Aspect-based opinion mining aims to model relations between the polarity of a document and its opinion targets, or aspects. While explicit aspect extraction has been widely researched, limited work has been done on extracting implicit aspects. An implicit aspect is the opinion target that is not explicitly specified in the text. E.g., the sentence “This camera is sleek and very affordable” gives an opinion on the aspects appearance and price, as suggested by the words “sleek” and “affordable”; we call such words Implicit Aspect Indicators (IAI). In this paper, we propose a novel method for extracting such IAI using Conditional Random Fields and show that our method significantly outperforms existing approaches. As a part of this effort, we developed a corpus for IAI extraction by manually labeling IAI and their corresponding aspects in a well-known opinion-mining corpus. To the best of our knowledge, our corpus is the first publicly available resource that specifies implicit aspects along with their indicators.
