Expert Finding And Engineering Empirical Knowledge (EEK) Recommendation Mechanism

Systematic and dynamic renewed engineering knowledge is one of its core competence.Enterprises are turning to online community and forums for empirical knowledge which should be revised for several times before sharing,thus ensuring its accuracy and practicability.In the knowledge accumulation period,how to carry out standard scrutiny into engineering empirical knowledge is a key to the efficiency of knowledge accumulation and improvement in its quality.With the accumulation of knowledge,information overwhelming occurs,and hinders people to filter needed knowledge efficiently.There are two direct problems on engineering empirical knowledge management process.Taking into account the characteristics of MediaWiki platform and empirical knowledge accumulation mechanism,an expert recommendation mechanism which considers the knowledge relevance between users and items and authority among users simultaneously,adopts Okapi BM 2500 model to calculate users' knowledge relevance,refines the PageRank algorithm for better performance of users' authority calculation and obtains the final experts list with Cascade collation is proposed.Based on the User-Item-Time(UIT)3-Dimension model and regarding to the individuality of users' short-term interest,we develop the Collaborative Filtering Based on Dynamic Trust Algorithm(CFBDT)to recommend engineering empirical knowledge(EEK),considering the dynamic trust between users in the virtual community and specified time windows,in order to improve the performance of existing collaborative filtering recommendation system.Users' rating matrix of UIT model suffers from sparsity severely as the increase of users and items.Therefore,a pre-filling algorithm to pre-process the rating matrix sequentially is developed to avoid the rapid accuracy loss of recommendation system.Moreover,loyalty L is defined to measure the probability of which a target user would be convinced by an experienced user,since trust between users varies dynamically according to their historical interactions.Based on the theoretical research,an EEK management system is designed to implement in real industrial production.The system architecture is equipped with knowledge accumulation and knowledge recommendation functional modules.Individual scenes are provided to show the process of expert finding for engineering empirical knowledge revising,and that of EEK collaborative recommendation considering dynamic trust in the social network.To sum up,major contributions of the dissertation and some advices for further work are presented.