Entity Resolution in Texts Using Statistical Learning and Ontologies

Štajner, Tadej; Mladenić, Dunja

doi:10.1007/978-3-642-10871-6_7

Cited by 16 publications

(15 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…One of the reasons is that the chosen news domain is ill-suited for this approach as described in section 6.3. Improvement potential lies in the way we compute the link weights: instead of using the Clusterscore, other metrics such as the Relatedness metric could be more fruitful [15].…”

Section: Discussionmentioning

confidence: 99%

“…It would be worthy to conduct further evaluations or introduce alternative link weighting (e.g. relatedness [15]). …”

Section: Interpretation Of the Ranking Metricsmentioning

confidence: 99%

“…In [15] the Instance Participation Selectivity metric is extended and applied to named entity recognition in text documents. Starting from existing disambiguated entities, this metric is used to identify these candidate entities with the highest probability according to the document context.…”

Section: Link-based Ranking Approachesmentioning

confidence: 99%

“…In many cases it would be useful to have a more informative statement about the strength of a connection in order to be able to assign weights to the edges in the knowledge graph. We have identified a series of ranking metrics from the literature that haven been evaluated in different semantic-based applications [12] [10] [15]. These metrics can be combined in order to provide a more flexible ranking approach that can be adapted to different scenarios.…”

Section: Link-based Ranking Approachesmentioning

confidence: 99%

See 3 more Smart Citations

Improving Semantic Search through Entity-Based Document Ranking

Großmann¹,

Todor

Paschke

2015

Proceedings of the 5th International Conference on Web Intelligence, Mining and Semantics

View full text Add to dashboard Cite

Traditional keyword-based IR approaches take into account the document context only in a limited manner. In our paper we present a novel document ranking approach based on the semantic relationships between named entities. In the first step we annotate all documents with named entities from a knowledge base (for example people, places and organisations). In the next step these annotations in combination with the relationships from the knowledge base are used to rank documents in order to perform a semantic search. Documents that contain the specific named entity that was searched for as well as other strongly related entities, receive a higher ranking. The inclusion of the document context in the ranking approach achieves a higher precision in the Top-K results.

show abstract

Section: Discussionmentioning

confidence: 99%

“…It would be worthy to conduct further evaluations or introduce alternative link weighting (e.g. relatedness [15]). …”

Section: Interpretation Of the Ranking Metricsmentioning

confidence: 99%

Section: Link-based Ranking Approachesmentioning

confidence: 99%

Section: Link-based Ranking Approachesmentioning

confidence: 99%

See 2 more Smart Citations

Improving Semantic Search through Entity-Based Document Ranking

Großmann¹,

Todor

Paschke

2015

Proceedings of the 5th International Conference on Web Intelligence, Mining and Semantics

View full text Add to dashboard Cite

show abstract

“…We carry out experiments with collective entity resolution Štajner and Mladenić 2009) and take into account entity co-occurrence across the documents. All references within the same document are related to each other by a co-occurrence relationship.…”

Section: Collective Er With Co-occurrences Of Referencesmentioning

confidence: 99%

Population Reconstruction

Kirby¹,

Carson²,

Dunlop³

et al. 2015

View full text Add to dashboard Cite

PrefacePeople shape societies. They are linked to each other by family ties and networks with social, economic and religious dimensions. People live together in households and form communities. Some own a house, land and other properties, often related to their profession. And all this is in continuous change. People are born, marry, have children and die, and they change houses and addresses, and build careers. For the study of a society in all aspects, people are at the heart of the problem and should be known in the context of their complex relationships. Even today, it is not easy to get this information in an all-enfolding way, but for populations in the past, it is a real challenge. And that is what this book is about. The book addresses the problems that are encountered, and solutions that have been proposed, when we aim to identify people and to reconstruct populations under conditions where information is scarce, ambiguous, fuzzy and sometimes erroneous.It is not a single discipline that is involved in such an endeavour. Historians, social scientists, and linguists represent the humanities through their knowledge of the complexity of the past, the limitations of sources and the possible interpretations of information. The availability of big data from digitised archives and the need of complex analyses to identify individuals require the involvement of computer scientists. With contributions from all these fields, often in direct cooperation, this book is at the heart of digital humanities and hopefully a source of inspiration for future investigations.The process from handwritten registers to a reconstructed digitised population has three major phases which shape the three sections of this book. The first phase is that of data transcription and digitisation while structuring the information in a meaningful and efficient way. Little of this phase can be automated. With archives that comprise easily tens of millions of records, the help of volunteers for transcription and digitisation is indispensable, but requires a rigorous management. Experiences from Denmark demonstrate the complexity of this task in Chap. 1. Spelling variation, aliases, abbreviations, errors and typos all generate difficulties in further processing and require data cleaning. Similarity measures can be helpful to v

show abstract