Case-based Reasoning Approach For Job Recommendation

With the increasing popularity of the Internet, more and more recruitmentmessages are posted on the web. Job seekers have to spend a lot of time and effort tofind a proper job that they can meet the job requirement and satisfies themselves fromthe massive recruitment information. That leads to the emergency of the scramble forjobs and information mismatch. Information mismatch occurs when unemployedgraduates do not acquire information on relevant existing vacancies, and firms do nothave the information necessary for finding persons with adequate qualifications. Tosolve the problem, many recruitment sites carry on various reforms. They recommendpositions for job seekers in order to attract the job seekers and make them only focuson their job sites. However, job seekers are not satisfied with the job recommendedbecause it is not targeted. Traditional recommendation approach, which is based onbrowsing and query, cannot meet the needs of the seekers and cannot enablepersonalization. For many job seekers, there is a common problem that they cannotcorrectly assess themselves. Job seekers are blind in the job-seeking process sincethey cannot really understand themselves, understand the business, or understand thejob. To overcome the above problems, this paper carries out an in-depth researchmainly from the following aspects:1. It introduces the case-based reasoning approach for a job recommendationsystem. This approach is helpful for the job seekers to improve the possibility of theirsuccess to find a job by utilizing the successful job applicants’ cases. It proposes acase representation that can expresses more completely and effectively the user profileand job-seeking intention of the job seekers.2. It classifies the job-seeking cases by the application of the Na ve Bayesclassification techniques. The industry information is obtained by the classification ofa job seeker’s major, which means that the case retrieval can be conducted in aspecific category. Therefore, it reduces the retrieval time and improves the efficiencyof the retrieval.3. It uses the analytic hierarchy process to determine the weight of the case attributes, introduces a similarity calculation method for the interval attributes andhierarchical attributes to make the result more accurate.4. It introduces the trust in the social network to the job recommended system.In addition to the similar reasoning on user profiles and job intentions, thesupplemented trust metric of similar cases improves the degree of user satisfaction ofthe recommendation.5. It utilizes job intentions context in the recommended system to perceive thechange of the active job seeker’s career intention and preferences timely and adjuststhe recommendation result to achieve personalized service.6. It realizes the algorithms with statistical analysis software R and MicrosoftSQL Server2008database management system that stores recruitment and jobseekers transactions from2005to2010from a talent market web site. A preliminaryjob recommendation system is implemented.In short, this paper explored the Case-Based Reasoning method, Bayesianclassification, contextual information retrieval techniques, and trust-basedrecommendation algorithm and their application to the job recommended system. Itobtains a satisfied result of the job recommendation for seekers. This study is helpfulfor the research and development in e-commence and recommender systems.