Expert systems in agriculture domain

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.


The problem of water management in agriculture requires cognition from many disciplines apart from the capability of an individual expert. Computerized tools and systems can give a shoulder to the operation and maintenance of the plantation. The integrated computer systems can combine many areas of human and system cognition to deliver the output in a functional manner. This may help a farmer in predicting the unprecedented events for management of the farm, with an appropriate usage of key components of this efficient system including databases, graphic play, involvement of expert systems and sensors. The penetration of computing systems in farms and water resource management department are in a nascent stage in many countries, and is often done without taking the whole picture, thus the question here is how it can be done with a more holistic approach involving data from various other farms, rivers, dams, weather reports etc. along with the farm related data. To understand the question we have suggested a model of an expert system in order to bring an effective water management approach in a farmer's work, to an extent of making the disguised unemployment in farming redundant. We argue that research in designing and execution of such systems of AI has several benefits:

It can open new channels in field of agricultural research showing new ways of food production like green revolution[1] as in a simulated environment it would be easier to see the impacts of how crossed seeds can work in various farms.


An expert system is a computer application that can solve complicated problems, with the involvement of subject matter experts in a specific domain. These expert systems represent cognition as set of rules or data within a computer. Expert systems are more common in many domains and they are a part of Artificial Intelligence.

An intelligent computer program that uses knowledge and inference procedures to solve problems that was difficult enough to acquire significant human expertise for their solutions (Feigenbaum).

Expert systems have many components associated with it; the knowledge base is a region where the knowledge from the domain expert is saved, the interference engine is the one which manipulates the cognition in the knowledge base and passes to the user interface which would best fit the result. There are three individuals associated with the expert system, the first and the foremost priority is given to the end-users who use this programmed system for problem solving assistance.

The other two roles are in building and maintenance of this system and they are the problem domain expert who plays a major role in building the knowledge base and finally the knowledge engineer who assist the subject matter experts in determining their representation of their knowledge and puts this sort of cognition in to an explanation module and then defines the interference technique required to obtain problem solving activity. In this expert system, knowledge is gathered from the human experts who can't be expected to be with us always. Important case is that one or two experts is not enough for areas like agriculture because so many fields are been covered by one major field and many experts are requires to design a specific expert system for the agriculture domain.

Expert systems are been used in agriculture since 1980s, several systems have been designed in different countries like USA, Egypt and other European countries for diagnosis, management and production aspects. The general features of an expert system are; can assist the farmers to take a onetime decision, and helps them in well planning before they start to do anything on their area. Secondly it should help him in designing an irrigation system for the plantation use. The next thing that it should consider is that to select, the most suitable crop variety for the season or crop suitable for the market outfit. Next is to guide with a set of financial accounts. And then it should be able to predict some disastrous events such as thunderstorms, frost and rain. The last and the foremost thing is to suggest a sequence of decision making throughout the production of a crop such as crop protection and nutrition decision, livestock feeding and so on. Therefore from the above mentioned points, we are clear that the emergence of expert system can help a farmer in a better way than the traditional method of documentation did. 

Need for expert systems in agriculture domain

There is a need for an expert system in the field of agriculture, because there are few problems with traditional system of agricultural management. In this paper we have analyzed few problems and have given few suggestions, on how use of an expert system can overcome those problems [3].

STATIC INFORMATION:  Research on stored and available information in the agriculture domain, brought to the light that the information is static and might not be effective for the growers need. The extension documentations, give a general outline and recommendation because it is quite impossible for it to have all the content in it. Wherein, an expert system produces advices based on its knowledge cases and its reasoning mechanism, which is far behind all the developed extension documents. Moreover when a user feeds the data of his/her crop plantation to the system, appropriate feedback is given, and there are no limitations on the generated recommendations or advices this is how an expert system can manage the problems of static information.

INTEGRATING SPECIALITIES: The extension systems have the capability of handling specific problems related to the specialty for example, entomology, plant pathology, diagnosis, nutrition and some other problems.  There are possibilities that the problem might have occurred due to more than one cause, and requires the collaboration of cognition behind the extension document and books. But an expert system has a knowledge acquisition system, facilitates the integration of subject matter expertise having experience and knowledge in different specialties. For example the agriculture expert system has the cognition of specialists in nutrition, plant pathology, breading and so on. So that we a problem arises the system would be very helpful in identifying the problems in a much more efficient way

COMBINING THE INFORMATION SOURCES: In traditional system i.e. the extension documents matching images, with some factors to reach an accurate diagnosis was quite difficult. Whereas the, expert system is integrated with other resources such as image bases and textual bases to make the work precise. For ex, images can be useful in identifying and describing few symptoms, which may be confusing for a normal farmer if it is in words.

UPDATION: There are many changes in the industry, alterations in the chemical, dozes and the environment should be considered and this information should be often updated whether it is stores in documents or audio tapes or videos which were not available in the traditional system. Somehow the expert systems with knowledge base can be maintained in a better way than a manual document. The problem of updating versions of plantation or agriculture relevant things are eliminated if the expert system is been connected to a network. The undocumented experience and cognition can be acquired and stored in the knowledge base of an expert system for a certain specialty and/or commodity

INFORMATION LAGGING: Sometimes the available or gathered information may not be enough; that sort of information may be available only with experienced farmers and human experts. Since agriculture is the backbone of many countries there is a need of transferring the information from certain experts to the general public of farmers using some new technologies. But the case is that there are very less experts in new technologies than their demand.


In this paper we don't intend to explain the details of the various methodologies for the implementation of expert system, rather we analyze two aspects which would be helpful for farming; they are the methodological aspect and the supplication aspect. Methodology is further categorized in to first generation expert systems and the second generation expert systems. The first generation expert system is based on the use of a commercial system shell, where knowledge is acquired through traditional acquisition and then rapid prototyping is done. The second generation methodology is based on human cognition, which means developing a model at the human level problem clearing approach and not at computational approach. The domain of application aspect is analyzed by taking agriculture as domain and the task type to classify the given application.


Many systems work under the principle of first generation expert systems. We have just analyzed one example that come under the first generation expert systems. An agroforestry expert system (UNU-AES) was invented in order to support growers, research, land use officials and other people who benefited using this system [5]. It was UNU-AES to take a first attempt to apply expert systems to the Agroforestry, this system is mainly used for addressing the option of alley cropping, a purely agroforestry technology. Alley cropping is nothing but to plant the crops in alleys or interspaces between woody perennials. By including more climatic, geographic and socio-economic data, UNU-AES can be used to provide advisory recommendation on alley cropping in more diverse geographical and ecological conditions and this expert system uses EXSYS shell and the documentation of this methodology is based on modified waterfall model [6].

There are few approaches that follow the second generation expert system approach; there are two methodologies namely the KADS and generic task methodology. The cucumber and citrus management expert systems were developed using KADS methodology [7][8], whereas generic task methodology was used to develop expert system for wheat management [9][10]. The second generation expert systems have generic models for different type of tasks such as diagnosis, planning, design and others. In agriculture domain we have two main models namely the diagnosis and scheduling tasks, the KADS methodology provides a library of expertise models for each of these tasks [7], whereas the generic task methodology, the hierarchal classification model is used for the diagnosis and the routine design model is used for scheduling [11]

Application aspect

The field of agriculture can be classified in to many domains: animal production, plant production, management of resources such as land and water. Here in this paper we concentrate on the domain plant production since many of the expert systems are developed in this subdomain.

There is one more way to classify agricultural expert systems by considering the domain specific tasks such as irrigation, fertilization, diagnosis of disease and so on. 

May be ill still take half a page more...... and still I haven't done lot of changes... u proceed with ur work Proposed Model

The proposed model is based on the concept of second generation expert systems and is based on the generic task approach with focus on diagnosis and scheduling concepts. It also makes use of the library of experts defined in the KADS approach. The framework is being created keeping in a more holistic view of being flexible to be open for continual improvement and more importantly to be capable of collaboration with key stakeholders, future expert systems of domains outside the field of agriculture. To attain this, the suggestion is to make use of service oriented architecture (SOA) [12] and cloud computing [13] concepts that can make it available over the web with the benefits of independent code development and reuse opportunities. The initiative is to make use of smart image 'sensor array' [14] to get the data of the soil to be used for scheduling irrigation. The farmer in this part will have a user interface that will be designed with a system, dialogue and tool perspective [15]. By system perspective it is to be understood that both farmer and the expert system will co-ordinate with each other to attain the final stated goal of irrigation. The dialogue perspective has always been associated with almost all expert systems to date. This will include factors like one to one questions that will be prompted to the farmer for the generic task analysis & execution. On completing the analysis the farmer will get indications of the unfinished task that need human intervention. The expert system shall be connected with water pipes across the field and on reading the data provided by the sensors about the current moistness of soil will send an information alert to the user indicating critical situations about different sections of the farm. The process of irrigating the farm then could be automated by using the water pipes. Similar indications based on live data could be 'scheduled' and sent to the user about other generic tasks need like cultivation, ploughing etc. Unlike irrigation such tasks needs a human intervention of the SME, and this is where the expert system is seen as tool perspective for the farmer.

The second part of the expert system will involve an avatar cum gaming model.  Based on the data received from the images of the sensors, a graphical model of the farm both inside the land and outside the land will be reconstructed by using image reconstruction algorithms. Such algorithms are presently kept out of scope of this paper. Once the graphical image is reconstructed the farmer will then be provided with an avatar of himself to play inside the farm. The user provided with a toolbar [15] then will have options of virtually doing all their activities in this gaming environment to visualize the impacts and analysis of various actions that the user might want to choose. All this data would be of status quo to provide a real impact. The user than foreseeing the consequences of various tasks could then take proactive measures and plan accordingly for the time ahead. The gaming model described here is in a nascent independent stage , however going forward it will definitely co-ordinating with applications outside its domain, will require live data feeds, and thus it must be integrated over the web using SOA approaches.

[2] Mackinson, S. 2000. An adaptive fuzzy expert system for predicting structure, dynamics and distribution of herring shoals. Ecological Modelling, 126, 155-178.

[3] Ahmed Rafea, Central Laboratory for Agricultural Expert Systems "Expert System Applications: Agriculture"

[4] Fikret Berkes, Robin Mahon, Patrick McConney, Richard Pollnac, and Robert Pomeroy "Managing Small-scale Fisheries"

[5] Warkentin, M., Nair, P., Ruth, S. and Sprague, K. (1990) . "A Knowledge-Based Expert System for Planning and Design of Agroforestry Systems". Agroforestry Systems 11(1): 71-83.

[6] Rafea, A., El-Dessouki A., Hassan H., Mohammed, S. (1993). Development and Implementation of a Knowledge Acquisition Methodology for CropManagement Expert Systems. Computers & Electronics in Agriculture. 8(2): 129-146

[7] Rafea, A., Edrees, S., El-Azhari, S., Mahmoud M. (1994). A Development Methodologyfor Agricultural Expert Systems Based on KADS. Proceedings of the Second World Congress on Expert Systems.

[8] Rafea, A., El-Azhari, S., Hassan, E.(1995). Integrating Multimedia With Expert Systems For Crop Production Management. Proceedings of the Second International IFAC Workshop on Artificial Intelligence in Agriculture, Wageningen, The Netherlands.

[9] Schroeder, K., Kamel, A., Sticklen, J., Ward, R., Ritchie, J., Schulthess,U., Rafea,A.,Salah,A. (1994). Guiding Object-Oriented Design via the Knowledge Level Architecture: The Irrigated Wheat Testbed. Mathl.Comput. Modeling, 20(8):1-16.

[10] Schulthess,U. (1996). NEPER-Weed: A Picture-Based Expert System for Weed Identification. Agron. J. 88: 423-427.

[11] Kamel, A., Schroeder, K., Sticklen, J., Rafea,A., Salah,A., Schulthess,U., Ward, R. and Ritchie, J. (1994). Integrated Wheat Crop Management System Based on Generic Task Knowledge Based Systems and CERES Numerical Simulation. AI Applications 9(1):17- 27.