Product reviews are critical for businesses to understand the needs of their customers. This paper proposes an opinion mining model that generates a multi-aspect opinion summary based on aspect-based sentiment analysis (MAOS) as an information synthesis form. This summary is based on senti-ment analysis according to different aspects of the product using a deep learning method supported by ontology. Both sentiment analysis and aspect detection tasks of the MAOS model depend not on individual words or sen-tences but on a strategy of aspect-based information synthesis from opin-ions. The summary, as the information synthesis form, helps opinion miners capture a significant amount of information without spending as much time reading as the summary based on an extractive or abstractive summarization strategy. This study's experimental results show that embedding the ontology into the corpus as part of the knowledge mining method greatly improves performance in tasks like finding aspects and combining information in opinions, even with a small corpus, when compared to standard methods.