A Case for Automated Large Scale Semantic Annotation

Dill, Stephen; Eiron, Nadav; Gibson, David; Gruhl, Daniel; Guha, Ramanathan V.; Jhingran, Anant; Kanungo, Tapas; McCurley, Kevin S.; Rajagopalan, Sridhar; Tomkins, Andrew; Tomlin, John A.; Zien, Jason Y.

doi:10.2139/ssrn.3199010

Cited by 13 publications

(18 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Even though it is not possible to draw a detailed parallel with other systems in the literature, due to the lack of homogeneity in data sets, domains, computing measures and used resources, a few conclusions can be drawn from a qualitative comparison with related approaches: 1. The systems described in [8][9][10][11] do not perform any syntactic analysis. The precision obtained in [9] is similar to ours, but in [8], precision and recall are lower.…”

Section: Resultsmentioning

confidence: 99%

“…Pattern-based methods can be further divided into two subgroups: those in which extraction rules arise from an initial set of tagged entities [3][4][5], and those in which extraction rules are manually defined [6,7]. Probabilistic machine learning methods rely on statistical models to predict the location of entities in texts [8][9][10][11], while inductive machine learning methods produce entity recognition rules from the syntactical analysis of texts [12][13][14][15][16].…”

Section: Related Researchmentioning

confidence: 99%

“…Regarding the nature of the extracted information, only a few methods [3,9,17] deal with aggregated instances and relations between entities. However, discovered entity associations only involve one aggregation level, that is, they do not attempt to find entity compositions through relationship chains, logic constructors (e.g.…”

Section: Related Researchmentioning

confidence: 99%

See 2 more Smart Citations

Generating complex ontology instances from documents

Dánger¹,

Berlanga²

2009

Journal of Algorithms

View full text Add to dashboard Cite

Section: Resultsmentioning

confidence: 99%

Section: Related Researchmentioning

confidence: 99%

See 1 more Smart Citation

Generating complex ontology instances from documents

Dánger¹,

Berlanga²

2009

Journal of Algorithms

View full text Add to dashboard Cite

“…The Semtag and Seeker system [10] uses the domain ontologies extracted from automatic wrappers [16] to annotate Web pages. Such systems extract relationships and facts that match these patterns from the natural language text segments of the Web pages.…”

Section: Related Workmentioning

confidence: 99%

“…Our system works without the requirements that (a) the Web pages need to share a similar presentation template or (b) that they need to share the same set of metadata among each other. Hence, we cannot readily use previously developed wrapper induction techniques [4,16,18] which require that the item pages should be template driven or the ontology driven extraction techniques [8,10,11] which require that an ontology of concepts, relationships and their value types is provided apriori in order to find matching information.…”

Section: Introductionmentioning

confidence: 99%

Information Extraction from Web Pages Using Presentation Regularities and Domain Knowledge

2007

View full text Add to dashboard Cite

World Wide Web is transforming itself into the largest information resource making the process of information extraction (IE) from Web an important and challenging problem. In this paper, we present an automated IE system that is domain independent and that can automatically transform a given Web page into a semi-structured hierarchical document using presentation regularities. The resulting documents are weakly annotated in the sense that they might contain many incorrect annotations and missing labels. We also describe how to improve the quality of weakly annotated data by using domain knowledge in terms of a statistical domain model. We demonstrate that such system can recover from ambiguities in the presentation and boost the overall accuracy of a base information extractor by up to 20%. Our experimental evaluations with TAP data, computer science department Web sites, and RoadRunner document sets indicate that our algorithms can scale up to very large data sets.

show abstract

Towards Large Scale Semantic Annotation Built on MapReduce Architecture

Laclavik

Seleng

Hluchý

2008

Computational Science – ICCS 2008

View full text Add to dashboard Cite

Abstract. Automated annotation of the web documents is a key challenge of the Semantic Web effort. Web documents are structured but their structure is understandable only for a human that is the major problem of the Semantic Web. Semantic Web can be exploited only if metadata understood by a computer reach critical mass. Semantic metadata can be created manually, using automated annotation or tagging tools. Automated semantic annotation tools with the best results are built on different machine learning algorithms requiring training sets. Another approach is to use pattern based semantic annotation solutions built on NLP, information retrieval or information extraction methods. Most of developed methods are tested and evaluated on hundreds of documents which cannot prove its real usage on large scale data such as web or email communication in enterprise or community environment. In this paper we present how a pattern based annotation tool can benefit from Google's MapReduce architecture to process large amount of text data.

show abstract

A Case for Automated Large Scale Semantic Annotation

Cited by 13 publications

References 24 publications

Generating complex ontology instances from documents

Generating complex ontology instances from documents

Information Extraction from Web Pages Using Presentation Regularities and Domain Knowledge

Towards Large Scale Semantic Annotation Built on MapReduce Architecture

Contact Info

Product

Resources

About