CAREER ADVISORY USING DISTRIBUTED NEURAL NETWORKS Amose Suwali, Supervised by Mr. A.Ndlovu.Department of Computer ScienceSCHOOL OF INFORMATION SCIENCES ANDTECHNOLOGYHARARE INSTITUTE OF TECHNOLOGY ABSTRACTThe project aims to develop an artificial intelligent web-based softwareagent that learns to recommend career fields to students based on theirpersonality type, past academic performance, and interests. The projectattempts to address the problem of the lack of personalized career guidanceraising due to the shortage of human and time resources that the processdemands through a machine learning web-based expert system. Furthermore, since the project is low budgeted it seeks to address thelack of high powered computer system through the design of a horizontallyscalable solution architecture model that distributes computational workproportionally to the nodes in hybrid network architecture (client-server andpeer to peer network).

                 1.                       INTRODUCTIONCareer adviceis a necessity every student ought to receive at any academic level. Unlessthis advice is personalized and tailored to one`s personal attributes thisadvice proves ineffective. There is a rise in the lack of quality personalizedcareer guidance due to the shortage of human and time resources that theprocess demands.

The projectaimed at developing a career advisory expert system that recommends careerfields to a student based on personalized career recommendation model. Thesystem implements machine learning techniques to develop a progressivelyimproving model that learns to recommend a career field for a student.However,implementing such a system requires high computational resources and introducesanother limitation to the success of this low-budgeted project. Thus, theproject aims to design a system architecture that distributes computationalwork of the above-proposed system.                   2.                       GOALS AND OBJECTIVES This survey paper is compiled basedon a research project aimed at developing a career advisory expert system with thebelow listed goals and objectives.

A career advisory solution that  ? Highly values and considers the user`s interest and passions. ? Provides personalised career advice tailored to the user`s academic strengths and abilities. ? Continuously improves/learns to provide relevant advice based on user`s feedback ? Utilizes leverageable processing power available at the client nodes of the used RESTful client-server architecture.

 RELATED WORKResearch work was perform based on existing solutionsin two domains highlighted in the goals and objectives. These are        i.           Career Guidance Models.      ii.           Processing load distribution in client-server architectures.                  3.                       CAREER GUIDANCE MODELS.

·        AutomatedCareer Counselling System for Students using CBR and J48This researchintends to solve the career assortment problems by making use of the CBR(Case-Based Reasoning) and Decision Tree J 48 algorithm. The system establishesan automated process similar to a one-to-one meeting with a career counsellorand aids to ‘plan’ a career true to the student’s grade, IQ, hobbies and,predominantly, gender. Students can later determine a career from the proposedoptions and the illustration of related jobs. The system’s distinction is tonominate Universities offering education for the recommended careers. 1·        iAdviceis a Career Advisory expert system designed by Chathra Hendahewa et al.

to guidestudents for faculty of B.Sc. IT students of Moratuwa University, engaged intheir higher education to determine their career paths and to select theircourse subjects to be in-line with their career goals. 16. The Systemconsists of three components viz; knowledge base, Inference Engine, and userinterface.

This expert system uses features such as reasoning ability,providing explanations, alternative solutions, uncertainty and probabilitymeasures, questioning ability and also forward chaining, backward chaining andrule-based inference in designing expert system. This system was divided intotwo main subsystems i.e.

Career known subsystem and Career unknown subsystem.The first subsystem provides advice to students who have a specific career goaland the second subsystem provides advice to students who are unclear aboutcareer objectives. Past examination performance, student preferences, andskills, industry alignment with subjects, are the main factors considered by ahuman expert in providing career guidance. 3·        CPSRSCareer PathSelection Recommendation System (CPSRS) was proposed and developed by Razak,Hashim, Noor, Hazwam, & Halim.

This system was developed using the fuzzylogic technique. CPSRS was designed for providing direction and guidance tofinal year students for faculty of computer and mathematical science, UniversityTeknologi MARA (UiTM) students of Malaysia for choosing a suitable career.Factors considered for career selection are student’s strengths, skills, andpersonality, interest, past academic records. The use of fuzzy logic approachhelps students by giving career recommendation based on career test. They usedFuzzy Associate Memory (FAM) as fuzzy inference because FAM will contain theknowledge from an expert that is believed to be able to reach nearly any sortof control objectives. The operator MIN and intersection AND is used as aninference rule. 4CGM- (CareerGuidance Model)Winston developed expert system model (CGM) for African high school students.

The survey was carried out to estimate the level of professional satisfactionwith the task and nature of their career and also determined what careerpractices are carried out in Kenyan high school. It was found thatapproximately 90% of public high school students in Kenya were not gettingreasonable career guidance due to limited resources and time. The proposedmodel consists of three sections, personality analysis, decision makingregarding selecting specific job category (simulation of activity), andScholastic Aptitude Testing (SAT) for evaluating one’s cognitive ability.Personality analysis model created knowledge and rules based on Myers-BriggsTypology Indicator (MBTI). The proposed system was designed using visual basicand Access. 4·        CMS(Career Master System):Balogun,Thompson specifies the development of career master DSS that  counsellors can use to help students in identifyingthe right discipline for secondary school leaving students of Nigeria who haveproblems with their choice of careers as they intend to study at tertiaryinstitutions of their choice. This career master system implemented usingVisual Basic. This system is designed for desktop and counselors, and systemrecommendation was based on parameters such as ability, skills, IntelligentQuotient, interest, parents and friends influence, preferences, parentoccupation and hobbies, past academic performance.

For the development of thiscareer master system author considers four databases subject, study, pass, andcourse. Author checked this DSS system with career counselors result, it foundthat developed system recommendations are correlated with counsellor’srecommendations. The system provides the desktop for the counsellors to enhancethe duty of choosing the best and most appropriate discipline for clients.(Balogun, Thompson, & State, 2009) 5·        Rule-BasedDSS:MuhammadZaheer Aslam, et al. presents the design and development of a proposed rulebased Decision Support System that helps students in selecting the bestsuitable faculty/major decision while taking admission in Gomal University.They designed a model using visual basic for testing and measuring thestudent’s capabilities like intelligence, understanding, comprehension,mathematical concepts his/her past academic record, intelligence level. Theydivided tests into two parts one for testing capabilities and abilities andanother for testing intelligence. Capability and ability test consists of 100questions i.

e. 20 questions each for English, Mathematics, Physics, Chemistry,Computer Science / Biology and intelligence test consisting of 50 questions.They applied model resulting into a rule-based decision support system todetermine the compatibility of the available faculties/majors in GomalUniversity. These DSS identify the most suitable faculty or major for thestudent based on his abilities and capabilities extracted from the test moduleresults. They used CLIPS language to store knowledge base. Rules can be mademore customized and more criteria may be added to it for more data minedresults. It can be extended to include other universities faculties and majorsto be able to serve more students wishing to be enrolled in other universitiesand make the criterion customized for that university. 7.

·        Designof an online expert system for career guidanceThe systemwill have the knowledge-base which contains the details about the colleges inPondicherry. This information is acquired from web pages using pattern matchingand jSoup parsing technique and the knowledge-base is constructed automaticallywithout manual efforts. Rules are framed and an inference engine is developedwhich makes the Expert System. The constructed knowledge-base can be queriedwith domain related queries and the Expert System provides the most relevantdetails for the query. 8 4          LOAD DISTRIBUTION TECHNIQUES IN CLIENT-SERVER ARCHITECTURES·        AutomatedCareer Counselling System for Students using CBR and J48This researchintends to solve the career assortment problems by making use of the CBR(Case-Based Reasoning) and Decision Tree J 48 algorithm. The system establishesan automated process similar to a one-to-one meeting with a career counsellorand aids to ‘plan’ a career true to the student’s grade, IQ, hobbies and,predominantly, gender. Students can later determine a career from the proposedoptions and the illustration of related jobs.

The system’s distinction is tonominate Universities offering education for the recommended careers. 1 ·        iAdviceis a Career Advisory expert system designed by Chathra Hendahewa et al.to guidestudents for faculty of B.Sc.

IT students of Moratuwa University, engaged intheir higher education to determine their career paths and to select theircourse subjects to be in-line with their career goals. 16. CONCLUSIONS       i.           Career Guidance Models.Based on therelated work research conducted currently the leading career guidance solutionsare based on the below-listed recommendation models·        LogicDriven·        Rule-Based·        FuzzyLogic·        KnowledgeBase·        Case-BasedReasoning & Decision Tree J48• Thesemodels above are implemented without the ability to improve or learn theirrecommendation throughput over time.

• Theabove-mentioned models have an input domain of 37 parameters which include astudent`s examination marks, parent’s occupation, cost of tuition fees, homelocation, students attitude etc. Thecombination of the parameters input into those models does not map to a careerrecommendation that is tailored to a student`s personal strength, skills andinterests.• Theexisting models do not consider the non-linear relationship between the inputparameters themselves but instead, they attempt to linearize them, for example:A student`sacademic performance trend is not linearly related to his or her personalitytype, thus a linear if-then-else function cannot phantom the pattern that mapsthe two parameters but rather attempts to transform this non-linearrelationship into one which only has two possible states (True or False).     ii.           Load distribution techniques in client-server architectures A student`s academic performancetrend is not linearly related to his or her personality type, thus a linearif-then-else function cannot phantom the pattern that maps the two parametersbut rather attempts to transform this non-linear relationship into one whichonly has two possible states (True or False).

Database Trained Execute Trained JS Model Stub Server Side Client Side JavaScript Web Workers Web Browser Logic Engine    REFERENCES1. MahaNawaz, Anum Adnan, Unsa Tariq, Jannat Fatima Salman, Rabia Asjad, and MariaTamoor, Department of Computer Sciences Kinnaird College for Women Lahore,Pakistan, “Automated Career Counseling System for Students using CBR and J48″2. C.Hendahewa, M. Dissanayake, S. Samaraweera, A. Ruwanpathirana, and A. S.

Karunananda, “Artificial Intelligence Approach to Effective Career Guidance,”no. September, pp. 32–42, 2006.3. T. R.Razak, M. A.

Hashim, N. M. Noor, I. Hazwam, and A. Halim, “Career PathRecommendation System for UiTM Perlis Students Using Fuzzy Logic,” 2014.4.

O.Winston and M. Lawrence, “Career Guidance Using Expert System Approach,” pp.123–131, 2000.

5. V. F.Balogun, A. F. Thompson, and O. State, “Career Master: A Decision SupportSystem ( DSS ) for Guidance and Counseling in,” vol. 10, no.

2, pp. 337–354,2009.6.

M. Z.Aslam and A. R. Khan, “A Proposed Decision Support System / Expert System forGuiding Fresh Students in Selecting a Faculty in Gomal University,” Ind. Eng.Lett.

, vol. 1, no. 4, pp. 33–41, 2011.7. S.Saraswathi, M.

Hemanth Kumar Reddy, S. Udaya Kumar, M. Suraj, Sk. Khaja Shafi,Information Technology, Pondicherry Engineering College, Pondicherry, India,”Design Of An Online Expert System For Career Guidance”8.

K.Addison, T. Antwi, and F. Amissah, “Factors that influence junior high schoolstudents’ choice of programmes for senior high school and its implication fortheir career choice in Ghana.,” J. Educ.

Curric. Dev. Res., vol. 2, no.

September, pp. 1–14, 2014.9.

OSORO,B. ET AL. 2000.

Career decision-making of high school students in Kenya. AnInternational Journal for the Advancement of Counseling10. K.

I.N. N. A. O.

ISSA, “Factors Affecting the Career Choice of Undergraduates inNigerian Library and Information Science Schools,” Nigeria, 2008.11. OjengeWinston and Muchemi Lawrence, “Career Guidance Using Expert System Approach”..

AUTHOR BIOGRAPHYANU MARIA is an assistant professor in thedepartment of Systems Science & Industrial Engineering at the StateUniversity of New York at Binghamton. She received her PhD in Industrial Engineering from the University of Oklahoma. Herresearch interests include optimizing the performance of materials used inelectronic packaging (including solder paste, conductive adhesives, andunderfills), simulation optimization techniques, genetics based algorithms foroptimization of problems with a large number of continuous variables, multicriteria optimization, simulation, and interior-point methods.