Challenges of Designing with Open Data: The Case of Cultural Data

2. 1. 1. Open Data During the Discover & Define Phases

Researchers have relied on data to understand users. In particular, they have used open data to understand the context of the target user. For instance, the Data Material Book explains the use of government data sets (data sets from different ministries from sectors such as state budget, traffic, or environment) during the discover and define phases (Otto, 2018). However, while the open data provides useful information, in a relatively inexpensive way, it reflects the difficulty of locating specific information among a large amount of data sets.

Among the studies in the HCI (human computer interaction) field, there are cases where the target user is defined through utilization of data. Of course, this is not an instance of open data, but it does exemplify the mode by which designers come to understand the problem situation and users with large amounts of data. For example, Zhang et al. presented a quantitative bottom-up data-driven approach to create objective personas (Zhang, Brown, & Shankar, 2016). They assembled user behavior via clicks gathered automatically from telemetry data related to actual product use in the field. They then aggregated 3.5 million clicks from 2400 users into 39, 000 clickstreams. Then, using a statistical approach, the data were used to generate five representative personas. According to the study, the five generated personas were validated by user behavior experts (via interviews), who verified that these personas did, in fact, represent archetypal users of our product.

2. 1. 2. Open Data During the Develop Phase

During the develop phase, design concepts are created through a process of idea generation. While there are various design methodologies used for concept generation, a methodology that relies on the utilization of open data for concept generation is rare. Notwithstanding research on data visualization, not many studies provide guidance on how to generate a design concept that uses open data as a core component of a product or service.

Although the Design Human Data Kit is not designed for open data use alone, it furnishes useful tools for the systematic analysis of data and design concept generation (Otto, 2018). The Design Human Data Kit is a set of tools and information that enables designers and their interdisciplinary teams to create sustainable and user-centered data services. This includes basic personal data, as well as tools for the design process, to facilitate discussions on the creation of data services. In particular, the kit includes the following design tools: Personal Data Cards, Privacy Model, and Data Canvas.

• • Personal Data Cards is a compilation of cards to analyze and discuss aspects of people, ethics, and data service security. This compilation of cards covers different aspects besides these requirements and places the focus on humans in form of personal data sources, security measures and ethical aspects.
• • Data Canvas is a framework to examine the developed ideas from different perspectives. Similar to business model canvas, Data Canvas can be used as a template for developing new service models or documenting existing ones, on the basis of data sets.
• • Privacy Model helps to assess the sensitivity of people’s personal data. This model shows the different types and levels of personal data from humans considering transparency, value and control.

The kit can help designers attain experience in and develop attitudes toward the use of personal data, including ethical issues. The Personal Data Cards will be especially useful at helping designers understand the source of certain data and how to store and handle such data in the context of their design concepts. Further, the Personal Data Canvas and Privacy Model may help designers to validate their design ideas from various perspectives.

2. 1. 3. Open Data During the Deliver Phase

Finally, open data may be also used in the deliver phase of design. Indeed, for concept validation, open data sets can only be used to test a service concept. For example, recommendation systems are often tested against standard data sets, such as MovieLens (http://grouplens.org/datasets/movielens/) which has gathered user ratings across many movies. Designers and developers can then test their new algorithms against the data set to verify predictive accuracy regarding user ratings and evaluate its performance, relative to the performance of numerous other algorithms. Likewise, the use of open data may not be reflected by the final service. However, the data are acknowledged to play a critical role in enhancing the quality of the service and ensuring its viability.

3. 1. 1. Participants

We conducted a design workshop with 40 design-major students (age = 23.2, SD = 2.2). Fourteen groups, each composed of two to three students, were organized during a workshop. On average, participants had studied industrial design and UX design for 3.4 years. Of the 40 participants, 13 respondents asserted knowledge of the concept of open data. However, none of the 40 respondents claimed to have conducted a design project using open data. These participants were suitable for our study, due to their experience designing various services, products, and mobile applications and knowledge of the general design process. Throughout the design workshop, we were able to ask the participants to compare the general design process with the design process that uses open data. We also analyzed their use of open data during the design process.

3. 1. 2. Workshop Materials

In the first week of the design workshop, we briefly introduced the concept of open data and showed participants the website where they could find open data.

For the purposes of our research, we narrowed our focus to the use of open data in the context of various cultural areas. Open data on culture was suitable for use in our experiments, because it included geographical data, textual data, images, and so forth. Such attributes render cultural open data similar to general open data, except for its restriction to culture-related topics. In particular, because we provided the data sets through one data portal (http://www.culture.go.kr/data/), participants easily searched and explored the data sets during the experiment.

The cultural data portal provides open data on art and culture, cultural heritage, cultural industry, tourism, sports, books, policy support, and cultural promotion. This site provides information on the organizations active in each field and introduces open data sets provided by such organizations provide. Users can search and retrieve various cultural data, including items on local festivals, cultural assets, performances, exhibitions, books, sports facilities, copyright, cultural-content industry, and the Korean language through each institutions API and website.

Although, in our study, we asked participants to principally use cultural data, we encouraged them to seek out other open data sets and incorporate data from those as well.

4. 3. 1. Open Data Filter for Target Group

The most popular type of outcome designed by participants was a mobile application or website that helped consumers search for culture-related information. Because of the overwhelming magnitude of available open data, these cases were designed to provide filtered data, tailored to a target audience, on cultural events, attractions, or historical sites.

For instance, Great Mom, a mobile application for mothers with infant children, allows users to browse places or events they may want to visit; it operates by filtering information on the basis of their region, as well as their children's tastes and ages. Another case is Road Friend, which provides information on cultural experience programs, by region, and enables users to book them. The One Bowl application introduces Korean food recipes and allows users to order ingredients (Figure 1). Unlike the two previous cases, this application also utilized multimedia data, such as photographs and images.

4. 3. 2. Open Data-based User Participation Platform

Another type of design case was a platform, based on open data, which supported interactions between users within the platform

For example, Party Together is a web-based platform that links users who want to host an event with companies that help prepare and administer such events. This service uses open data to register the schedules of nationwide events and related business listings. The platform generates revenue by mediating user-company encounters: “Open data was essential to devise our idea. It will be possible to create a foundation of our service platform with open data and use it to drive users to our services when there is insufficient data on our platform in the early stages of service” (P39, designer of Party Together). Similar to open-data filter applications, the Traview application, which recommends tourist attractions (Figure 2), is also a platform where users can upload realistic video reviews and generate a user community. To encourage more participation, users receive bonus points, which can be used during travel.

We found that designers tried to build their services on a foundation of open data, which is free to the public and represents an extensive amount of information

4. 3. 3. Open Data as Aesthetic Material

Among the cultural data sets, some could be utilized as aesthetic material. Open data on traditional patterns, in particular, were used twice among the design cases. For instance, Postcard Printer is a kiosk at cultural heritage sites or sightseeing spots that helps visitors craft unique postcards. Visitors can design postcards by combining traditional patterns and sightseeing photos through the kiosk interface. Similarly, My Phone Case is a kiosk that engraves traditional patterns on smartphone covers (Figure 3). Using the kiosk, users can generate custom designs and engrave them on smartphone covers. The participants noted that they wished to generate visually aesthetic graphics by editing the open data sets to adapt the traditional beauty of the patterns to a more modern style sensibility.

In addition, data in the form of photographs of historical remains and artifacts were also frequently used. Play Pamphlet used image data on cultural sites or artifacts to make visually appealing pamphlets. By printing images on foldable structures, users can assemble the pamphlets into souvenirs: “We gathered free image data of the ruins in the data portal. Then, we tried to create additional value by editing them” (P9, designer of Play Pamphlet).

Similar to the cases above, several open data sets, in the form of images, were used to produce objects that people could physically own without employing designers’ complex data alterations and aggregations.

4. 3. 4. Copyright-free Open Data Provider

In the case of cultural data, some open data consisted of novels, poetry, and paintings, which represents data of substantial economic and cultural value. In particular, some data sets lacked legal copyrights, thereby rendering them available for use by anyone. Some participants utilized this data to create a design type that enabled the public to easily access copyright-free data sets.

Hand-span Novel is a service that requests payment from passengers of public transportation to read a portion of an article or novel (Figure 4). The participants who designed this service explained that they had discovered a considerable amount of copyright-free literature from which they could profit.

This type of design case identified and distributed open data of intrinsically high economic value. The designers thereby constructed a service channel to facilitate the distribution of valuable open data in a way that avoids the cumbersome processes of data searching, packaging, and transformation.

5. 2. 1. Limitations of Current Types of Data Utilization

Data science can be understood in terms of seven stages: acquire, parse, filter, mine, represent, refine, and interact (Fry, 2004). In other words, data requires many processes, including aggregation, to be meaningful. Tracking the number of steps that people take in a day is worthless, unless they compare that number to a reference standard, and tracking the steps taken in a single day yields little information about overall activity or fitness. While digital environments increase the potential scale of aggregation tremendously, our experiments show that designers have difficulty performing data aggregation (such as transforming or combining open data sets). As shown in the design cases, our participants tended to focus on the types of open data that seemed immediately applicable. In other words, their lack of knowledge on how to utilize open data in various ways mandated their regard of the instant availability of the data itself as the most important factor.

In particular, design types Open Data as Aesthetic Material and Copyright-free Open Data Provider demonstrated that designers tended to focus on instant availability. These two design types relied on aesthetically and economically valuable data sets. The former utilized data on image types with high visual value. For example, designers directly applied images of traditional patterns or artifacts provided through the data portal. The second type used text-type data with a high literacy value. However, one of the biggest challenges associated with those types is the limited amount of high-quality data. These types of cases may be successful or forced to terminate, depending on whether government agencies continue to discover and provide new data to fuel the designs. Thus, designers need to investigate novel ways to create additional value through data aggregation.

The Open Data Filter for Target Group was the most common design type in our study. Participants designed mobile/web applications that provided users with open data on restaurants and tourist sites. However, relative to existing applications on the market that use the same data source, the competitiveness of the design cases is not high.

Moreover, effectively designing the provision of open data is not just a matter of attractive presentation. It is to list more than one kind of data sequentially. Like prior research projects, designers must be able to assemble insights from among vast quantities of open data to build services that help users browse for data as they prefer (J. Kim et al., 2015; N. W. Kim et al., 2016). Another possible method involves deriving a new meaning from alternative ways of combining additional data. For example, recommending a tourist route that accounts for the weather, by including weather data, would be more helpful to users than simply presenting tourist information, a feature many other platforms have.

5. 2. 2. Proposal of Design Methods

As described above, despite the open data’s vast potential, designers who were unfamiliar with data use tended to lack understanding on how to utilize open data. In our experiment, no participant group performed any high-level analysis of open data provided by an institution, and they rarely synthesized various types of open data to generate additional value from the data they were using. This may be due to the fact that, in general, designers rarely understand the nature of open data and how to use it in design process.

According to the work conducted by Feinberg (Feinberg, 2017), designers should be able to understand the context of open data-how it was collected and for what ends-and the utilization techniques of data. One possible way to improve designers’ data literacy would be to focus on design methods. The role of design methods is to support design work, the aims of which can vary. Such aims may include gaining key insights or unique essential truths that result in more holistic solutions, thereby achieving better experiences for users, through products, services, environments, and the systems on which they rely.

One possible method is to adopt the design of Data Mixer, which was devised to support the combination of diverse open data (Figure 6). It can help designers identify associations and values among different types of open data that designers may previously have regarded as irrelevant. For instance, rather than just listing data on a tourist site, designers may be able to combine open data on weather and on the number of visitors to a tourist site with additional personal data on individuals’ consumption or driving history. If there is a design method that helps designers anticipate the outcome of combining data sets, they may be able to create a service that recommends tour routes suitable under diverse conditions.

Another possible method is to learn from the Data Recipe Book, a collection of recipes describing various ways of utilizing the most popular open data. The current data portal already shows good examples that utilize open data. However, it lacks sufficient descriptions on the types of open data being used and how that data is applied to a given design case. In a format similar to that of a Personal Data Canvas (Otto, 2018) (Figure 7), designers may be able to share information on collecting, transforming, and storing open data. Then, designers will be able to learn from data recipes that provide various examples of how designers and data scientists can filter, analyze, and combine open data.