Applied Research in Science and Technology 4 .
: 61-76 2024 Contents lists available at openscie.
E-ISSN: 2776-7205
Applied Research in Science and Technology DOI: 10.
33292/areste.
Journal homepage: https://areste.
org/index.
php/oai Web-Based Expert System for Dragon Fruit Disease Diagnosis Using Bayes Method Dionisius Raffi Koa1.
Sisilia Daeng Bakka Mau1.
Alfry Aristo Jansen Sinlae1* Computer Science Study Program.
Faculty of Engineering.
Widya Mandira Catholic University.
Indonesia *Correspondence E-mail: alfry.
aj@unwira.
ARTICLE INFO
ABSTRACT
Article History:
Received 30 October 2024 Revised 21 November 2024 Accepted 27 December 2024 Published 29 December 2024 Dragon fruit cultivation in Kampung Daun Baumata has faced significant challenges due to plant diseases, with farmers reporting a 15% yield reduction in 2022.
This study addresses this critical agricultural problem by developing an innovative web-based expert system that utilizes Bayesian probability theory for accurate and timely disease diagnosis.
The system provides farmers and agricultural stakeholders with an accessible digital tool to identify common dragon fruit diseases, including stem rot, anthracnose, and fungal infections, through symptom analysis and probability Implemented using PHP programming language and MySQL database, the expert system offers several advantages over traditional diagnostic methods.
It operates independently of human experts, delivers real-time results, and provides prevention The Bayesian approach enables the system to process uncertain information and update disease probabilities as new symptom data becomes available, significantly improving diagnostic accuracy compared to conventional methods.
Field testing demonstrates the system's effectiveness in supporting farmers' decision-making processes, enabling early disease detection, and facilitating appropriate treatment measures.
The implementation of this technological solution has the potential to reduce economic losses, improve crop yields, and promote sustainable farming practices in dragon fruit cultivation.
bridging the gap between farmers and agricultural expertise, this research contributes to the digital transformation of agricultural disease management in developing regions.
Keywords:
Agricultural Technology.
Bayesian Method.
Dragon Fruit Disease.
Expert System.
Web Application.
To cite this article: Koa.
Mau.
Sinlae.
Web-based expert system for dragon fruit disease diagnosis using bayes method.
Applied Research in Science and Technology, 4.
, 61Ae76.
This article is licensed under a Creative Commons Attribution-ShareAlike 4.
0 International License A2024 by author/s Introduction Dragon fruit (Hylocereus spp.
) has emerged as a globally important tropical fruit crop due to its nutritional benefits and economic value (Anitha et al.
, 2023.
Martynez-Rodryguez et al.
, 2022.
RojasSandoval & Praciak, 2021.
Verona-Ruiz et al.
, 2.
As a member of the Cactaceae family, this plant thrives in tropical and subtropical climates, with Indonesia cultivating four main varieties: Hylocereus undatus .
hite fles.
Hylocereus polyrhizus .
ed fles.
Hylocereus costaricensis .
uper red fles.
, and Selenicereus megalanthus .
ellow ski.
(Rojas-Sandoval & Praciak, 2.
The crop's popularity continues to grow, with global production increasing by 8.
7% annually since 2018 (Li et al.
, 2.
In Kupang Regency.
East Nusa Tenggara, the Farmer Group in Baumata Village has developed a 2hectare horticultural farm specializing in dragon fruit cultivation, maintaining 533 red-fleshed dragon fruit plants (H.
However, recent yield declines from 80% in 2021 to 65% in 2022 highlight significant production challenges.
Similar yield reductions have been reported in other tropical regions, primarily due to plant diseases and inadequate cultivation practices (Balendres & Bengoa, 2.
Dragon fruit cultivation faces numerous biotic stresses, including stem rot .
aused by Fusarium spp.
anthracnose (Colletotrichum gloeosporioide.
, and fungal infections (Bipolaris cactivor.
, which can reduce yields by 15-30% (S, 2.
Farmers often struggle with disease identification, typically relying on visual inspection without proper diagnostic tools, leading to inappropriate treatments that may exacerbate plant damage (Balendres & Bengoa, 2019.
Indumathi et al.
, 2.
This problem is particularly acute in developing regions where access to agricultural experts is limited (Verona-Ruiz et , 2.
The Baumata farmer group exemplifies these challenges, reporting difficulties in obtaining expert guidance for disease management.
This knowledge gap results in ineffective treatments and significant economic losses.
Similar situations have been documented across Southeast Asia, where smallholder farmers lack access to timely and accurate plant disease diagnostics (Drury et al.
, 2017.
Kamrani et al.
Recent advances in agricultural technology suggest that web-based expert systems can bridge this knowledge gap.
Bayesian probability models have shown particular promise in plant disease diagnosis, achieving 85-92% accuracy in similar applications (Drury et al.
, 2017.
Hendra Kurniawan, 2021.
Indumathi et al.
, 2025.
Li et al.
, 2011.
Maulana et al.
, 2023.
Ramadhan, 2020.
Riyawan et al.
, 2.
Such systems can process uncertain information and provide probabilistic diagnoses, making them ideal for resource-limited agricultural communities (Drury et al.
, 2017.
Indumathi et al.
, 2.
This study addresses these challenges by developing a web-based expert system incorporating Bayesian probability to improve dragon fruit disease diagnosis in Baumata Village (Kabu et al.
, 2023.
Lado et al.
, 2021.
Lelo et al.
, 2023.
Nugroho et al.
, 2.
The system aims to provide farmers with accessible, accurate diagnostic tools, potentially increasing yields by 15-20% through early disease detection and appropriate treatment recommendations (Li et al.
, 2011.
Riyawan et al.
, 2.
Methods This research develops a web-based expert system for diagnosing dragon fruit plant diseases by adapting a generic expert system architecture consisting of two main environments: the development environment and the consultation environment, as shown in Figure 1.
Consultation environment Development environment User Knowledge Base (KB) Knowledge engineer Knowledge Knowledge base editor User interface Expert knowledge Inference engine Explanation system Case specific Shell Figure 1.
Expert System Architecture The following is an explanation of each stage shown in Figure 1:
System Development Environment The system development phase begins with knowledge acquisition through in-depth interviews with dragon fruit agricultural experts in Kupang Regency and literature studies of journals related to Hylocereus spp.
diseases published in the last 5 years.
The data collected includes:
8 main disease types: Anthracnose Fruit Rot (Colletotrichum gloeosporioide.
Bacterial Soft Rot (Erwinia carotovor.
Stem Base Rot (Neoscytalidium dimidiatu.
Fusarium Wilt .
r Fusarium Rot/Fusarium oxysporu.
White Grub .
arva of scarab beetles/Leucopholis spp.
Stem Soft Rot (Bipolaris cactivor.
Orange Fruit Spot (Curvularia lunat.
, and Scab (Elsinoy ) as systematically classified in Table 1.
Table 1.
Dragon Fruit Disease Coding System Code Disease Name Anthracnose Fruit Rot Bacterial Soft Rot Stem Base Rot Fusarium Wilt White Grub Stem Soft Rot Orange Fruit Spot Scab Our study successfully documented 32 distinct symptoms associated with various diseases affecting dragon fruit plants (Hylocereus spp.
) as comprehensively cataloged in Table 2.
These symptoms were systematically observed and validated through field surveys conducted in dragon fruit plantations across Kupang Regency.
East Nusa Tenggara.
Indonesia.
Table 2.
Disease Symptom Coding for Dragon Fruit Plants Code
Symptom Name
Code
Premature G17
Shade-affected rot
G18
Fruit stunting G19
Symptom Name
Yellow mucoid stem rot
Stem base rot
White stem tomentum
G10
G11
G12
G13
G14
G15
G16
Fruit desiccation White/brown fruit spots Sunken orange fruit lesions Concentric black fruit rings Soil hyperhydration Circumferential stem rot
Plant dullness Golden-yellow stem rot
Unilateral stem rot
Wet rot with dry margins Insect-mimic stem lesions Plant wilting Wet rot
G20
G21
G22
G23
G24
G25
G26
G27
G28
G29
G30
G31
G32
Productive tip rot Dry rot Productive tip wrinkling Brown tip spotting Brown stem rot Root rot Growth retardation Fetid stem decay Stem soft rot Stem browning Rust-orange fruit spots Brown watery spots Reddish tendril roughness The disease probability weights presented in Table 3 were rigorously developed through a structured expert elicitation process involving a plant pathologists specializing in tropical fruit These weights represent the conditional probabilities of disease occurrence given observed symptoms in dragon fruit (Hylocereus spp.
), serving as critical parameters for our Bayesian inference model.
Table 3.
Disease Weight Values Based on Expert Knowledge Code Disease Name Anthracnose Fruit Rot Bacterial Soft Rot Stem Base Rot Fusarium Wilt White Grub Stem Soft Rot Orange Fruit Spot Scab Weight The knowledge base in this expert system is built upon three core components: .
facts represented by clinical symptoms .
, brown spots on stem.
, .
IF-THEN rules linking symptoms to specific diseases .
, "IF stem exhibits brown discoloration AND soft texture THEN diagnose stem rot"), and .
Bayesian methods calculating disease probabilities based on observed symptom combinations.
The system undergoes continuous knowledge refinement through validation by plant pathologists and accuracy testing against historical disease cases.
When discrepancies arise between system diagnoses and field observations, knowledge engineers adjust the rules or probability values to enhance precision.
The Bayesian probability values in Table 4 quantify the likelihood of eight dragon fruit diseases (P1-P.
based on 32 observed symptoms (G1-G.
This probability matrix reveals distinct diagnostic patterns: .
Anthracnose (P.
shows high probabilities .
for fruit-related symptoms (G1-G.
Bacterial Soft Rot (P.
is strongly associated with stem conditions (G8G14, 0.
Stem Base Rot (P.
links to root/stem decay (G15-G19, 0.
Other diseases demonstrate specific symptom correlations like Fusarium (P.
with branch tips (G20G.
and Stem Soft Rot (P.
with foul odor/texture (G27-G.
Notably, symptoms with probabilities Ou0.
7 serve as key diagnostic markers, while lower values .
indicate negligible associations.
This structured probability framework enables accurate disease prediction by weighting symptom combinations, where higher values reflect stronger pathological relationships validated through field studies.
Table 4.
Bayes Disease Probability Value from The Symptoms that Dragon Fruit Plants have Symptoms G10
G11
G12
G13
G14
G15
G16
G17
G18
G19
G20
G21
G22
G23
G24
G25
G26
G27
G28
G29
G30
G31
G32
System Consultation Environment The consultation environment provides an interactive web-based platform where farmers input observed symptoms to receive real-time disease diagnoses.
Designed for accessibility, the system features: .
symptom selection menus with visual guides .
, image-based symptom identificatio.
, .
automated Bayesian probability calculations that weigh symptom combinations against the knowledge base, and .
diagnostic reports displaying ranked disease probabilities with management recommendations.
Bayes' Theorem provides a mathematical foundation for predicting the likelihood of plant diseases .
based on observed symptoms .
In our dragon fruit expert system, this theory calculates the probability of specific diseases when certain symptoms are detected in the field.
The system implements three key applications:
Basic Case (Single Symptom - Single Diseas.
Formula 1:
ycy ( ya .
a ) O ycy ( ya ) ycy ( ya .
a ) = ycy .
a ) ycy.
= probability of hypothesis H occurring if evidence E occurs ycy.
= the probability of evidence E, if hypothesis H occurs = the probability of hypothesis H regardless of any evidence ycy .
a ) = the probability of evidence E regardless of anything ycy .
a ) Multiple Diseases (Single Symptom - Multiple Disease.
Formula 2:
aycn ) O ycy.
aycn ) ycy.
= ycu Ocyco=1 ycy.
ayco ) O ycy.
ayco ) ycy yaycn ya = probability of hypothesis Hi occurring if evidence E occurs ycy.
aycn ) = the probability of the emergence of evidence E, if the Hi hypothesis occurs ycy.
ayco ) = the probability of the emergence of evidence E, if the Hk hypothesis occurs ycy.
ayco ) = the probability of hypothesis Hk regardless of any evidence = number of possible hypothesis This formula compares probabilities across several diseases for one symptom.
Complex Cases (Multiple Symptoms - Multiple Disease.
Formula 3:
aycn ) O ycy.
aycn ) O A .
O ycy.
aycn ) O ycy.
aycn ) ycy.
a1 ya2A.
yayco ) = ycu Ocyco=1 ycy.
ayco ) O ycy.
ayco ) O A .
O ycy.
ayco ) O ycy.
ayco ) Used when analyzing symptom combinations (ECA,ECC,.
EC.
against multiple diseases.
Bayesian Disease Diagnosis Procedure for Dragon Fruit Plants This systematic diagnostic approach employs Bayes' Theorem to accurately identify diseases based on observed symptoms, following three key operational stages:
Step 1: Symptom Identification and Input The diagnostic process begins with systematic observation and recording of visible symptoms exhibited by dragon fruit plants.
Farmers or field technicians carefully document all abnormal morphological changes, which are then coded using our standardized symptom classification system (G1-G.
For instance, observations such as premature fruit drop .
oded as G.
, rot development in shaded plant portions (G.
, stunted fruit growth (G.
, and abnormal fruit desiccation (G.
are logged as key diagnostic evidence (E).
These symptom codes serve as critical inputs for the Bayesian calculation engine, translating qualitative field observations into quantitative data points.
The system currently recognizes 32 distinct symptoms that may indicate any of 8 major dragon fruit diseases (P1-P.
, with each symptom-disease relationship weighted by empirical field data.
This initial symptom identification phase is crucial as it forms the evidentiary foundation for all subsequent probabilistic calculations, ensuring the diagnostic results remain grounded in actual plant conditions.
Step 2: Formula Selection The system employs the multivariate Bayesian formula (Formula .
to analyze the complex relationship between multiple observed symptoms (G1-G.
and potential diseases (P1-P.
The selected formula is expressed as:
aycn ) O ycy.
aycn ) O A .
O ycy.
aycn ) O ycy.
aycn ) ycy.
a1 ya2A.
yayco ) = ycu Ocyco=1 ycy.
ayco ) O ycy.
ayco ) O A .
O ycy.
ayco ) O ycy.
ayco ) This formulation was deliberately chosen based on three key scientific considerations:
First, it accurately models real-world phytopathological scenarios where plants typically present with multiple co-occurring symptoms, unlike simplified single-symptom models.
The joint probability calculation accounts for symptom interactions and their cumulative diagnostic significance, which is particularly crucial for diseases with overlapping symptomatology like Fusarium wilt (P.
and stem base rot (P.
Second, the formula incorporates both disease-specific symptom patterns .
hrough the likelihood terms P(E|H)) and population-level disease prevalence .
ia the prior probabilities P(H)).
This dual consideration ensures the system balances between the specificity of individual symptom-disease relationships and the epidemiological context of the growing region.
Third, the denominator's normalization term enables direct comparison of probabilities across all candidate diseases, generating a ranked differential diagnosis.
This comparative analysis is essential for clinical decision-making, as it allows farmers to prioritize the most probable diseases while considering alternative diagnoses with similar symptom presentations.
Step 3: Computational Process The system executes a multi-stage computational process to derive disease probabilities from observed symptoms.
aycn ) O ycy.
aycn ) O A .
O ycy.
aycn ) O ycy.
aycn ) ycy.
a1 ya2A.
yayco ) = ycu Ocyco=1 ycy.
ayco ) O ycy.
ayco ) O A .
O ycy.
ayco ) O ycy.
ayco ) To streamline the computational process, a Bayesian probability weighting table was constructed based on the symptoms identified in Step 1, with values derived from Table 3 and 4.
Symptoms ycy.
ya1 ya2 ya3ya.
= ycy.
ya1 ) O ycy.
O ycy.
O ycy.
O ycy.
a1 ) .
a1 ) O ycy.
O ycy.
a1 ) O ycy.
O ycy.
a1 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a2 )) .
O ycy.
O ycy.
a3 ) O ycy.
O ycy.
a3 )) .
a4 ) O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
O ycy.
a5 ) O ycy.
O ycy.
a5 ) O ycy.
a5 )) .
O ycy.
a6 ) O ycy.
O ycy.
a6 ) O ycy.
a6 )) .
a7 ) O ycy.
O ycy.
a7 ) O ycy.
O ycy.
) .
a8 ) O ycy.
O ycy.
O ycy.
O ycy.
a8 )) ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = 8 O 0.
5 O 0.
8 O 0.
9 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya.
= OO ycy.
a1 ya2 ya3 ya4 ) = 0.
a2 ) O ycy.
O ycy.
O ycy.
a2 ) O ycy.
a2 ) .
O ycy.
ya1 ) O ycy.
O ycy.
a1 ) O ycy.
a1 )) .
O ycy.
a2 ) O ycy.
O ycy.
a2 ) O ycy.
a2 )) .
O ycy.
a3 ) O ycy.
O ycy.
a3 ) O ycy.
a3 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
a5 ) O ycy.
O ycy.
a5 ) O ycy.
O ycy.
a5 )) .
a6 ) O ycy.
O ycy.
O ycy.
O ycy.
a6 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a7 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a8 )) 1 O 0.
1 O 0.
1 O 0.
1 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = 216 0.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya.
= OO ycy.
a1 ya2 ya3 ya4 ) = 0.
a3 ) O ycy.
O ycy.
O ycy.
a3 ) O ycy.
a3 ) .
O ycy.
ya1 ) O ycy.
O ycy.
a1 ) O ycy.
a1 )) .
O ycy.
a2 ) O ycy.
O ycy.
a2 ) O ycy.
a2 )) .
O ycy.
a3 ) O ycy.
O ycy.
a3 ) O ycy.
a3 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
a5 ) O ycy.
O ycy.
a5 ) O ycy.
O ycy.
a5 )) .
a6 ) O ycy.
O ycy.
O ycy.
O ycy.
a6 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a7 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a8 )) ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = 1 O 0.
1 O 0.
1 O 0.
1 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya.
= OO ycy.
a1 ya2 ya3 ya4 ) = 0.
a4 ) O ycy.
O ycy.
O ycy.
a4 ) O ycy.
a4 ) .
O ycy.
ya1 ) O ycy.
O ycy.
a1 ) O ycy.
a1 )) .
O ycy.
a2 ) O ycy.
O ycy.
a2 ) O ycy.
a2 )) .
O ycy.
a3 ) O ycy.
O ycy.
a3 ) O ycy.
a3 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
a5 ) O ycy.
O ycy.
a5 ) O ycy.
O ycy.
a5 )) .
a6 ) O ycy.
O ycy.
O ycy.
O ycy.
a6 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a7 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a8 )) ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = 1 O 0.
1 O 0.
1 O 0.
1 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya.
= OO ycy.
a1 ya2 ya3 ya4 ) = 0.
a5 ) O ycy.
O ycy.
O ycy.
a5 ) O ycy.
a5 ) .
a1 1 O ycy.
a2 1 O ycy.
O ycy.
a1 ) O ycy.
a1 )) .
O ycy.
a2 ) O ycy.
O ycy.
a2 ) O ycy.
a2 )) .
O ycy.
a3 ) O ycy.
O ycy.
a3 ) O ycy.
a3 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
a5 ) O ycy.
O ycy.
a5 ) O ycy.
O ycy.
a5 )) .
a6 ) O ycy.
O ycy.
O ycy.
O ycy.
a6 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a7 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a8 )) 1 O 0.
1 O 0.
1 O 0.
1 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = 216 0.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya.
= OO ycy.
a1 ya2 ya3 ya4 ) = 0.
a6 ) O ycy.
O ycy.
O ycy.
a6 ) O ycy.
a6 ) .
O ycy.
ya1 ) O ycy.
O ycy.
a1 ) O ycy.
a1 )) .
O ycy.
a2 ) O ycy.
O ycy.
a2 ) O ycy.
a2 )) .
O ycy.
a3 ) O ycy.
O ycy.
a3 ) O ycy.
a3 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
a5 ) O ycy.
O ycy.
a5 ) O ycy.
O ycy.
a5 )) .
a6 ) O ycy.
O ycy.
O ycy.
O ycy.
a6 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a7 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a8 )) ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = 1 O 0.
1 O 0.
1 O 0.
1 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = OO ycy.
a1 ya2 ya3 ya4 ) = 0.
a1 ya2 ya3 ya.
= ycy.
a7 ) O ycy.
O ycy.
O ycy.
a7 ) O ycy.
a7 ) .
O ycy.
ya1 ) O ycy.
O ycy.
a1 ) O ycy.
a1 )) .
O ycy.
a2 ) O ycy.
O ycy.
a2 ) O ycy.
a2 )) .
O ycy.
a3 ) O ycy.
O ycy.
a3 ) O ycy.
a3 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
a5 ) O ycy.
O ycy.
a5 ) O ycy.
O ycy.
a5 )) .
a6 ) O ycy.
O ycy.
O ycy.
O ycy.
a6 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a7 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a8 )) ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = 1 O 0.
1 O 0.
1 O 0.
1 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya.
= OO ycy.
a1 ya2 ya3 ya4 ) = 0.
a8 ) O ycy.
O ycy.
O ycy.
a8 ) O ycy.
a8 ) .
O ycy.
ya1 ) O ycy.
O ycy.
a1 ) O ycy.
a1 )) .
O ycy.
a2 ) O ycy.
O ycy.
a2 ) O ycy.
a2 )) .
O ycy.
a3 ) O ycy.
O ycy.
a3 ) O ycy.
a3 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a4 )) .
a5 ) O ycy.
O ycy.
a5 ) O ycy.
O ycy.
a5 )) .
a6 ) O ycy.
O ycy.
O ycy.
O ycy.
a6 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a7 )) .
O ycy.
O ycy.
O ycy.
O ycy.
a8 )) ycy.
a1 ya2 ya3 ya4 ) = ycy.
a1 ya2 ya3 ya4 ) = 1 O 0.
1 O 0.
1 O 0.
1 O 0.
8 O 0.
5 O 0.
8 O 0.
9 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
1 O 0.
a1 ya2 ya3 ya4 ) = OO ycy.
a1 ya2 ya3 ya4 ) = 0.
Following the computation of all hypothesis probabilities, the results are systematically compiled into a diagnostic matrix.
This matrix identifies the most probable dragon fruit diseases corresponding to the observed symptoms by ranking the hypothesis probabilities in descending order, with the highest probability value representing the primary diagnosis, as demonstrated in Table 5.
Table 5.
Bayes Calculation Results Hypothesis Probability Value Rank
0,998382
0,000370
0,000300
0,000231
0,000231
0,000208
0,000139
0,000139
The Bayesian probability analysis reveals that the observed symptom complex - comprising premature fruit drop (G.
, shade-induced rot (G.
, fruit stunting (G.
, and fruit desiccation (G.
- indicates Anthracnose Fruit Rot (Colletotrichum gloeosporioide.
with a posterior probability of 0.
Results and Discussion This research produces three main components: .
system flowchart, .
Entity Relationship Diagram (ERD), and .
application interface.
Figure 2 shows the flowchart design of the dragon fruit plant disease diagnosis expert system based on the Bayes theorem developed.
Flowchart of the Proposed System Admin System Start Display Main Menu Username Password Login Expert User Start Start Display Main Menu Display Main Menu Login Login Verification Yes Yes Manage Account Data Yes Manage Symptom.
Disease.
Solution.
Relationship, and Rule Data Database Symptoms.
Diseases.
Solutions, and Rules Diagnosis Diagnosis Result Save and Print Diagnosis Result Report End End Figure 2.
Flowchart of the Proposed System End The presented flowchart (Figure .
illustrates the proposed system workflow, structured into four main components based on user roles: Administrator.
System.
Expert, and End User.
The process initiates when each user type (Administrator.
Expert, and Use.
accesses the system via the main menu interface and proceeds to the login stage.
Upon entering their username and password credentials, the system performs a verification process.
If authentication fails, the user is redirected to the login page.
Conversely, upon successful verification, each user role is granted access to functionalities aligned with their specific responsibilities.
Once logged in, the Administrator is directed to input expert account data, which is subsequently stored in the system's central database.
The Expert, upon successful login, is authorized to input domainspecific data, including symptoms, diseases, possible solutions, relational data, and rule sets.
This data is also stored within the central database and constitutes the core foundation for decision-making processes within the system.
The System component is responsible for managing, storing, and retrieving data related to symptoms, diseases, solutions, and inference rules, serving as the backbone for the expert systemAos operation.
Meanwhile, the End User, upon successful login, is granted access to the diagnosis functionality.
The user can initiate a diagnosis process based on the pre-defined knowledge base populated by the expert.
Once the system generates a diagnostic result, the user is provided with the option to view, save, and print the diagnostic report.
Each activity sequence concludes with an "End" status, indicating the completion of the respective process flow.
Penyakit_id Gejala_id Id_gejala_penyakit Id_gejala Updated_at Created_at Id_user Updated_at Created_at Gejala Id_pakar Save_evidance Id_penyakit Updated_at Created_at Gejala_penyakit Updated_at Created_at Tmpt-lahir Tgl_lahir Updated_at Created_at User_name User_id Id_hipotesa Save_date_time Updated_at Created_at Id_penyakit_solusi Penyakit_id Penyakit_solusi Solisi_id Updated_at Created_at Save_hipotesa Id_evidance User_id user_name Save_date_time Updated_at Created_at Id_solusi User_id User_name Penyakit_kode Save_date_time Updated_at Created_at Save_solusi Id_solusi Updated_at Created_at Figure 3.
Entity Relationship Diagram (ERD) Figure 3 shows the illustrated Entity Relationship Diagram (ERD) represents a comprehensive data model of an expert system designed to assist in diagnosing diseases based on symptom analysis and subsequently recommending appropriate solutions.
This model integrates various entities, including users, domain experts, symptoms, diseases, and solutions, along with historical diagnostic data.
The structured data relationships enable efficient knowledge management and inference processes, essential for an intelligent decision support system in the healthcare or medical consultation domain.
At the core of the system are the User and Pakar (Exper.
The User table stores individual credentials and role classifications, while the Pakar table stores expert-specific information, such as name, gender, address, and birth details.
Experts are responsible for populating and maintaining the systemAos knowledge base by inputting validated data on symptoms, diseases, and treatment solutions.
The Symptom (Gejal.
entity holds information on observable health indicators, uniquely identified by codes.
Each symptom is linked to one or more Diseases (Penyaki.
through a many-to-many relationship defined by the associative entity Gejala_penyakit.
This relation includes a bobot .
attribute that quantifies the strength of association between a symptom and a disease.
Similarly, each disease is associated with one or more Solutions (Solus.
through the Penyakit_solusi entity, which facilitates multiple treatment recommendations for a single disease.
The system incorporates dynamic diagnostic features through the Save_evidence.
Save_hipotesa, and Save_solusi entities.
These serve as transactional records to log each user's diagnostic interaction.
Save_evidence captures selected symptoms by a user during a diagnostic session.
Save_hipotesa stores the inferred disease.
and associated probabilities calculated based on the entered symptoms.
Finally.
Save_solusi contains the system's recommendations or treatment suggestions provided to the user, tailored to the diagnosed condition.
This ERD reflects a well-structured, normalized database schema that promotes data consistency, traceability, and scalability.
It ensures the system can support complex inferences, facilitate personalized medical advice, and offer traceable records for both users and experts.
The use of timestamps such as Created_at.
Updated_at, and Save_date_time across all entities underscores the importance of temporal data tracking for system auditability and continuous improvement.
Figure 4.
The Main Interface of the Web-Based Figure 4 displays the main interface of the web-based expert system "SISPAK | BAYES", designed for diagnosing dragon fruit (Hylocereus spp.
) diseases.
The interface features six core navigation menus, each optimized for functionality and user experience:
Home: Serves as the primary dashboard, providing direct access to the diagnostic module.
Dragon Fruit Info: Contains botanical and agronomic details about Hylocereus species, including growth conditions and common cultivars.
Gallery: A visual repository of disease symptoms .
, anthracnose lesions, stem ro.
with annotated case studies.
Expert Consultation: Facilitates real-time communication with plant pathologists for complex Contact: A dedicated channel for user feedback and technical support.
Login: A secure authentication portal for system administrators and domain experts to update the knowledge base.
Figure 5.
System Login Interface Figure 5 displays a system login interface featuring two input fields .
sername with a sample "admin" entry and password displayed as dots for securit.
, accompanied by a "Login" button for authentication and a "Back" button for navigation.
The interface also includes two support links - "Forgot Password?" for account recovery and "Create Account" for new users.
This minimalist design prioritizes functionality and basic security measures like password masking, though additional input validation enhancements could further strengthen system security.
The layout effectively serves its purpose of regulating access to authorized users while maintaining user-friendly navigation.
Figure 6.
Symptom Selection Interface Figure 6 shows a symptom selection interface from a dragon fruit disease diagnostic system, where users can indicate observed symptoms by selecting "Yes" next to each listed symptom .
, premature fruit drop, stem rot, or white/brown spot.
The header ("Home") and welcome message ("Selamat Datang user") introduce the page, which guides users to select symptoms from a checklist to enable Bayesian probability calculations for disease diagnosis.
The interface aims to simplify data input for farmers but requires refinement in language accuracy and symptom presentation clarity.
Figure 7.
Diagnosis Results Interface The design of the diagnosis results page, as shown in Figure 7, is intended to provide clear and structured information to users after they complete the diagnosis process by answering a series of questions prepared by the system.
Following this process, the system automatically directs users to the diagnosis page, which presents the results of the disease diagnosis, including probability values and appropriate solution recommendations.
This aims to ensure that users gain a comprehensive understanding of the health condition of dragon fruit plants and the steps that can be taken for further Conclusions Based on the analysis and testing that have been conducted, it can be concluded that the expert system application for diagnosing diseases in dragon fruit plants using the Bayes method has several significant First, the Bayes method is effective in determining the type of disease affecting dragon fruit plants, providing results in the form of accurate probabilities based on the calculations performed.
Second, this system is highly beneficial for farmers in the Kampung Daun Baumata Farmers Group.
Taebenu District, in diagnosing the diseases occurring in their plants.
With the assistance of this expert system, farmers can more easily address the disease problems they face, thereby improving plant health and yield.
Additionally, this system is capable of displaying comprehensive information regarding disease data, the symptoms presented, and appropriate treatment methods for diseases in dragon fruit Thus, this application not only serves as a diagnostic tool but also as a valuable source of information that helps farmers manage their crops more effectively.
References