This study aims to analyze and categorize user experience (UX) of smartphones by utilizing social media data (Twitter). Social media (e.g., Facebook, Twitter) can be helpful for observing natural UX through the users' words. It is a potentially valuable source of data that can be used to examine the thoughts of millions of people. To gather UXs of smartphones, mining social media data (Twitter) techniques were used. Collected UXs were categorized according to the product smartness and the relationship between the product smartness and UXs were identified. A total of 19,288 tweets involving the term "smartphone" were collected from 2014.06.01 to 2014.08.31. Among these, a total of 699 tweets were related to UXs of smartphones. In addition, 478 tweets were categorized according to the five dimensions of product smartness (Autonomy, Adaptability, Multi-functionality, Connectivity, and Personalization). Results found that many satisfactory experiences for all dimensions, but there were unsatisfactory experiences associated with the multi-functionality and connectivity. Findings suggest that mining techniques can be used to gather and analyze UX effectively and efficiently. Also, the results of this study can be helpful for understanding user's implicit needs concerning smartphones, and provide valuable insights for developing smartphones.

With the growing interest concerning user experience (UX), it has become a primary consideration to observe and identify users' implicit needs in product development (Kaulio, [19] ; Kraft, [22] ). Design with considering UX is crucial to improve customer loyalty. For example, Jordan ([15] ) suggests that if people have pleasurable experiences regarding a product, they are more willing to buy subsequent products from the same company. Moreover, as the user-centered design (UCD) process has become indispensable in product development, the notion and necessity of UX has been widely accepted (Hassenzahl, [10] ; Kujala & Miron-Shatz, [23] ; Lee, [25] ). UCD involves identifying users' implicit needs and considering context-of-use in the early stage of the new product development (NPD) process. Considering UCD in the NPD process, identifying user needs through observing UX is required. Therefore, it is important to observe and analyze UX for developing new products or services.

Observing UX is complex in that the amplitude of experience varies with each interaction, and the overall evaluation of products changes over time (Karapanos, Zimmerman, Forlizzi, & Martens, [18] ; von Wilamowitz-Moellendorff, Hassenzahl, & Platz, [47] ). In order to understand UX, various observational techniques have been used (e.g., field survey, ethnographic study, task analysis, questionnaires, diary-based method). Ethnographic studies are used to understand the context of users when using technology in real life. Field surveys enable researchers to investigate a large number of participants by use of questionnaires. Task analysis is conducted to observe the exact time for each behavior by use of video recording (Kuniavsky, [24] ). Focus group interviews (FGI) are usually conducted in the early development procedure through a structured group interview. Perception of users, such as priorities and desires, can be identified through FGI. Since experiential aspects change over time, diary techniques are increasingly used. The diary technique is known to be in a mediate position of observation in naturalistic setting and controlled settings (Hyldegrd, [12] ). It has its benefits on observing users' experience for long term, and able researchers to obtain quantitative as well as qualitative data (Kahneman, Krueger, Schkade, Schwarz, & Stone, [17] ).

However, the methods of observing UX discussed above have several limitations. First, it is challenging to understand the implicit needs of consumers through conducting questionnaires based on survey methods. This method does not often provide the underlying meaning of the outcome data, and the answers of participants can be induced by the experimenter as well as the wording of the question (Gable, [8] ). In addition to the limitations of the method itself, previous studies were limited to laboratory experiments in which subjects were first time users of products. In laboratory experiments, researchers cannot completely simulate the real, everyday context of product use. In addition, artificial restrictions on predefined choice sets can affect the outcomes of the experiment (Levitt & List, [26] ).

Hence, it can be more effective to collect unrefined and numerous UXs expressed in the subjects' personal social network system (SNS) platform (or personal channels) for better observation of the user's behavior. Social media (e.g., Facebook, Instagram, Twitter) can be defined as services that provide users the opportunity to establish a connection with friends and share information (Noordhuis, Heijkoop, & Lazovik, [32] ). Among personal SNS platforms, Twitter is a "real-time happening" sharing tool that allows users to post their thoughts or status (called tweets) and follow the status updates of other users in almost real-time (Schonfeld, [41] ). This could make it possible to infer people's opinions, both at an individual level as well as at a mass level, regarding any subject or event (Schonfeld, [41] ).

Observing greater amounts and more natural UX is needed in the development process of smart products. Smart products have been developed and improved by technology-oriented approaches. Developers have been concerned about the available technologies or functions as well as the level of degree and methods of integration (Chan, Estve, Escriba, & Campo, [4] ). As a result, numerous smart products designed in this way have surged to market and the sales volume of smart products continues to grow. Customers are willing to purchase smarter products, but the reality is that users rarely make the best use of them and the smart functions are not fully utilized. The technology-oriented approaches in product design do not sufficiently analyze user needs.

The objective of this study is to collect and analyze natural UX of smartphones using social media data (Twitter). First, we observed UX of smartphones and analyzed data based on product smartness. Subsequently, in order to classify the UX of smartphones, we re-organized the dimensions of product smartness based on the literature review and expert reviews. Finally, to identify the reason of user satisfaction or dissatisfaction regarding smartphone use, qualitative investigation was conducted. Even though many previous studies have been investigated the UX of smartphones concerning various aspects (Nascimento et al., [31] ; Park & Han, [34] ; Walsh, Nurkka, & Walsh, [48] ), most were based on survey methods. Thus, those studies could not effectively observe UX in a naturalistic environment.

Read the introduction. What is the article about and what is the knowledge that it adds to existing research?

How does figure two illustrate that this study is a "mixed-method analysis"? Explain what methods the article uses and how they are mixed in this specific case.