Toponym disambiguation in historical documents using semantic and geographic features

Ardanuy, Mariona Coll; Sporleder, Caroline

doi:10.1145/3078081.3078099

Cited by 23 publications

(18 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Performance measures in our experiments are Precision, Recall, F 1 -measure, and mean error distance (M). However, to ascertain whether an interpretation is correctly predicted, we also investigate the error distance between the predicted coordinates and the actual coordinates, as used in numerous studies [5,7,8,19,22,31,35]. This distance enables us to fare various systems against each other since they may select latitude/longitude of locations from different gazetteers or knowledge bases 10 .…”

Section: Evaluation Metricsmentioning

confidence: 99%

“…This distance enables us to fare various systems against each other since they may select latitude/longitude of locations from different gazetteers or knowledge bases 10 . We set the error distance to 10 miles, same as Adaptive method [22], whereas most researches tend to adopt a relaxed threshold (i.e., 161 kilometers) [5,8,31,35].…”

Section: Evaluation Metricsmentioning

confidence: 99%

“…Approaches using knowledge bases Wikipedia has been integrated as a knowledge base into more recent toponym disambiguation techniques [5,8,31,35,36]. Ardanuy and Sporleder [5] address toponym disambiguation in multilingual retrospective articles. They build a model to distill semantic features from Wikipedia information such as page title and article body.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

A Coherent Unsupervised Model for Toponym Resolution

Kamalloo

Rafiei

2018

Proceedings of the 2018 World Wide Web Conference on World Wide Web - WWW '18

View full text Add to dashboard Cite

Toponym Resolution, the task of assigning a location mention in a document to a geographic referent (i.e., latitude/longitude), plays a pivotal role in analyzing location-aware content. However, the ambiguities of natural language and a huge number of possible interpretations for toponyms constitute insurmountable hurdles for this task. In this paper, we study the problem of toponym resolution with no additional information other than a gazetteer and no training data. We demonstrate that a dearth of large enough annotated data makes supervised methods less capable of generalizing. Our proposed method estimates the geographic scope of documents and leverages the connections between nearby place names as evidence to resolve toponyms. We explore the interactions between multiple interpretations of mentions and the relationships between different toponyms in a document to build a model that finds the most coherent resolution. Our model is evaluated on three news corpora, two from the literature and one collected and annotated by us; then, we compare our methods to the state-of-the-art unsupervised and supervised techniques. We also examine three commercial products including Reuters OpenCalais, Yahoo! YQL Placemaker, and Google Cloud Natural Language API. The evaluation shows that our method outperforms the unsupervised technique as well as Reuters OpenCalais and Google Cloud Natural Language API on all three corpora; also, our method shows a performance close to that of the state-of-the art supervised method and outperforms it when the test data has 40% or more toponyms that are not seen in the training data.

show abstract

Section: Evaluation Metricsmentioning

confidence: 99%

Section: Evaluation Metricsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

A Coherent Unsupervised Model for Toponym Resolution

Kamalloo

Rafiei

2018

Proceedings of the 2018 World Wide Web Conference on World Wide Web - WWW '18

View full text Add to dashboard Cite

show abstract

“…They serve many purposes, he obvious need for unambiguous identification for navigation, but also for current laims and managing a society's past (e.g., to compare the renaming of streets or even entire ing a regime change) [3][4][5][6][7][8]. Toponyms frequently have deeper meanings, often involving semantics related to language and history [9][10][11], but many toponyms also describe the y name. Some example toponyms from Indonesia are derived from folklore tales (Mount perahu, Banyuwangi), historical names (Jakarta from Jayakarta), or names of persons that adjusted to the local language (Malioboro from General Malborough, or Sampur from [12][13][14].…”

Section: Pportunities For New Approaches To Collect Place Namesmentioning

confidence: 99%

A Citizen Science Approach for Collecting Toponyms

Perdana

Ostermann

2018

IJGI

View full text Add to dashboard Cite

Abstract:The emerging trends and technologies of surveying and mapping potentially enable local experts to contribute and share their local geographical knowledge of place names (toponyms). We can see the increasing numbers of toponyms in digital platforms, such as OpenStreetMap, Facebook Place Editor, Swarm Foursquare, and Google Local Guide. On the other hand, government agencies keep working to produce concise and complete gazetteers. Crowdsourced geographic information and citizen science approaches offer a new paradigm of toponym collection. This paper addresses issues in the advancing toponym practice. First, we systematically examined the current state of toponym collection and handling practice by multiple stakeholders, and we identified a recurring set of problems. Secondly, we developed a citizen science approach, based on a crowdsourcing level of participation, to collect toponyms. Thirdly, we examined the implementation in the context of an Indonesian case study. The results show that public participation in toponym collection is an approach with the potential to solve problems in toponym handling, such as limited human resources, accessibility, and completeness of toponym information. The lessons learnt include the knowledge that the success of this approach depends on the willingness of the government to advance their workflow, the degree of collaboration between stakeholders, and the presence of a communicative approach in introducing and sharing toponym guidelines with the community.Keywords: citizen science; volunteered geographic information (VGI); toponym; crowdsourced data collection; data quality Opportunities for New Approaches to Collect Place NamesPlace names (known as toponyms) are an indispensable component of our communication about geographic features or regions, both natural and man-made [1,2]. They serve many purposes, including the obvious need for unambiguous identification for navigation, but also for current territorial claims and managing a society's past (e.g., to compare the renaming of streets or even entire cities following a regime change) [3][4][5][6][7][8]. Toponyms frequently have deeper meanings, often involving complicated semantics related to language and history [9][10][11], but many toponyms also describe the features they name. Some example toponyms from Indonesia are derived from folklore tales (Mount Tangkubanperahu, Banyuwangi), historical names (Jakarta from Jayakarta), or names of persons that have been adjusted to the local language (Malioboro from General Malborough, or Sampur from Zandvoort) [12][13][14]. Other (natural) features can cross multiple linguistic regions, for example, the river "Danube" has several names: "Donau" in Germany and Austria, "Dunaj" in Slovakia, "Duna" in Hungary, "Dunav" in Croatia and Serbia, "Dunav" and " On the other hand, government agencies p working to produce concise and complete gazetteers. Crowdsourced geographic information d citizen science approaches offer a new paradigm of toponym collection. This paper addresses ues ...

show abstract

“…Adding the ability to distinguish between different types of geographical features when extracting place-name entities can, we propose, enable separable analyses of, for instance, lakes, rivers, buildings or regions of widely differing geographical scopes, scales and spatial dimensions. Finally, especially in historical corpora, issues of simple spelling variation in place-names, to say nothing of archaic, obsolete and other sorts of place-name variations, can cause significant problems for conventional NER and NED methods, as these methods tend not to recognize non-standard spellings of place-name entities [1,2,20]. To rectify this problem, we propose that the modern standard spelling of a place-name might be used as a primary identifier (or 'parent'), and that all known alternative forms of that place-name be linked to this parent using a unique numeric identifier.…”

Section: Introductionmentioning

confidence: 99%

A deeply annotated testbed for geographical text analysis

Rayson

Reinhold

Butler

et al. 2017

Proceedings of the 1st ACM SIGSPATIAL Workshop on Geospatial Humanities

View full text Add to dashboard Cite

This paper describes the development of an annotated corpus which forms a challenging testbed for geographical text analysis methods. This dataset, the Corpus of Lake District Writing (CLDW), consists of 80 manually digitised and annotated texts (comprising over 1.5 million word tokens). These texts were originally composed between 1622 and 1900, and they represent a range of different genres and authors. Collectively, the texts in the CLDW constitute an indicative sample of writing about the English Lake District during the early seventeenth century and the early twentieth century. The corpus is annotated more deeply than is currently possible with vanilla Named Entity Recognition, Disambiguation and geoparsing. This is especially true of the geographical information the corpus contains, since we have undertaken not only to link different historical and spelling variants of place-names, but also to identify and to differentiate geographical features such as waterfalls, woodlands, farms or inns. In addition, we illustrate the potential of the corpus as a gold standard by evaluating the results of three different NLP libraries and geoparsers on its contents. In the evaluation, the standard NER processing of the text by the different NLP libraries produces many false positive and false negative results, showing the strength of the gold standard.

show abstract

Toponym disambiguation in historical documents using semantic and geographic features

Cited by 23 publications

References 15 publications

A Coherent Unsupervised Model for Toponym Resolution

A Coherent Unsupervised Model for Toponym Resolution

A Citizen Science Approach for Collecting Toponyms

A deeply annotated testbed for geographical text analysis

Contact Info

Product

Resources

About