The number of news articles with various volumes is increasing, so they need to be summarized. This is to find important information in Indonesian language news articles. For this purposed, a lot of research has been done to build an automatic summation system. Steinberger et.al. propose the sentence selection method in summarizing text based on the Latent Semantic Analysis (LSA) method. The sentence selection was conducted based on the value of the concept that the first concept was considered to have the highest importance, so there was a possibility that sentences that the unimportant sentences were selected. This results in summaries containing incoherent information. This study proposed the sentence selection method based on the relatedness between concepts and sentences and referring to the structure of writing news articles. The experimental results showed that the proposed sentence selection was able to summarize the text well and achieved better performance than the result of Steinberger et.al. sentence selection method. In average the value of F-Measure increases by 42% for 10% compression rates, 39% for 30% compression rates, and increases by 24% in a 50% compression rate.