Aspect-based sentiment analysis summarizes what people like and dislike from reviews of products or services. In this paper, we adapt the first rank research at SemEval 2016 to improve the performance of aspect-based sentiment analysis for Indonesian restaurant reviews. We use six steps for aspect-based sentiment analysis i.e.: preprocess the reviews, aspect extraction, aspect categorization, sentiment classification, opinion structure generation, and rating calculation. We collect 992 sentences for experiment and 383 sentences for evaluation. We conduct experiment to find best feature combination for aspect extraction, aspect categorization, and sentiment classification. The aspect extraction, aspect categorization, and sentiment classification have F1-measure value of 0.793, 0.823, and 0.642 respectively.