Study On Citrus Pests Information Database And Expert System

The control of citrus pest in China is becoming more difficult due to more pest species and more serious damage they caused. In recent years, chemical pesticides caused environment pollution and ecological deterioration to a great extend as the result of that fruit famers lacked enough knowledge of scientific citrus pests control and sufficient guidance from the authorities. In addition, there was not enough agro-technique personnel and the way to obtain useful guide information in some citrus-growing area. Under the situation of the low-level pest management and development of productive forces, some dangerous citrus pests had been tending to spread widely. So it is necessary to develop a platform for the information intercommunion and knowledge sharing among fruit farmers, agro-technique people, expert and authorities. The information technologyis applied widly in pest management as well as in other agricultural areas could provide new tools and new resolutions in current situations.The thesis presents the study on information technology used to support the control of citrus pests. The basic information derived from prthe results of the researches and documents was stored in and managed by SQL Server 2005. The domain knowledge of the citrus pests were organized into ontology with the ontology edit platform Protege 4.0. Based on the ontology,an initiatory inference machine was developed by Jena to build a primary expert system for citrus pest diagnosis which could assist users to solve the problems of citrus pests control. The main results were showed as follows:1.The information database of citrus pestsThe information database of citrus pests included the entries of citrus pests, pictures. Acquiring pests'information and online browsing pests'pictures in this system were realized network application base on internet technology. Therefore, users could simply acquire pest information they wanted and operate expert system via the internet at any time. The database contained entries of 187 species,52 families,11 orders,3 classes(Insecta, Gastropoda, Arachnoidea) and 2 phylums(Mollusca, Arthropoda). The picture subdatabase contained 381 micro-graph and natural photo pictures covered 124 species.2. Morphological and damage symptoms dynamic retrieval of citrus pestsA key table of morphology and a key table of damage symptoms were built, which were linked to each collected pest and compiled with binomial formula to improve the system with practicality and professionality. Especially, the key table of damage symptoms was implemented according to damage location—leaf, flower and fruit, limb and root, budlet, while the key table of morphological according to taxonomic such as specy, family, order, class and phylum. On the basis of the consulting rules and method of classification codes for vertebrate in China, every retrieval entry in the key tables was encoded and then was stored in the database. The key table of morphology was encoded with 16 numbers and its specific form was phylum code+class code+subclass code+ order code+suborder code+superfamily code+family code+genus code+entry number. The 2-numbers code started from 01 and the one-number code from 1. Fields of the table which stored retrieval entry were designed as follows:ID, cname, lname, description, image_url, left_ID, right_ID. The key table of damage symptoms was encoded with 5 numbers and its specific form was as follows:location code+ entry number. Fields of the table were defined as follows:ID, diagnosis, symptom, left_ID, right_ID, image, imagetype. Responsing to of the entry selected by users, the corresponding content would be displayed dynamically for user to find out the satisfied target entry. Specially, the entry of the family which only contained one specy will be directly corresponded to the entry of that specy. Only the single entry or a set of entries of specy could be retrieved in in the key table of damage symptoms. With the help of Web technology, function that online morphological retrieval and damage symptoms retrieval was successfully achieved.3. Ontologies of morphology identification and harmful diagnosis of citrus pestsBase on consulting " seven_step" method developed by Stanford University, the development process of ontology was extended to 8 steps in the thesis. The eighth step was to consummate and evolve the ontology. According to the key table of morphology, 664 concepts were extracted involving 621 classes and 43 properties. Five top classes were defined as follows:Mollusca, Arthropoda, Organs, Corpus and Identification_features, which covered 90% pests in the entry database of citrus pests. The citrus pests ontology of harmful diagnosis was built according to the key table of damage symptoms. The ontology contained 231 classes and 19 properties(?)Five top classes were defined as pests, damage location, damage stage, damage phases and damage symptoms in this ontology. The two ontologies were created and edited with the Protege 4.0 and saved as OWL file.4. The diagnosis expert system of citrus pestsA network-based expert system would assist user to diagnose the damage by some uncertain damage symptoms. The diagnosis rule and knowledge bases were builted based on the ontologies of the morphology and damage symptons which had been integrated byJena to implement the inference process. The system will provide brief summary of damage symptoms and diagnosis result inferred by the information of typical characters or combination charactersthat were selected from the user interface.