A study of two methods of delivery computer aided diagnostic assistance for turfgrass problems on the web
Issues in Information Systems
International Association for computer Information Systems
The demand for assistance for tu rf diagnosis far outstrips the availability of agricultural agents to offer this service. To meet this de mand, thousands of Web Sites have been developed to assist with the diagnosis and treatment of turf grass problems. The two major types are Web Documentation Sites (DS’s) and Expert Systems S ites (ES’s). Documentation Sites consist of a library of relevant documents with a search engi ne. The user simply uses the search engine to find documents which will help with the problem. Ex pert Systems consist of a user interface, a computer readable knowledge base and an infere nce engine. The user answers questions about problem symptoms; this may include evaluating pi ctures of lawns with various problems. The Expert System then presents a description of the most likely problems and describes treatments. The purpose of this study was to empirically comp are the diagnostic effectiveness of a DS with an ES for turf grass problems. Three groups were used—a control group, a DS group, and an ES group. The Control group received no diagnostic assi stance. (The control was necessary since many people have developed some knowledge of grass problems in caring for their own or others’ lawns). All groups attempted to correct ly diagnose three common lawn problems. The DS group used five excellent Document websites to aid with the diagnosis. The ES group used a well-regarded Expert System. Di agnostic correctness was measured by rating the similarity of the participant’s written diagnosis with correct di agnosis obtained from a Turf Grass expert. The mean diagnostic correctness of the DS group and the mean diagnostic correctness of the ES group were both highly significan tly better than the mean diagnostic correctness of the Control group. There was no significant difference between the mean diagnostic corr ectness of either the DS or the ES group. If this result were to be su pported by further resear ch, the DS method might be effectively used in place of the much more expensive ES method. If an easy to use, highly effective search engine were provided in the DS it might yield better performance than any ES.
Brewer, Harlan; Visonhaler, John; and Johnson, Paul, "A study of two methods of delivery computer aided diagnostic assistance for turfgrass problems on the web" (2006). CWEL Publications. Paper 1.