Design And Implementation Of Content Based Service Work Order Recommendation System

With the continuous development of the market economy,people will have more appeals besides the issue of food and clothing.Whether a company or a local government needs to respond to this change in a timely manner,to better serve "customers." Therefore,more and more organizations are paying attention to work orders.Although many organizations currently have their own work order systems,the work orders are all manually assigned by the customer service staff to various processing employees or departments,resulting in inefficiency and failing to respond to the demands of the complainants in a timely manner.It may even cause the outbreak of similar work orders.Therefore,the establishment of a recommendation system for service work orders is of great significance.The accuracy of the work order recommendation algorithm and the ranking of the recommendation list directly affect the work efficiency of the work order recommendation system.The main work accomplished in this paper is as follows:(1)A hybrid recommendation model LDTF is proposed.Through the use of the compulsive work order data segmentation,using the hybrid data model LDTF to extract the feature vectors of the work orders,and the establishment of the employee's preference vector,the work staff is recommended to select the appropriate work orders.At the same time,this paper proposes a new method for dimension reduction of feature vectors.In order to further improve the recommendation accuracy of worksheet recommendation algorithm,this paper proposes an improved feature word extraction algorithm PG-TF/IDF.It proposes the concept of part-of-speech ratio and word-based information gain to more accurately calculate employee processing.The interest vector of the ticket,thereby improving the accuracy of the recommendation algorithm.Through a series of experiments,the proposed improved algorithm,model,and method will have higher recommendation accuracy and recall rates on the work order data set,as well as fewer run-time and low-dimensional feature vectors.(2)The sorting learning algorithm is applied to the ticket-based recommendation algorithm,and a new sorting learning model LTE is proposed,which integrates the sentiment analysis,the text content of the work order,the turnover time of the work order,and a new sorting learning model is obtained.LTE,through a series of experiments,this sorting model can accurately predict the order of employee processing work orders.(3)Designed and developed work order recommendation system.The system includes a work order processing subsystem and a back office management subsystem.The work order processing part includes work order display,work order processing,and return of single work order.Management section includes work order creation,work order allocation,work order recommendation and other modules.