2. The environment in which AGRIS is operating has understandably changed dramatically since the early seventies. In fact if one takes a closer look at the role of information, knowledge and technology in the development process today one will see that it is becoming increasingly systematic and its role in decision making at all levels is continuously growing in importance.

3. With the recent developments in information technology - such as the advent of Internet - together with recommendations made by the Expert Consultation on the Redesign of AGRIS and CARIS held in January 1998, it is felt that the AGRIS system itself, and its data processing mechanisms, should be readdressed and restructured in order to better integrate new information technologies but also to meet AGRIS’ original objective of developing the Member Nations information management capacities.

4. This paper aims to suggest a technically feasible model for restructuring the AGRIS system and will propose a series of steps and tools which will ease the transition of AGRIS from a centralized data processing system to a decentralized system where each participating centre is in charge and responsible for the processing of the information which it possesses and has collected.

5. A number of tools and processes will be designed to assist the participating centres during this period of reform. It should be stressed that we have and will take into consideration that not all participating centres enjoy the same degree of equipment and technologies, with some centres being in possession of all the most up to date information technology tools and others having a simple PC with which to carry out their work.

6. This proposal aims to give equal opportunities to all countries to play a full role in the future AGRIS as well as facilitating the wider objective of capacity building.

Figure 1 : Current AGRIS Procedures

11. Arguably the most revolutionary development in information development in recent years is the advent of the Internet which, due to its very nature, has opened the doors to the possibility of creating decentralized information systems in an effective and efficient manner.

12. In order to maximise the exploitation of this information technology tool in the context of AGRIS, the AGRIS methodology and architecture for data processing and dissemination needs to be re-evaluated and decentralized. Such a restructuring will greatly increase the data and information flows between AGRIS and the participating centres, between the members of the network itself and between the AGRIS network and end-users. Furthermore the benefits of increasing and facilitating information flows, generated from such structural reforms, will be mostly felt in isolated areas where access to information is most limited.

13. In order to facilitate this move towards a decentralized approach, the participating centres will be equipped with new modern information management tools and methodologies to be used in support of their own national agricultural information systems, using national means and capacities. External assistance will only be sought for the initial transfer of knowledge, technology, and training. The type of tools provided to each participating centre will depend on what technologies each centre possesses, ensuring that each centre’s needs are fully and satisfactorily addressed. Each participating centre will be provided with a customized package, called the AGRIS Information Server (AIS), that will enable the participants to input, manage, and disseminate AGRIS-related information during the transitional period. The AIS will be constructed using Internet-based technologies, but will also support use on a stand-alone basis. The proposed new architecture, presented in this paper, will be easy to use and applied during the entire transitional process thereby guaranteeing full transparency.

14. The aim of the new AGRIS data processing architecture is to provide the participating centres with tools so that they are able to analyse, process and disseminate their own data. The participating centres will assume the responsibility for the correctness and relevance of the data and for all data preparation and processing. Initially, AGRIS (through FAO) will continue to provide the necessary means for data processing, but this will increasingly be shifted from FAO to the participating centres and AGRIS will provide the participating centres with a version of the AIS that can be used for data preparation and processing purposes.

15. FAO/AGRIS will continue to be responsible for the dissemination of AGRIS data. In exceptional circumstances participating centres, in need of real assistance, will be given the option to use the AGRIS system’s dissemination modules to disseminate their own data which will enable them to render their information resources available to the end-users.

16. The two main modules of the system are the "Data Entry Layer", which is the interface for the agricultural analyst to analyse and enter data according to AGRIS metadata rules, and "The Dissemination Layer", which disseminates the processed information through different media. (See Figure 2)

• Data Entry Layer: This component is intended for the user who prepares and enters AGRIS data. Its main function is to provide an intuitive and friendly Graphical User Interface (GUI) and to perform on-line automatic checking of the entered data. The on-line check performs two tasks. The first one is the formal check, which checks that data has been entered according to the AGRIS metadata rules (or data entry rules) and the second one is the authority check, which consists of some look-up tables such as country codes, subject categories, AGROVOC, language codes, etc. For the implementation of the checking mechanism an easy but powerful Metadata Description Language (MDL) will be designed, which will be able to represent all AGRIS metadata rules. The MDL interpreter will apply all the rules described in the MDL-Base and generate an error list that will guarantee the consistency and quality of the data.
• Dissemination Layer: Once information has been entered into the data pool successfully, it will pass to the Dissemination Layer to be processed so as to render AGRIS information available to end-users. This layer generates data outputs for searching purposes. The information processed by the Dissemination Layer will be inserted into a database which will be accessible via the Internet, on a CD-ROM, or a special output format that can be used for loading the AGRIS database into an external system.

The transition to the envisaged future AGRIS system has been subdivided in three stages:

• the first phase is to ensure that the existing AGRIS activities will continue without interruption;
• the second phase will involve installing and implementing tools, procedures and mechanisms to guarantee a smooth transition from the existing AGRIS to a decentralized system;
• the third phase will be the achievement of real decentralization using especially designed tools for assisting the participating centres to reach this goal.

17. AGRIS Information Servers (AIS) will be customized and applied during these three phases in order to meet the needs, assist and empower the participating centres during the transitional period.

18. Three versions of the AIS will be produced. Versions 1 and 2 will mainly be employed during the first two transitional stages (each participating centre applying the AIS which best suits its needs) while version 3 should be regarded as the optimum model, which will fully satisfy the needs of an effective and efficient decentralized AGRIS system operating at the centre of a possible future AGRIS Network (as described in GIL:AGRIS/CARIS/TC/4/2).

19. The three versions of the AIS system are:

1. Central AGRIS System - The participating centres which are connected to Internet will have access to the AGRIS main server and will therefore use the Central AGRIS Sstem. Access to this version of the AIS will only require the ability to access the Internet and a Web browser. Whoever has authority to process AGRIS data can use the data entry layer to enter new data into the system; the Browser is the main interface for the users of this system where data entry and retrieval can be performed and most importantly under this architecture, the dissemination of AGRIS data is literally instantaneous. In fact the dissemination of all AGRIS information will be done automatically by the main server allowing immediate access to AGRIS users. The Data Entry and Dissemination Layers operate on a central server, which can be accessed by all end-users and the system’s search option is open to the general public.
2. Stand-alone AGRIS System - The participating centres which do not have access to Internet will be provided with a stand-alone version of the AIS in which they can input and process their own data. The system architecture of this AIS is similar to the complete version with some restrictions which enables it to run on a stand-alone PC. The processed data will then be transferred to the AGRIS main server for dissemination purposes. However it will contain all checking modules. The UNDP is running a programme called SDNP which aims to provide connectivity to those who wish to gain access to Internet, even those in the most remote areas, and is at the forefront of activities to ensure that everyone in the near future will be offered the possibility of accessing Internet.
3. Distributed AGRIS System - The participating centres which are in possession of their own Web-Server will be equipped with a version of the distributed AIS that permits them to input and process their own data locally. The participating centres will be provided with the data entry module, authority files and data entry rules. If they also possess dissemination facilities, they can disseminate their own data. If not, data dissemination will be achieved through the AGRIS Main Server. This version of the AIS will be a distributed application designed to run in different environments. The data entry module of this version will run on the client side (i.e. the participating centre) and the production module will run on a main server. The two modules will communicate via a "Transmission Layer" which will act as a bridge between the participating centres and the AGRIS Server. The transmission layer monitors the participating centres’ data pool, looking for new entries to process. When new records are found, the data is automatically transferred to the AGRIS main server. After the transfer has been completed, a subsequent module will be triggered, that will pick up the transferred data and prepare it for dissemination. For each country, this should be seen as the first major milestone towards an eventual national agricultural information system.

20. The future structure of AGRIS is based on a decentralized model employing Internet technologies and new information technologies with much emphasis given to developing the capacities of Member Nations in information management in the light of the recognition of the importance of access to information in achieving sustainable agricultural development.

21. In order to favour this process and in order to ensure its success in an ever evolving future, the AGRIS system itself needs to be modified to comply with this process.

22. The decentralization of the data processing mechanisms from one central processing unit to the participating centres themselves fully meets the two-fold objective of the decentralization of AGRIS and capacity building in the Member Nations in information management.