AKTors.org
AKTors.org AKTTechnologiesPublicationsRelated ProjectsPeople
AKTors.org AcquisitionModellingRetrievalReusePublishingMaintenance
Melita from The University of Sheffield

The Semantic Web is dependant on the production of machine-readable content, but this is extremely laborious to produce. The is a growing need for automated document annotation. Melita is a semi-automatic annotation tool that has an Adaptive Information Extraction engine (Amilcare ) integrated in it. Its purpose is to support the user in the process of annotation. Melita aims to gradually change the role of the user from one of annotator to one of supervisor. The system is pro-active in the sense that it takes the initiative to do any pre-processing which will be used in the future. The novelty of Melita is the possibility of tuning the Adaptive Information Extraction system so as to provide the desired level of pro-activity and intrusiveness.

Melita fact-file

Owner  :  The University of Sheffield
Researchers
(listed alphabetically)		  :  Dr Fabio Ciravegna [Browse, RDF], Mr Alexiei Dingli [Browse, RDF]
Description  :  http://www.dcs.shef.ac.uk/%7Ealexiei/Melita
Screencam  :  http://www.aktors.org/technologies/melita/melshort3.html
Builds on  :  Java, Amilcare
Addresses challenges  :  Knowledge Acquisition

What's the Problem?

  • Machine readable content is needed for the Semantic Web
  • Most actual or potential users of the Semantic Web are not experts in document annotation
  • Manual annotation is difficult, slow, time-consuming, tedious and costly.
  • There exists a growing necessity of automated support for document annotation.
  • Current IE technologies require skilled human effort for annotation.
  • Many users are knowledgeable about their domain but have limited knowledge when it comes to computing and natural language processing.

Towards a Solution

Melita is a semi-automatic annotation tool that has an Adaptive Information Extraction engine (Amilcare ) integrated in it. Its purpose is to support the user in the process of annotation. Melita aims to gradually change the role of the user from one of annotator to one of supervisor. The system is pro-active in the sense that it takes the initiative to do any pre-processing which will be used in the future. The novelty of Melita is the possibility of tuning the Adaptive Information Extraction system so as to provide the desired level of pro-activity and intrusiveness. This tuning is done by adjusting two slide bars which alter precision and recall, without the user needing to understand these concepts or to have any knowledge about natural language processing. Melita also contains a document sorting mechanism which dynamically sorts documents after every annotation in order to find the document that best covers the unexplored areas of the domain. Documents with the least number of tags are taken to cover unexplored areas of the domain where new rules can be learned if they are annotated. This approach has led to a quicker convergence of the learning algorithm whilst overcoming the problem of data sparseness.

Take a Guided Tour

A General Introduction video, in Shockwave Flash (0.5 Mb).
A Detailed Tutorial video, in Shockwave Flash (1.5 Mb).

Obtaining the Technology.

Please contact the developers, Alexiei Dingli and Fabio Ciravegna.

Technical requirements: Melita is a client-server system so the client requires very low technical specifications. The server requires the same specifications as Amilcare: Windows 2000, XP, Java Runtime Environment 1.3, 512 Mb RAM, 800 MHz Processor

Example Applications

Further Reading

Fabio Ciravegna, Alexiei Dingli, Daniela Petrelli and Yorick Wilks : "User-System Cooperation in Document Annotation based on Information Extraction" in 13th International Conference on Knowledge Engineering and Knowledge Management (EKAW02), 1-4 October 2002 - Sigüenza (Spain)
Available in the eprints archive.

Fabio Ciravegna , Alexiei Dingli , Daniela Petrelli and Yorick Wilks : " Timely and Non-Intrusive Active Document Annotation via Adaptive Information Extraction" in Semantic Authoring, Annotation & Knowledge Markup (SAAKM 2002) , ECAI 2002 Workshop July 22-26, 2002 , Lyon, France

Alexiei Dingli : " Next Generation Annotation Interfaces for Adaptive Information Extraction " in 6 th Annual Computer Linguists UK Colloquium (CLUK03) , January 6-7, 2003 , Edinburgh, UK

Fabio Ciravegna , Alexiei Dingli , Yorick Wilks and Daniela Petrelli :
"Using Adaptive Information Extraction for Effective Human-centred Document Annotation" in R. Skuppin (ed.): Text Mining (preliminary title), book published by Springer Verlag, to appear in 2003

Posters:

Semantic representation

View in the AKT Triplestore Browser or as RDF.

Also available in DOAP RDF (Description Of A Project)