarisation from the English KBA corpus only. Ranks and then selects top sentences based upon the query and considers redundancy.

UWMDSqlec2t25¶

Participants | Proceedings | Input | Appendix

Run ID: UWMDSqlec2t25
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: 413e8a20077b9cbc3c8ef285d6df9f8c
Run description: Used LMD and Query Expansion with 25 expansion terms

UWMDSqlec4t50¶

Participants | Proceedings | Input | Appendix

Run ID: UWMDSqlec4t50
Participant: UWaterlooMDS
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: sus
MD5: de962e5d22124059ca68c0c6796a9c0e
Run description: Used LMD and Query Expansion with 50 expansion terms

ValueTask¶

Participants | Proceedings | Input | Appendix

Run ID: ValueTask
Participant: ICTNET
Track: Temporal Summarization
Year: 2013
Submission: 9/4/2013
Task: vt
MD5: 98e3b4cf32996cecd32444138fa9733e
Run description: we used the "KBA 2013 english-and-unknown-language streamcorpus", which is from October 2011 to February 2013.

VT1¶

Participants | Proceedings | Input | Appendix

Run ID: VT1
Participant: wim_GY_2013
Track: Temporal Summarization
Year: 2013
Submission: 9/1/2013
Task: vt
MD5: e5814d7eaf45fe97b4d133aff0fe09fa
Run description: We come from the Zhengzhou Information Science and Technology Institute. This is the first run we provide for the Vaule Tracking in TS2013. The run extracts the attribute values according to hand-coded rules.

VT2¶

Participants | Proceedings | Input | Appendix

Run ID: VT2
Participant: wim_GY_2013
Track: Temporal Summarization
Year: 2013
Submission: 9/3/2013
Task: vt
MD5: 5503dcdf41e65a143f0b3a89db17cdc4
Run description: We come from Zhengzhou Information Science and Technology Institute. This is the second run we submit for the Vaule Tracking in TS2013. The run extracts the attribute values according to hand-coded rules with the help of CoreNLP.