The agricultural ontology servicePresentation Transcript
The Agricultural Ontology Service : A Proposal to Create a Knowledge Organisation Framework in the area of Food and Agriculture
FAO’s interest in Knowledge Management
We need to organize our own information production to enhance the productivity of FAO officers in the world
We need to make FAO’s information accessible to Decision Makers and Technical Specialists all over the world
We need to make all relevant Information sources accessible in the World for our member countries and FAO officers.
The main problems
Information retrieval is inaccurate and incomplete.
Little domain interoperability, cross domain searching needs high efforts
The work of knowledge organization is done without coordination and by duplicating efforts
Machine systems are inefficient or tend to be proprietary or closed in a specific application
The new possibilities of the web and the electronic availability of Information Objects are not fully exploited
The Evolution of Knowledge Management Card and computer catalogues Machine indexing and document annotating Human reading, checking and ordering Human indexing and document annotating Web catalogues on the templates of traditional catalogs Machine readable metadata (RDF) Semantic Web Web Pre-Web Full Text Search engines (Semantic text analysis) Full Text Search engines (Statistical text analysis) Implicit knowledge based web portals Formal Knowledge based web portals Ontologies Thesauri, Classification Schemes, Glossaries, Keyword Lists, Text Mining by Robots
What is an Ontology?
It is a buzz word, but also something very simple
Everyone creates “Ontologies”
A directory structure on a hard disk is an ontology
A personal agenda is an ontology
A bibliographical catalogue is an ontology
Humans might survive without ontologies, but humans have intuition
Machine have no intuition
Machine need formal information
Do we need Ontologies?
An ontology is a formal knowledge organization (representation system).
It contains concepts
Information about these concepts
Relations between these concepts
And it contains information about the relations between instances and concepts
Meronymic Relations--used to describe part-of relationships.
Temporal Relations--define the time interval and time point aspects of an ontology.
Spatial Relations--used to describe space relationships between objects.
Influence Relations--expresses an impact or effect one object has on another.
Dependency Relations--describes an objects dependency on another.
Case Relations--provides ability to express a knowledge structure without which the relation cannot exist.
Defining relations in an ontology- Important relationships that should be introduced in the AOS
Sample ontology for crop pest management
In which areas Ontologies can help (1)
Teaching machines to have some intelligence
Automatic indexing and text annotation tools
Better machine translation
Text Mining on the Web (meaning-oriented access)
Full text search engines that create meaningful classification (FAO-Schwartz not related to FAO) (semantic clustering)
In which areas Ontologies can help (2)
Better structure and organization of knowledge on the web
Guided discovery of knowledge
Easy retrievability of information without using complicated Boolean logic
1. To build dynamic taxonomies (ordered classification) through automatic classification operation
2. Support for Natural Language Processing through concept matching and query processing
locate a concept based on a description
find other similar objects in the ontology
Applications of ontologies
3. To aid information retrieval through automatic document clustering
....closely associated documents tend to be relevant to the same requests
Information Management tasks that can be performed using an ontology
1 .Expressing information needs using natural languages
Example: what insects damage soybean leaves?
2. Finding the concept in the ontology
Insect Damage Soybean leaves
3. Analysing the grammatical pattern with in the stated sentence (syntactic analysis)
4. Mapping the grammatical structure into objects in the ontology (semantic analysis)
5. Drawing inferences between the user’s query and objects in the database
6. Displaying the results to the user
Steps in using ontology for Natural Language Processing- a simplified view
AOS: Possible Use... (1) Records found: 5 1 . xxxxxxxxxxx 2 . xxxxxxxxxxx 3 . xxxxxxxxxxx 4 . xxxxxxxxxxx 5 . xxxxxxxxxxx Biotopes Cropping systems using forests Economics of forest production Forestry equipment Soil science You may also be interested in... What would you like to view? Forest rights issues Parasites of forests Pesticides used in forests Types of forest products Uses of forest products Geographic area You can further limit by: x Africa Web page Type of resource
AOS: Possible Use... (2) Conservation agriculture Farmers like it because it gives them a means of conserving, improving and making more efficient use of their natural resources About camels and llamas Descendants of the same rabbit-sized mammal, they have become two of humanity's most versatile domestic animals Agribusiness and small farmers Well managed contract farming contributes to both increased income for producers and higher profits for investors Toward biosecurity Biological and environmental risks associated with food and agriculture have intensified with economic globalization Urban food marketing In the “century of cities”, a major challenge will be providing adequate quantities of nutritional and affordable food for urban inhabitants Crop science and ethics In order to continue their contribution to human development, crop scientists must regain credibility Use your right mouse button to learn more about an italicized word on the page. Biosecurity : management of all biological and environmental risks associated with food and agriculture, including forestry and fisheries See also : Biosafety Food Safety Risk Management Or are you interested in... : Food Security Biological Diversity Agricultural Web Page
What is the starting point?
FAO maintains the multilingual thesaurus AGROVOC since the early 80s
Other consistent thesauri are maintained by CABI in England and the National Agricultural Library in the States
Various other knowledge organization systems are scattered around the world
The existing systems are language biased with English as the leading language
None of the systems is satisfactory for resource description and discovery purposes
Thesauri Contain Knowledge
Thesauri were mostly used only for indexing and to help users in searching
But thesauri are already knowledge organization systems
Not only the vocabulary of concepts, but also the defined relations (BT, NT, RT, UF …) contain domain knowledge
To leverage this knowledge in the context of Web technologies we need to develop them further
The Origin of the AOS - Project
Born as the AGROVOC Taxonomy Server
Agronomists were upset with the word taxonomy
IT people were upset about the word server
After XML2000 the word ontology started to become sexy... And fundable
Now we are in the phase to define the project
Why an Agricultural Ontology Service (1) AGROVOC NAL Thesaurus CABI Thesaurus Dedicated KOSs Non-dedicated KOSs e.g., ASFA thesaurus e.g., the Multilingual Forestry Thesaurus e.g., the Sustainable Development website classification e.g., biological taxonomies such as NCBI and ITIS GEMET Other thematic thesauri Existing Thesauri and Knowledge Organization Systems (KOSs)
Why an Agricultural Ontology Service (2)
FAO can be a neutral point of co-ordination
Central access-point to domain knowledge
Re-usability of domain knowledge
Involvement of a high number of subject specialists
Clear and distributed responsibilities for maintenance
Federation of Institutions to manage this service
Possibility of business plan to assure sustainability
AOS: A reference point on the web Concept Attributes Responsible Party Definitions Labels Relations ?????? URI, e.g., www.agri-ontology.org/2050.xml
An architecture for federated AOS Normalised Ontology Normalised Ontology Merged Ontology- THE AOS Local Ontology Local Ontology Forestry Portal (application on merged ontology) OneFish Portal (application on merged ontology) Crop Portal (application on merged ontology) Common data model for local ontologies Ontology + metadata repository System 1 Ontology + metadata repository System 2
AOS: Possible Use... (3) Create your own ontologies using the AOS The AOS provides the necessary building blocks to create your own ontologies. Follow the following simple steps to create your ontology instantly. Domain Authority Ontology Content & Structure Language Representation
AOS: Iterative Knowledge Registration KOS application KOS uses components to build an application Discussions and choices for amendments to components KOS partner Agricultural Ontology Service (AOS) Federated storage and description facility Components: terms, definitions, relationships KOS partner Components: terms, definitions, relationships Users search and browse application using components User feedback
AOS: where we are
The concept note has found interest in the domain area -- comments are mostly encouraging from all subject specialists, ontology developers and users
A Launch Group has been established at the first workshop. The Launch Group is in charge to write a definite project proposal and to define the possible collaborations and the necessary management structures of the project
A proposal for a fundable project (6 th framework) should be written
Partnership of the most important stakeholders must be established
Pilot projects are ongoing
http://www.fao.org/agris/aos [email_address] [email_address] [email_address] Information and Contacts
Possibility of mapping to register quicker and to accept different terminology
How can local content made available
We need examples about searches, at it is now and as it would be with the AOS.
The usage slides have to be integrated
Introduction of terminology precision…..
Microsoft for funding
We need to do something (once, well, together)
Importance of bibliographical databases: there is a kind of illiteracy between undergraduates who do not consider material that is not online available.
Prototype Searchengine, that searches the web in arabic or chineses, using agrovoc