*2.1. Hearing Loss*

Hearing loss is the most common sensory deficit. Among 432 million people with hearing loss, 93% of them are adults, and 54% of them are male—over 5% of the world's population have hearing loss. Furthermore, it is estimated that the number of people with hearing loss will rise to over 900 million by 2050. Untreated hearing loss can negatively affect individuals' communication performance and thus the quality of life in individuals and their families. Hearing loss can be reduced speech understanding, declined acoustic information, and impaired localization of sound sources [12]. Hearing loss is associated with comorbidities such as social isolation, loneliness [13], depression [14–16], balance problem [17], acoustic neuroma(vestibular schwannoma), multiple sclerosis, cardiovascular disease [18], and diabetes [19,20]. Recently, a growing body of research has shown that hearing loss and dementia are related [21–24]. Hearing loss in later life is one of the factors that play an important role in decreasing cognitive ability and developing dementia.

Hearing loss can be caused by damage to any portion of the peripheral and central auditory systems. The main causes of sensorineural hearing loss are degenerative processes associated with aging, genetic mutations, noise exposure, exposure to therapeutic drugs that have ototoxic side e ffects, and chronic conditions [25]. The most common cause of hearing loss is aging [17]. Age-related hearing loss, generally referred to as presbycusis, typically arises from gradual changes in the inner ear, a ffecting the sum of sensory, neural, and metabolic causes. Additionally, other factors such as ear diseases and the e ffects of noise exposure may a ffect people at all ages and stages in life [26]. Noise-induced hearing loss is caused by loud noise exposure for more extended periods. It has been suggested that more than 12% of the global population is at risk for hearing loss from noise [27]. The WHO estimates that one-third of all cases of hearing loss can be attributed to noise exposure.

Several options are available for hearing loss, ranging from medical treatment to listening devices such as hearing aids and cochlear implants. Treatment depends on the cause and severity of hearing loss. For age-related and noise-induced hearing loss, hearing cannot be treated, but hearing can be restored after using hearing aids or cochlear implantation [25].

#### *2.2. Social Q&A Community*

Social Q&A is an online question-and-answer platform enabled by Internet and Web technologies. It is a community-driven platform that allows online users to exchange information by asking questions and providing answers [28,29]. It is open to the public, where interested parties submit questions to be answered by other fellow online users around the world. Over the last decade, social Q&A has gained popularity, and the number of visits to the top Q&A sites such as Yahoo! Answers has increased dramatically [30]. For example, Yahoo! Answer includes more than 300 million questions and 90 million unique users worldwide as of 2012 since the service launched in 2005 [10]. Another popular social Q&A site launched in 2006, Wiki Answers, has 17 million answers posted. Questions and answers on topics ranging from education to diet become a source of rich experience and opinion for anyone with similar concerns or problems. Naver is the largest online platform in South Korea, referred to as the "Google of South Korea", and Naver provides the Social Q&A platform, Knowledge-iN. Users can post any topics on Knowledge-iN, and professionals or people who know the issues make comments and provide information and solutions, in content-centered platforms, and then users select the most valuable answers, and respondents earn awards or points.

#### *2.3. Social Q&A Log Analysis*

According to the Pew Internet and American Life Project data, more than 70% of Internet users use the Internet to search for health or medical information [7]. Increased use of such online platforms such as social Q&A sites leads to the generation of unprecedented volumes of information about symptoms, treatments, and health directly from patients, which is generally referred to as electronic patient-authored text [6,31]. As the volume of potentially valuable patient-authored text on social Q&A is growing, more researchers have paid attention to identifying the potential of online data sharing platforms for education and health service. Online patient narratives are a reliable data source for detecting disease trends and identifying medical terms [31]. Moreover, novel insights into patients' treatment decisions and drug-treatment e ffects were discovered on PatientsLikeMe [32].

A review of the literature on the electronic patient-authored text on the social Q&A community indicates that the existing research streams can be divided into content-centered (e.g., question and answer narratives) and user-centered (e.g., questioners, answerer, and the community) studies. The content-centered studies have mainly focused on three areas: (1) detection of diverse types of health-related questions and answers [9,33,34]; (2) identification of medical concepts in the patient-authored text [31]; and (3) evaluation of the quality of questions and answers with a distinct set of criteria [35,36].

The first type of research has examined electronic patient-authored questions and answers from social Q&A sites to detect health-related hot topics [33,34,37]. Lu et al. (2013) applied text clustering techniques to detect disease topics such as lung cancer, breast cancer, and diabetes, and related symptoms, medical tests, drugs, procedures, and complications. Sadah et al. [33] identified a set of popular topics and associated sentiments based on the patients' demographics. The second type has focused on identifying medically relevant terms and mapping words from the patient-authored text to medical concepts [9,31]. A language gap between patients and health care professionals is known to hinder effective communication between the two groups, so identifying and bridging the vocabulary gap is crucial [31]. Park et al. [8] applied the named entity recognition method to identify medical terms in their collected diabetes dataset and then map the identified terms to the formal medical vocabularies in the Unified Medical Language Systems (UMLS). Lastly, with concerns about the quality of both health-related questions and answers, researchers have proposed a diverse set of quality criteria and empirically examined them [36,38]. Harper et al. [37] employed supervised machine learning algorithms to distinguish information and conversational intent questions automatically. Their findings show evidence that conversational questions yield a lower archival value than informational questions.
