This research aims to address the issues surrounding the allocation of employee benefits at TVRI Nusa Tenggara Timur. Currently, the decision-making process for benefits is solely based on recommendations from supervisors, without taking into consideration employee performance, potentially leading to social jealousy and suboptimal outcomes. In this study, we develop a Decision Support System (DSS) utilizing the Analytical Hierarchy Process (AHP) method to assess employee performance. AHP allows for a hierarchical and objective breakdown of multi-factor problems. The criteria used for evaluation include attendance, behavior, length of service, and dependents. It is expected that the implementation of this DSS will enhance transparency and objectivity in decision-making, minimize employee dissatisfaction, and provide a more robust basis for benefit allocation decisions. This research offers practical contributions to human resource management and decision-making at TVRI Nusa Tenggara Timur.

Attention to spices and flavorings among the younger generation is still low. The strategy that can be used to overcome this problem is a programmed and computerized arrangement of spices and flavorings using Convolutional Neural Network (CNN) calculations. In this exploration there are 300 images of spices and flavors which will be characterized into 3 classifications. Namely ginseng, ginger and galangal. Information in each classification is divided into two, namely preparation information and testing information with a proportion of 80%: 20%. The CNN model used in computerized grouping of spice and flavor images is a model with 2 convolutional layers, where the first convolutional layer has 10 channels and the second convolutional layer has 20 channels. Each channel has a 3x3 portion frame. The channel size in the pooling layer is 3x3 and the number of neurons in the secret layer is 10. The actuation capability in the convolutional layer and secret layer is tanh, and the actuation capability in the result layer is softmax. In this model, the accuracy of preparation information is 0.9875 and the loss value is 0.0769. The precision of the test data is 0.85 and the loss value is 0.4773. Meanwhile, testing new information with 3 images for each classification resulted in an accuracy of 88.89%.