Content area
Full text
1.
Introduction
The explosive growth of social media on the World Wide Web dramatically changes people's methods of expressing their opinions, and, consequently, how they make decisions in their lives. The medical domain is not immune of this fact. With the vast amount of medical online information and the rapid growth of social media in this field, people no longer take a drug before going to the internet to learn something about it. However, extracting and analyzing opinions manually from the huge volume of texts is a formidable or even impossible task. The aim of the automatic sentiment analysis (opinion mining) system is to provide the user with the attitudes of people toward entities and their attributes or the overall polarity of a document. Opinions can be classified into two groups according to the type of presentation in the sentences (Liu and Zhang, 2013). An explicit opinion is one that is expressed explicitly in a subjective sentence, e.g., "Methadone works great for me." On the other hand, an implicit opinion is one that is implied in an objective sentence. These sentences usually describe desirable or undesirable facts, e.g., "Methadone makes me feel nauseous." Unlike implicit-opinion mining, a significant amount of research has been done on explicit-opinion mining.
One issue that remains almost unexplored in medical-opinion mining is detecting implicit-sentiment words and phrases that show desirable or undesirable facts about one medicine in the drug review. One group of these words and phrases are drug side effects that cover a large portion of drug reviews. Side effects can imply both positive and negative opinions about one drug. Even so, talking about drug side effects is rarely positive, and positive terms are more related to drug effectiveness. On the other hand, most drug reviews are narratives and contain patients' experiences. Thus, just detecting these kinds of words individually and without considering the context cannot be an appropriate solution to this problem. For instance, some sentences show the general status of patients or their symptoms before taking a drug. As an example, in the sentence "I take alprazolam when I'm having anxiety about whatever it may be," "anxiety" is a disease symptom, and the reason for taking this psychoactive drug by the patient. Indeed, in this sentence the...