Development of Indonesian Language Speech Recognition Algorithm Model in Knowledge Database

Research on humanoid robots in Indonesia is developing very rapidly, marked by research on humanoid robots having the ability to learn independently using Indonesian and vocabulary to receive and provide responses from humans based on Natural Language Processing. The general objective of research focuses on speech recognition systems to be able to interact with people naturally. Humanoid robots must be able to consider the question style and infer answers through automated, unreliable conversations in noisy environments. In development, there are still shortcomings that is if the answer is not in the knowledge base, the robot will look for the context that is most similar to the knowledge base, instead of giving an answer, the robot gives an answer `not found'. Sometimes, the wrong answer will be displayed because the cosine value between the question and the wrong answer is influenced by the same context, that the truth is not the actual answer. Based on these problems, this study was conducted with the aim of producing structured data and information knowledge according to the topic of the question, producing knowledge storage designs, optimizing models based on design and knowing the performance of knowledge representation storage models. Based on these objectives, this research will design algorithms and modeling representation of Indonesian language knowledge that will be applied to humanoid robots. The expected output is an algorithm design and storage of knowledge representation on humanoid robots.