Talk given at CLEF 2007, Budapest, Hungary

1. Exploring Location Indicators for
Geographic Information Retrieval
Johannes Leveling and Sven Hartrumpf
Intelligent Information and Communication Systems (IICS)
University of Hagen (FernUniversität in Hagen)
58084 Hagen, Germany
firstname.lastname@fernuni-hagen.de
CLEF 2007 Workshop, Budapest, Hungary

2. Exploring
Location
Indicators for
Geographic
Information
Outline
Retrieval
Johannes
Leveling
and
Sven 1 Introduction
Hartrumpf
Introduction 2 Location Indicators
Location
Indicators
Location
3 Location Indicator Normalization
Indicator
Normalization
Semantic 4 Semantic Analysis for GIR
Analysis for
GIR
GeoCLEF 5 GeoCLEF 2007 Experiments
2007
Experiments
Conclusion
and Outlook
6 Conclusion and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 2 / 17

3. Exploring
Location
Indicators for
Geographic
Information
Introduction
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
• Traditional information retrieval (IR):
stemming is applied to all words in a text
Introduction
Location
• Geographical information retrieval (GIR):
Indicators use named entity recognition and classification;
Location
Indicator
avoid stemming location names (typically, proper nouns
Normalization
only); employ geographic knowledge
Semantic
Analysis for • GIRSA (Geographic Information Retrieval by Semantic
GIR
GeoCLEF
Annotation):
2007 aims at a broader GIR approach not solely based on
Experiments
Conclusion
location names, but on location indicators
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 3 / 17

4. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location
Indicator
Normalization
Semantic
Analysis for
GIR
GeoCLEF
2007
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

5. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location • Adjectives corresponding to a location.
Indicator
Normalization Example:
Semantic
Analysis for
tunesisch →Tunesien
GIR (Tunisian →Tunisia)
GeoCLEF
2007
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

6. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location • Demonyms, e.g. the name for inhabitants originating
Indicator
Normalization from a location.
Semantic
Analysis for Example:
GIR
Franzose, Französin →Frankreich
GeoCLEF
2007 (Frenchman, Frenchwoman →France)
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

7. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location • Codes for a location name.
Indicator
Normalization
Example:
Semantic
Analysis for HU21 →Tolna County, Hungary (FIPS region code)
GIR
GeoCLEF
2007
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

8. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location • Abbreviations and acronyms for a location name,
Indicator
Normalization including adjectives.
Semantic
Analysis for
GIR
Example:
GeoCLEF
franz. →französisch →Frankreich
2007
Experiments
(French →France)
Conclusion TX →Texas
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

9. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location • Orthographic variants, exonyms, historic names.
Indicator
Normalization
Semantic
Example:
Analysis for
GIR
Lower Saxony →Niedersachsen
GeoCLEF
2007
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

10. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location • Unique entities associated with a geographic
Indicator
Normalization location, e.g. headquarters of an organization,
Semantic persons, buildings.
Analysis for
GIR
GeoCLEF
Example:
2007
Experiments
Eiffel Tower →Paris
Conclusion Moliére →France (?)
and Outlook
VW →Wolfsburg (?)
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

11. Exploring
Location
Indicators for
Geographic
Information
Location Indicators
Retrieval
Johannes
Leveling
and
Sven Definition
Hartrumpf
Location indicators are text segments from which the
Introduction geographic scope of a document can be inferred.
Location
Indicators
Location • The location names itself (full names and short
Indicator
Normalization forms).
Semantic
Analysis for
GIR
Example:
GeoCLEF
Republik Korea →Südkorea
2007
Experiments
(Republic of Korea →South Korea)
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 4 / 17

12. Exploring
Location
Indicators for
Geographic
Information
Location Indicator Normalization
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
Normalization on surface (character), morphologic,
syntactic, semantic, and lexical level.
Introduction
Location
Indicators Character level
Location
Indicator • Diacritical marks replaced with non-accented
Normalization
characters
Semantic
Analysis for • Orthographic variants normalized by selecting a
GIR
GeoCLEF
representative
2007
Experiments Example:
Conclusion
and Outlook
Québec →Quebec
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 5 / 17

13. Exploring
Location
Indicators for
Geographic
Information
Location Indicator Normalization
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
Normalization on surface (character), morphologic,
syntactic, semantic, and lexical level.
Introduction
Location
Indicators Morphologic level
Location
Indicator • Inflectional endings are identified and removed
Normalization
Semantic • Morphologic variations of location names are reduced
Analysis for
GIR to their base form
GeoCLEF • Derivational morphology: adjective →location name
2007
Experiments
Conclusion
Examples:
and Outlook des Roten Meer(e)s →Rote Meer
References bayrisch →Bayern
dänisch →Dänemark
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 5 / 17

14. Exploring
Location
Indicators for
Geographic
Information
Location Indicator Normalization
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
Normalization on surface (character), morphologic,
syntactic, semantic, and lexical level.
Introduction
Location
Indicators Semantic level
Location
Indicator • Prefixes are separated from the name
Normalization
Semantic
• Location indicators are mapped to location names
Analysis for
GIR Examples:
GeoCLEF
2007
Norddeutschland →Nord-Deutschland
Experiments exception:
Conclusion
and Outlook
Südafrika →Südafrika
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 5 / 17

15. Exploring
Location
Indicators for
Geographic
Information
Location Indicator Normalization
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
Normalization on surface (character), morphologic,
syntactic, semantic, and lexical level.
Introduction
Location
Indicators Lexical level
Location
Indicator • Name variations are normalized using synset
Normalization
representatives
Semantic
Analysis for
GIR Example:
GeoCLEF Burma →Myanmar
2007
Experiments Birma →Myanmar
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 5 / 17

16. Exploring
Location
Indicators for
Geographic
Information
Semantic Analysis for GIR
Retrieval
Johannes
Leveling
and • Extension of semantic network matching approach,
Sven
Hartrumpf GIR-InSicht (Leveling et al. (2006)),
Introduction
derived from the deep question answering (QA) system
Location
InSicht (Hartrumpf and Leveling (2007))
Indicators
• Query semantic network was allowed to be split in
Location
Indicator parts at specific semantic relations, e.g. at a
Normalization
LOC ( ATION ) relation
Semantic
Analysis for
GIR
• Query decomposition:
GeoCLEF a query can be decomposed into two dependent
2007
Experiments queries, the subquery and the main query
Conclusion • The subquery is answered by the QA system InSicht;
and Outlook
References
answers are integrated into the main query
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 6 / 17

17. Exploring
Location
Indicators for
Geographic
Information
Semantic Analysis Example
Retrieval
Johannes
Leveling
and
Sven Topic 10.2452/57-GC
Hartrumpf
Whiskyherstellung auf den schottischen Inseln/
Introduction “Whiskey production on the Scottish Islands”
Location
Indicators
Location
Indicator Inferential query expansion followed by query
Normalization
decomposition
Semantic
Analysis for →Subquery Nenne schottische Inseln/
GIR
GeoCLEF
“Name Scottish islands”
2007
Experiments
Subquery Nenne Inseln in Schottland/
Conclusion
“Name islands in Scotland” (inferences)
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 7 / 17

18. Exploring
Location
Indicators for
Geographic
Information
Semantic Analysis Example
Retrieval
Johannes
Leveling
and
Sven Topic 10.2452/57-GC
Hartrumpf
Whiskyherstellung auf den schottischen Inseln/
Introduction “Whiskey production on the Scottish Islands”
Location
Indicators
Location
Indicator Answering the subqueries on the GeoCLEF corpus
Normalization
and the German Wikipedia
Semantic
Analysis for →Partial answers Iona and Islay
GIR
GeoCLEF
→Better gazetteer entry points
2007
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 7 / 17

19. Exploring
Location
Indicators for
Geographic
Information
Semantic Analysis Example
Retrieval
Johannes
Leveling
and
Sven Topic 10.2452/57-GC
Hartrumpf
Whiskyherstellung auf den schottischen Inseln/
Introduction “Whiskey production on the Scottish Islands”
Location
Indicators
Location
Indicator New queries (paraphrased)
Normalization
→New queries Whiskyherstellung auf Iona/
Semantic
Analysis for “Whiskey production on Iona”
GIR
GeoCLEF
and Whiskyherstellung auf Islay /
2007
Experiments
“Whiskey production on Islay”
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 7 / 17

20. Exploring
Location
Indicators for
Geographic
Information
Semantic Analysis Example
Retrieval
Johannes
Leveling
and
Sven Topic 10.2452/57-GC
Hartrumpf
Whiskyherstellung auf den schottischen Inseln/
Introduction “Whiskey production on the Scottish Islands”
Location
Indicators
Location
Indicator
Normalization
→In total, 80 different subqueries were produced for the 25
Semantic
topics
Analysis for
GIR
GeoCLEF
2007
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 7 / 17

21. Exploring
Location
Indicators for
Geographic
Information
Experimental Setup
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
• GeoCLEF 2007 documents: 275,000 German
Introduction
newspaper articles from Frankfurter Rundschau,
Location
Indicators Schweizerische Depeschenagentur, and Der Spiegel
Location from the years 1994 and 1995
Indicator
Normalization • GIRSA evaluated on 25 GeoCLEF topics with a title (T),
Semantic
Analysis for
a short description (D), and a narrative part (N)
GIR
• Setup similar to previous GIR experiments on
GeoCLEF
2007 GeoCLEF data Leveling et al. (2006); Leveling (2007)
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 8 / 17

22. Exploring
Location
Indicators for
Geographic
Information
Methods for GIR (1/3)
Retrieval
Johannes
Leveling
PoS-Tagger/NERC (TnT, Lingpipe etc.):
and
Sven • Andogah, Bouma et al. (U. Groningen)
Hartrumpf
• Buscaldi, Rosso (U. Valencia)
Introduction
• Ferrés, Rodríguez (U. Catalunya)
Location
Indicators
• Kölle, Heuwing et al. (U. Hildesheim)
Location
Indicator • Lana-Serrano, Villena-Román et al. (U. Madrid)
Normalization
Semantic • Overell, Magalhães et al. (IC London)
Analysis for
GIR • Perea-Ortega, García-Cumbreras et al. (U. Jaén)
GeoCLEF
2007 List lookup:
Experiments
• Leveling, Hartrumpf (U. Hagen)
Conclusion
and Outlook
• Larson (U. C. Berkeley)
References
→Only part of the solution, but GIRSA needs this, too!
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 9 / 17

23. Exploring
Location
Indicators for
Geographic
Information
Methods for GIR (1/3)
Retrieval
Johannes
Leveling
PoS-Tagger/NERC (TnT, Lingpipe etc.):
and
Sven • Andogah, Bouma et al. (U. Groningen)
Hartrumpf
• Buscaldi, Rosso (U. Valencia)
Introduction
• Ferrés, Rodríguez (U. Catalunya)
Location
Indicators
• Kölle, Heuwing et al. (U. Hildesheim)
Location
Indicator • Lana-Serrano, Villena-Román et al. (U. Madrid)
Normalization
Semantic • Overell, Magalhães et al. (IC London)
Analysis for
GIR • Perea-Ortega, García-Cumbreras et al. (U. Jaén)
GeoCLEF
2007 List lookup:
Experiments
• Leveling, Hartrumpf (U. Hagen)
Conclusion
and Outlook
• Larson (U. C. Berkeley)
References
→Only part of the solution, but GIRSA needs this, too!
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 9 / 17

24. Exploring
Location
Indicators for
Geographic
Information
Methods for GIR (2/3)
Retrieval
Johannes
Gazetteers/GKB (GNS, WordNet etc.):
Leveling
and • Andogah, Bouma et al. (U. Groningen)
Sven
Hartrumpf • Buscaldi, Rosso (U. Valencia)
Introduction
• Cardoso, Cruz et al. (U. Lisbon)
Location • Ferrés, Rodríguez (U. Catalunya)
Indicators
• Guillén (CSU)
Location
Indicator
Normalization
• Lana-Serrano, Villena-Román et al. (U. Madrid)
Semantic • Larson (U. C. Berkeley)
Analysis for
GIR • Li, Wang et al. (Microsoft Asia)
GeoCLEF
2007
• Nasikhin, Adriani (U. Indonesia)
Experiments
• Overell, Magalhães et al. (IC London)
Conclusion
and Outlook Small name lists (about 250,000 entries):
References • Leveling, Hartrumpf (U. Hagen)
→GIRSA does not use geographic knowledge, yet.
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 10 / 17

25. Exploring
Location
Indicators for
Geographic
Information
Methods for GIR (2/3)
Retrieval
Johannes
Gazetteers/GKB (GNS, WordNet etc.):
Leveling
and • Andogah, Bouma et al. (U. Groningen)
Sven
Hartrumpf • Buscaldi, Rosso (U. Valencia)
Introduction
• Cardoso, Cruz et al. (U. Lisbon)
Location • Ferrés, Rodríguez (U. Catalunya)
Indicators
• Guillén (CSU)
Location
Indicator
Normalization
• Lana-Serrano, Villena-Román et al. (U. Madrid)
Semantic • Larson (U. C. Berkeley)
Analysis for
GIR • Li, Wang et al. (Microsoft Asia)
GeoCLEF
2007
• Nasikhin, Adriani (U. Indonesia)
Experiments
• Overell, Magalhães et al. (IC London)
Conclusion
and Outlook Small name lists (about 250,000 entries):
References • Leveling, Hartrumpf (U. Hagen)
→GIRSA does not use geographic knowledge, yet.
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 10 / 17

26. Exploring
Location
Indicators for
Geographic
Information
Methods for GIR (3/3)
Retrieval
Johannes
Leveling
and Blind Feedback:
Sven
Hartrumpf • Cardoso, Cruz et al. (U. Lisbon)
Introduction
• Ferrés, Rodríguez (TALP) – Relevance Feedback
Location • Guillén (CSU)
Indicators
Location • Kölle, Heuwing et al. (Hildesheim)
Indicator
Normalization • Larson (U. C. Berkeley)
Semantic
Analysis for • Nasikhin, Adriani (U. Indonesia)
GIR
GeoCLEF
• Overell, Magalhães et al. (IC London)
2007
Experiments No Blind Feedback:
Conclusion • Leveling, Hartrumpf (U. Hagen)
and Outlook
References →GIRSA will not utilize ad-hoc blind feedback!
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 11 / 17

27. Exploring
Location
Indicators for
Geographic
Information
Methods for GIR (3/3)
Retrieval
Johannes
Leveling
and Blind Feedback:
Sven
Hartrumpf • Cardoso, Cruz et al. (U. Lisbon)
Introduction
• Ferrés, Rodríguez (TALP) – Relevance Feedback
Location • Guillén (CSU)
Indicators
Location • Kölle, Heuwing et al. (Hildesheim)
Indicator
Normalization • Larson (U. C. Berkeley)
Semantic
Analysis for • Nasikhin, Adriani (U. Indonesia)
GIR
GeoCLEF
• Overell, Magalhães et al. (IC London)
2007
Experiments No Blind Feedback:
Conclusion • Leveling, Hartrumpf (U. Hagen)
and Outlook
References →GIRSA will not utilize ad-hoc blind feedback!
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 11 / 17

28. Exploring
Location
Indicators for
Geographic
Information
Experimental Setup
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
Different indexes:
Introduction S: All words in the document text are stemmed
Location
Indicators SL: Location indicators are identified and normalized to a
Location base form of a location name
Indicator
Normalization
SLD: In addition, decompounding is applied to the words in
Semantic
Analysis for the text
GIR
GeoCLEF
O: Documents and queries are represented as semantic
2007 networks and GIR is seen as (a form of) QA
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 12 / 17

29. Exploring
Location
Indicators for
Geographic
Information
Results and Discussion
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
Run Parameters Results
Introduction
index fields rel_ret MAP P@5
Location
Indicators
FUHtd1de S TD 597 0.119 0.280
Location FUHtd2de SL TD 707 0.191 0.288
Indicator
Normalization FUHtd3de SLD TD 677 0.190 0.272
Semantic FUHtdn4de SL TDN 722 0.236 0.328
Analysis for FUHtdn5de SLD TDN 717 0.258 0.336
GIR
FUHtd6de SLD/O TD 680 0.196 0.280
GeoCLEF
2007 GIR-InSicht O TD 52 0.067 0.104
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 13 / 17

30. Exploring
Location
Indicators for
Geographic
Information
Results for Monolingual German
Retrieval Precision
Johannes
Leveling 0.60
and ♦ FUHtd1de: 0.119 MAP
Sven FUHtd3de: 0.190 MAP
Hartrumpf 0.50 FUHtd6de: 0.196 MAP
× GIR-InSicht: 0.067 MAP
Introduction
Location
0.40 ♦
Indicators
Location 0.30 ♦
Indicator
Normalization
♦
Semantic 0.20
Analysis for ♦
GIR ×
×
GeoCLEF 0.10 × × ♦
2007 × ♦ ♦ ♦
Experiments × × × × ♦
× ×
♦
Conclusion 0.00 ♦
and Outlook 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
References Recall
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 14 / 17

31. Exploring
Location
Indicators for
Geographic
Information
Conclusion
Retrieval
Johannes
Leveling
and
Sven
Hartrumpf
Introduction • Baseline run (FUHtd1de) is clearly outperformed
Location
Indicators • Adding selected location names (from the narrative)
Location notably improves performance
Indicator
Normalization • Hybrid approach (with GIR-InSicht) for GIR proved
Semantic
Analysis for
interesting:
GIR even a few additional relevant documents were found
GeoCLEF
2007
Experiments
Conclusion
and Outlook
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 15 / 17

32. Exploring
Location
Indicators for
Geographic
Information
Outlook
Retrieval
Johannes
Leveling
and
Sven
Planned improvements for GIRSA:
Hartrumpf
• Estimate the importance (weight) of different location
Introduction indicators, possibly depending on the context:
Location Danish coast →Denmark, but
Indicators
German shepherd → Germany
Location
Indicator
Normalization
• Apply part-of-speech tagger and named entity
Semantic recognizer to identify location names
Analysis for
GIR • Investigate the combination of means to increase
GeoCLEF
2007
precision (metonymic uses of location names)
Experiments with means to increase recall (normalizing location
Conclusion
and Outlook
indicators)
References
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 16 / 17

33. Exploring
Location
Indicators for
Geographic
Information
Selected References
Retrieval
Johannes
Leveling Hartrumpf, Sven and Johannes Leveling (2007). Interpretation and
and
Sven normalization of temporal expressions for question answering. In
Hartrumpf
Evaluation of Multilingual and Multi-modal Information Retrieval: 7th
Workshop of the Cross-Language Evaluation Forum, CLEF 2006
Introduction
(edited by et al., Carol Peters), volume 4730 of LNCS, pp. 432–439.
Location
Indicators Berlin: Springer.
Location Leveling, Johannes (2007). Experiments on the exclusion of metonymic
Indicator
Normalization
location names from GIR. In Evaluation of Multilingual and
Multi-modal Information Retrieval: 7th Workshop of the
Semantic
Analysis for Cross-Language Evaluation Forum, CLEF 2006 (edited by et al.,
GIR
Carol Peters), volume 4730 of LNCS, pp. 901–904. Berlin: Springer.
GeoCLEF
2007 Leveling, Johannes; Sven Hartrumpf; and Dirk Veiel (2006). Using
Experiments semantic networks for geographic information retrieval. In Accessing
Conclusion Multilingual Information Repositories: 6th Workshop of the
and Outlook
Cross-Language Evaluation Forum, CLEF 2005 (edited by et al.,
References Carol Peters), volume 4022 of LNCS, pp. 977–986. Berlin: Springer.
Johannes Leveling and Sven Hartrumpf Exploring Location Indicators for Geographic Information Retrieval 17 / 17