Sentiment classification of delta robot trajectory control using word embedding and convolutional neural network

Sentiment classification (SC) is an important research field in natural language processing (NLP) <span lang="EN-US">that classifying, extracting and recognizing subjective information from unstructured text, including opinions, evaluations, emotions, attitudes. Human-robot interaction (HRI) also involves processing, knowledge representation, reasoning by utilizing deep learning, cognitive science, robotics. However, sentiment for HRI rarely implemented, especially to navigate a robot using the Indonesian Language which semantically dynamics when written text. This paper proposes of Bahasa Indonesia that supports delta move particular trajectory directions. Navigation commands were vectorized word embedding method containing two-dimensional matrices propose classifier pattern such as convolutional neural network (CNN). The result compared architecture CNN, GloVe-CNN, Word2Vec-CNN. As method, CNN models trained, validated, tested with higher accuracy are 98.97% executed less than minute. produces four navigation labels: right means </span><em><span lang="EN-US">'kanan'</span></em><span lang="EN-US">, left lang="EN-US">'kiri</span></em><span lang="EN-US">', top lang="EN-US">'atas</span></em><span bottom lang="EN-US">'bawah</span></em><span multiplier factor. utilized transform any into direction along end-effector coordinates.</span>