I wonder… Inquiry techniques as a method to gain insights into people’s encounters with visual art

Lucia Marengo, Queen Mary University of London, UK, George Fazekas, QMUL, UK

Abstract

Keywords: visual art, information design, inquiry techniques, user requirements, online collections, interaction design

1. Introduction

The digitization of art collections is a great opportunity for museums and galleries to widen the horizon of their interaction with their audience, and with the community they are part of. Since the late 1990s, the creation of online collections has been encouraged and promoted for its potential impact on the accessibility, visibility, and ultimately democratization of culture (Kahn, 2005). A growing number of institutions are making their collections available online to better attract, build, and engage their audiences (Axiell, 2017); access to digital collections is not limited by physical distance and users are not bound by time constraints.

Part of people’s cultural and artistic diet now develops online, in mobility, and on personal devices. In this context, online collections are one of the assets museums and art galleries can use to reach their audience. Online experiences can be an alternative to the physical visit, as in the case of virtual exhibitions; they can assume new forms and interaction concepts to adapt to the possibilities of the new medium, as in the case of Rijksmuseum, SFMOMA, and Clyfford Still Museum, among others. User studies have shown that, alongside professionals and scholars, online collections already attract casual users or “cultural snackers”; people that consume contents in the context of leisure activities, without a specific information need, and with experiential motivations that range from boredom to improving one’s own personal knowledge (INTK, 2013 and Europeana, 2014). Engaging audiences with online collections poses interesting challenges, not least because motivations to interact—the social and physical context of the online interactions—differ from the physical visit; it thus becomes relevant to understand how users might want to interact with the content offered online. Researchers have mostly contributed to the study of the interaction needs of scholars and professional users, and this has been applied to the development of interfaces for search-oriented tasks (Hearst, 2009), while research is not equally rich on contributions about leisure interactions.

Adopting a human-centered approach, we seek to collect insights from people’s experiences as a source of inspiration and guidance to develop products and services that resonate with the intended audience. In a human-centered framework, designers seek to understand the users’ perspectives as an important step to imagine and improve their creative solutions.

In the present study, we focus on uncovering information requirements—the additional content that accompanies single paintings. We choose to derive insights from people’s curiosities that emerge when they encounter an artwork, following the inquiry technique developed for visible thinking strategies.

In two studies, we asked participants recruited during public events to submit questions that reflect their curiosities about a set of selected artworks; 138 questions were collected and analyzed using a thematic analysis. We developed a taxonomy to classify the questions, and we derived the main themes; finally, using statistical analysis, we segmented the questions based on variables of age, previous exposure to the painting, frequency of going to museums, content, and the style of the painting itself, in order to highlight trends. Results from the insights were used to characterize information interests of casual users, and to discuss design implications. Finally, we discuss the value of the chosen method and possible design opportunities.

2. Background and related works

Academic research that deals with online collections is multidisciplinary and scattered in multiple fields, predominantly design research and computer science (information science and HCI). In the next paragraphs, we review research in two selected areas: design of exploratory interfaces for casual-leisure users, and user studies of online collections. Our review includes projects from online libraries, history museums, and other image collections.

2.1 Designing exploratory experiences for casual-leisure users

The research on casual-leisure exploratory search is fairly recent; Wilson and Elsweiler introduced the following scenario in 2010: The casual-leisure user does not approach the information system with an information goal; if it exists, this is transient, and the user’s need can be met even if no information is found. Needs include a desire to change mood or physical state; to pass time; to further knowledge; or to interact socially (Elsweiler, 2011). Dörk (2011) elaborates on this scenario introducing the information flaneur, a human-centered model for exploratory attitudes in information practices. The information flaneur is intrigued by “evolving information landscapes”; explores information as part of their daily environment; enjoys “bumping into information”; and cultivates an “open and questioning mind.” Their experience is highly personal, and draws from both the real and fictional world. Some researchers drew from these information models to develop design requirements and interfaces, especially to define collections’ visualization and navigation features. Following a prototype-based research, Whitelaw (2015) developed a visual framework specifically addressing accessibility of users with casual browsing behaviors: his generous interfaces offer a full view of the whole collection; visually show the relative number of artifacts each category contains; and surface the relationships between the items. Coburn (2016) developed and tested an exploratory interface to navigate the Tyne and Wear Archive and Museum based on the information flaneur‘s characteristics; the interface is based on the concept of infinite scrolling, and surfaces more or less related items depending on the user’s scrolling activity. Mayr (Mayr et al., 2016) instead offers a list of design requirements to visualize cultural heritage data: providing conceptual orientation; supporting exploratory search; keeping the cognitive load low; and enabling intuitive modes of interactions. All are proposed in response to the non-goal oriented, exploratory user group.

Outside the strictly academic research, museums and art galleries are leading the experimentation on online experiences, proposing a variety of designs and interactions beyond the “exploration of a digital catalog.” On Rijksstudio, users can contribute to the collection by selecting items to curate for their own exhibition, that can be made available to other users too (https://www.rijksmuseum.nl/en/rijksstudio). SFMOMA recently launched “Send me…,” a playful take on the search features to connect their diverse audience with the ever-growing collection: a 1-to-1 text service sends paintings to the users based on keywords and emoticons they submit (Mollica, 2017). The Clyfford Still Museum introduced the slow looking online feature, that allows users to experience paintings from a close and dynamic perspective (cogapp, 2017).

2.2 Collecting information requirements for exploratory interfaces

Many of the methods to improve the knowledge of users’ needs have been developed in goal-oriented scenarios. Choi and Rasmussen (2003) used a “Wizard of Oz” method by manually collecting natural language searches from faculty members in the Library of Congress American Memory image archive. Queries were classified based on the type of information sought as specific/general, abstract/nameable, and subjective, with more than half falling in the “general-nameable.” In a larger scale study, Hale (2016) looked at 3,459 logs in the Auckland Art Museum to determine the most frequently employed search categories. Results show, for example, how department classification had a higher usage than information about the medium, and that people searched by subjects rather than by formal titles.

Online pop-up surveys are a popular method to investigate real-users’ needs. Clough et al. (2017) studied what information visitors were looking for on the Europeana website. 240 answers were categorized according to the search task they entail, following the facet approach by Armitage and Enser (1997); results showed that “general topical search” was the most frequent activity, and that academics submitted a higher proportion of specific-item searches, while cultural heritage enthusiasts mostly engaged in browse/explore searches. Skov (2009) took a similar approach: a Web survey was distributed to users of The Military Museum in Copenhagen. Results helped to differentiate between collectors that mostly submitted specific-item searches, and art enthusiasts users, that display ill-defined information needs. Simulated search task situations can also be used: 34 users in a lab were given a description of a real-life context, and asked to perform a search to meet their information needs (Skov, 2009).

Another group of studies focused on information requirements people have after accessing a specific item in the collection. Kravchyna’s Doctoral Thesis (2004) identifies users’ preferences on the information available for current exhibits in selected museums in Texas. 550 self-selected respondents from the general public, scholars, and curators completed a Web survey including the question “what information about a work of art would you like to access on the museum website?” Participants selected one or more options from a closed list that included, title, name of creator, dimension, etc. The research showed that, while title and artist are interesting to the majority of users (92-91%), contextual and historical information is relevant to the scholar user group alone. Similarly, Skov (2009) asked hobbyist users to indicate their usual preferred data elements, choosing from a list of options. The results were classified as factual, object related, visual, and historical-context related. Results show that collectors in their leisure tasks mainly need object-related information, while art enthusiasts are interested in a wider range of information types. Finally, Jörgesen (1998) and Hollink et al. (2004) generated the category classes for an online collection based on the keywords defined by users to describe a selection of images of the collection itself. While it has been argued that the keywords are not indicative of an actual search need, and thus that the method might not be appropriate to define indexing and browsing categories for task-based searches, this approach can be relevant to investigate how users read and engage with images, and also in non-task based activities, as is the case in our research.

3. The study

In previous studies, interfaces have often been informed by user models in information science and not by direct user research, while in other cases, it is unfortunately difficult to derive the research and design concepts that led to the development of the interface. Contributions in the academic field focused on experimenting navigational features, while no similar interest was found for other aspects of the exploratory experience. Finally, methods to derive information requirements are designed around search-oriented scenarios (they often aim at improving the collection’s indexing and retrieval system), but they might not be able to capture casual-leisure users’ ones.

Our work contributes to the field by adopting a human-centered approach and by looking at people’s interaction with the information that surrounds the art piece as a source of insights for the design of meaningful experiences. Human-centered design methods are increasingly popular in the museum science field, though they are not yet extensively applied to online collections.

Instead of search-based requirements, we collected curiosities as insights into the user-item interaction: What surfaces in the mind of the users when they see a work of art? Intrinsic curiosities, that do not depend on external factors (including the use of a specific system), are a powerful source of motivation for people’s exploratory activities (Loewenstein, 1994). We assume that designing around them may be a powerful leverage for exploratory experiences.
The method is inspired by Project Zero’s Visible Thinking strategies, that are used to formalize and share internal thoughts when exploring a work of art (http://www.pz.harvard.edu/projects/visible-thinking).

To summarize, our study does the following:

• investigates casual-leisure interactions with artworks;
• adopts a human-centered approach;
• assumes intrinsic curiosities as a source of knowledge;
• derives insights from information requirements in the interaction with single items.

3.1 Methodology

The study was conducted in a mixed controlled space in two iterations. Participants were selected on an opportunistic basis during two public events: the Festival of Communities at Queen Mary University, and a Cryptoparty workshop at Mile End Pavillion, London. Both events feature stands where people can engage in interactive activities. In the first case, the stands were presenting research developed at Queen Mary; in the second one, workshops on the theme of encryption and privacy were offered. We invited visitors to explore a mini art gallery with printed versions of selected artworks; after choosing one of them, they completed a Post-it that started with the sentence “I wonder…” and stuck it next to the print. Finally, participants were asked to complete a post-experience questionnaire, designed to collect basic demographics and feedback about the process.

The design was slightly different during the two iterations, with the first one serving as a test to fine-tune the design of the second one. In the first iteration, participants were presented with 16 different paintings. Our selection aimed at providing a range of styles, time periods, and infamy (see Table 1 for details). Participants were free to contribute to one painting or more. It must be noted that, although we were not able to include their data in the analysis, the activity was offered to kids, too, and their Post-its were stuck alongside the adults’, and visible to all. More importantly, an element of playfulness was added for the kids; they were selecting the starting words of their questions by casting two big foam dice, which included the words “who,” “what,” “how,” “when,” “might,” “could,” “was,” and “will.”

The second iteration followed a formal within-subject design; six artworks were selected, and each participant had to write at least a question for each painting. Contributions from persons under 16 years of age were also included in the analysis, and we eliminated the use of the foam dice. Selection criteria for the paintings included reputation (half of the paintings were renowned); the presence of living figures; and style (in particular, the level of realism)—two aspects that in the first iteration seemed to have determined different types of questions. Techniques and materials were not considered, as the reproduction quality did not allow appreciation of differences in strokes/canvas (see Table 2 for details).

Data were first analyzed in a qualitative way. We performed a content analysis following an open coding approach in two iterations. The code structure was improved with data from the second iteration using a constant comparative method. Since subjectivity is intrinsic to the qualitative analysis methods (Braun, 2006), a second coder performed the analysis on a subset of 50 paintings. The resulting themes were then discussed and compared with the existing list, and the final categories were structured into a taxonomy.

Finally, we performed statistical analysis on the data from the second iteration with the aim of exploring possible users’ segmentations into groups with specific interests. Age and level of expertise were chosen as grouping variables because they identify traditionally relevant user groups in the design of museums’ experience. Previous knowledge was derived from the definition of curiosity itself, something people do not know about: we hypothesized that people with higher knowledge about the painting would come up with different questions, possibly more specific, or more abstract ones. To partially account for the countless differences between the paintings, we evaluated the effect of the subject depicted, and the painting style. Statistical analysis was performed using a Chi-square test with significance set at 0.05. When the count of cases for each variable combination was lower than five, (thus rendering the Chi-square test unreliable), and when possible, we opted for the Fisher Exact test. We used SPSS to run the analysis of both the Post-its and the post-experience questionnaires.

3.2 Results

22 people participated in the first study, and 15 in the second, contributing 56 and 82 questions respectively. Figures 1 and 2 describe the participants’ demographics:

Figure 1: descriptive statistics of participants for Study 1

Figure 2: descriptive statistics of participants for Study 2

Overall, participants did not have particular difficulties with the activity; the mean of Easiness, measured in the questionnaire, supports the use of such method (mean 3.7 over 5 and STD 0.66). Observations showed that visitors that were not fluent in English had limitations in expressing their curiosities. Others declared they were not comfortable experiencing art without information. Finally, some people might not be comfortable in verbalizing what is a personal contemplative experience.

Some people might have been influenced by what previous participants had written, but overall, existing Post-its worked as positive elements: People were attracted to the Post-its, and enjoy reading them. This indicates an interest in a playful approach to art, and that other people are a source of inspiration.

3.2.1 Content analysis: classifying the users’ questions

Results from the thematic analysis help in describing the focus of the questions. Figure 3 summarizes the codes into a taxonomy that describes “What information is the participant looking for?” The classes are described below.

Most questions, 68.8%, were about the depicted scene. They considered the painting as a window to a represented world. The context the images were presented in (not in a museum, without a frame, not in the original size) probably influenced this interpretation, though it reflects fairly well how images appear in digital environments, too. The thematic analysis in the next section outlines this class in more detail. 8.7% of the questions considered the physical nature of the artwork and included questions about its dimensions, material, and texture; about the process that led to its creation; how long it took; whether the subject was depicted in real life or not (painting as material object). The artist was the main subject in 10.1% of the questions, which investigated how the artist came to paint the artwork, and what inspired them, both in terms of life events and emotional baggage. 10.9% of the questions focused on the historical context in which the painting was created, including its date and place of origin; style or movement it belongs to; and its connections to other artworks or historical events. People were also interested in the artistic value that has been attributed to the painting and how it came to be part of a museum. Two questions did not fall into the previous categories and were open-ended questions on broader themes the painting inspired: “How long is the future?” (ID:7).

Figure 3: schema developed to describe users’ curiosities

Below we give a richer description of themes emerged from curiosities regarding the scene (the upper branch of the schema in Figure 3):

• Description: people looked for a description of the scene; they wanted to identify characters, time and place, and events related to the image: “Who is the woman?”(ID:5); “When did this take place?”(ID:1); “What is going on?”(ID:6).
• Meaning & Symbols: people went beyond the literal representation and looked for metaphors, symbolism, and generally for the message the artist wanted to convey with the artwork: “What do the colors mean?”(ID:13); “Why are the stars massive and yellow?”(ID:17); “Why is it a mix of a human and a deer?”(ID:15).
• Emotion & Empathy: people wondered about the inner world of the characters, their emotions, and thoughts. Their questions sometimes implied empathy and compassion towards the characters: “What was her life like?”(ID:5); “I wonder what in her life was so bad that she felt she did not want to carry on living”(ID:9). This theme emerged in the artist category, too: “What was the mood of the artist as they painted it?”(ID:17); “What goes through the artist’s mind when they painted this?”(ID:14).
• Immersion: people imagined themselves as part of the painting and wondered how they would live or feel in the scene: “What could I do if I were living in that palace?”(ID:1) could be interpreted as another type of personal involvement, similar to empathy.
• Narrative & Storytelling: people imagined events around the depicted scene, focusing on what happens in the characters’ past and future: “Will the two people get on the boat?”(ID:10); “What could happen if he falls?”(ID:7).
• Implicit elements: people sometimes asked questions about an element not directly represented in the painting, but that was somehow suggested by visual hints, or that they purely imagined. The arrows in the deer’s body suggest the presence of persecutor in “I wonder if the killers have a story to tell?”(ID:14), and a misplaced object suggests the presence of someone in “I wonder who knocked over the bowl?”(ID:12).

The process described in the last two themes resembles what we do with photographs and cinematography, where we consider the image as a real world that lives beyond the moment captured in the painting. The last three themes show a trend among participants towards imagination, and were further grouped into the general theme of Fantasy.
The 94 questions about the scene plus four questions about the artist were assigned a theme. Description appears in 51% of them, Meaning 28.5%, Emotion 10.2%, and Fantasy 10.2%.

Additional themes emerged when looking not only at the content of the questions but also at their form; how the information was sought. 26.1% of the questions were not purely informational; instead, they included the participant’s interpretation or hypothesis on the matter, and the question was somehow a request to clarify their doubts. Such questions could be expressed in the form Is it true that…? We found people building theories in the use of color for a cubist painting: “I wonder if the variation might be associated with the environment (eg red=sun, blue=water, green=grass)?”(ID:13); identifying the place a scene is set; “Is that Venice?”(ID:11); linking the scene to the life of the artist; “Did the artist have a daily view of this scenery?”(ID:8). Sometimes the hypothesis was extremely specific, and probably drew from partial information the participant had or had heard of, for example,  “Did Van Gogh have a medical condition that made him see the world in such a unique way?”

Most questions seemed implicitly addressing art experts or the museum itself; interesting exceptions were questions directed to other visitors or peers: “Does anyone else feel cold looking at this?”(ID:17); and others still were self-reflective or open-ended: “What could I do if I were living in that palace?”(ID:1).

82.6% of the questions were “complex” in that they required more than a one-word answer. No information was provided about the painting, yet nobody asked for the title; only one person asked the name of the artist, and five people asked about the time and place in which the painting was created.

The categorization was ambiguous at times, as natural language analysis is a lengthy process that requires some degree of interpretation. For example, the question “What volume of paint did the artist use up for this painting?” could be coded in the Artist or Object category. The question “Why (are) there worm stars?”(ID:17) has been interpreted as “What is the choice behind representing stars this way?” and therefor coded in the Meaning and Symbols theme; if we assume the participant is rather wondering about events or rules of the depicted world that led to the stars’ appearance, we could code it in the Narrative category.

3.2.2 Statistical analysis: which factors might help segment the users’ interest?

Study design: themes of immersion, narrative, and implicit elements appear only during the first iteration. We hypothesized that adults were inspired to write more creative questions, after reading the playful ones kids wrote after rolling the dice.

The statistical analysis was performed on data from the second iteration alone. Participants were included in this analysis if they submitted at least half of the requested Post-its (3/6).

Age: 29% of the users did not submit any question including hypothesis, while the rest submitted a mix of passive and active questions; adults submitted 19.2% more active questions compared to underage participants; participants under 16 years old focused more on questions about the scene and less on the painting as object or the artist, which adults seemed more interested in. Four kids’ questions about historical context are either simple ones (asking the year), or they came from one kid that seemed to have a specific interest in how we define an artwork to be such. Emotional and narrative questions came exclusively from adults that had a medium-high previous knowledge about the painting. Since these themes had a low count, we focused the remaining analysis on descriptive and meaning themes. Participants under 16 years old focused significantly more on the meaning and symbols of paintings, with questions around surreal elements, and their appearance; adults proportionally asked more descriptive questions (p=0.016).

Frequency visiting museums: we found a negative correlation between the presence of active questions and the habit of visiting museums (p=0.027). People that rarely go to museums submitted 38.6% more active questions compared to medium visitors, and 51.9% more than regular visitors. We address this peculiar result in the discussion section.

Previous knowledge: active questions were 14.4% more common when participants had a low-to-medium previous knowledge of the painting compared to well-known artworks. The descriptive theme is more common if the painting is familiar, otherwise meaning related questions are more common (p=0.011).

Painting subject and style: as expected, the focus of questions was slightly different for each painting (Figure 4 shows values in detail). When grouping paintings by style, questions on the (single) abstract painting focused 29.5% more on its physical characteristics, and less on the artist or the historical context. Paintings with surreal elements attracted more questions about the scene and about the artist (although we believe the reason might be Frida Kahlo was the subject of the painting we chose). Realistic paintings mainly attracted questions about the scene. The presence of living figures does not seem to influence the focus of the questions. Realistic paintings attracted fewer questions on meaning, which are more common in the surreal and abstract ones; the opposite is valid for descriptive questions. Figure 5 shows the frequency of themes by painting. Finally, realists paintings attracted 12.7% less active questions than paintings with abstract or surreal elements.

Figure 4: frequency of questions’ content by painting

Figure 5: frequency of questions’ main themes by painting

4. Discussion: developing insights from the results

First, we summarize the results and characterize the curiosities we collected:

1. Curiosities are contextualized: people’s interest is in information closely related to the painting they are watching, more than in general information. Meaning, they asked about the artist’s motivation and inspiration in creating THIS artwork, not general information about their life;
2. Images are interpreted as scenes, regardless of the subject and style. This might relate to the context the image was presented in (not in a museum, without a frame, not in the original size). Less frequently, people interpreted the image as its physical counterpart or as an artist’s work;
3. People are drawn to meanings, including symbols and the artist’s message;
4. Curiosity for art develops on the emotional level too; we isolated empathy with the characters and with the artist as two types of emotional connection;
5. Curiosity can develop on the imaginative level, though for adults this seems related to external factors (the presence of imaginative Post-its), that work as primers towards a playful attitude;
6. People assume an active role and engage in active interpretations;
7. Questions’ complexity: people rarely looked for basic information or one word answers. When people had little knowledge about the painting, and especially kids, they asked more meaning related questions than descriptive ones;
8. Characters in the interaction: questions seem to be addressed mainly to museums (or art experts); interesting exceptions are peer visitors and reflective questions.

The insights above can be used to derive some implications for the design of exploratory experiences, such as the following:

• Atomic information (e.g., tabular data) might not be suited for casual users with no previous knowledge of the painting. If people do not know much about a painting, the name of the artist alone does not give them much more knowledge or space for reflection; it remains sterile or it might require them to drift off to find more information. Rich, self-contained information might be more fulfilling. Atomic descriptive information might be more suitable when people can use it to build on top of existing knowledge (i7).
• Contextual instead of general information: e.g., initially people might be more curious about the life of the artist contextualized for the specific painting they are observing, not their whole biography (i1).

However, our study explored people’s curiosities—it did not test their behavior using a real system. The findings should not be directly interpreted as implications, and would need to be tested in dedicated studies. On the contrary, the insights are relevant if interpreted as design opportunities to develop new human-centered experiences and systems. We propose some of these ideas below:

• Engage visitors with information on the emotional and symbolic level (i3-4);
• Structure information as storytelling around the scene, including characters and events before and after the depicted scene (i5);
• Interactive discovery of information: the exploratory experience as an active journey, e.g. the information is initially hidden, and the user is asked to select the correct option by comparing the current painting to similar ones (i5);
• Consider that the dialogue with art information falls on a range between social connection and personal/intimate/reflective experiences (i7).

Some insights echo design concepts that are applied in digital experiences on-site and online; the VR Experience of the Ochre Atelier at Tate relates to empathy with Modigliani (http://www.tate.org.uk/whats-on/tate-modern/exhibition/modigliani/modigliani-vr-ochre-atelier); the physicality of paintings is highlighted through high-resolution versions of the digital paintings in the Gigapixel project by Google Art and Culture. Researching how these insights can be embedded in online experience, and how the design choices affect the visitors’ interest and engagement, is of great interest. Lastly, the study opens up interesting questions that challenge the design and testing of new interfaces:

• Are people that often go to museums less prone to submit active questions? Are (some) museums communicating an authoritative identity that is not open to other interpretations (people “do not dare” to give their opinion); and is this present on online interfaces too? Can we change this perception by design?
• Are current interfaces pushing the users’ interest towards the scene alone? How can we design specific experiences that focus on the painting as a physical object and as a work of art?
• Paintings have potential to engage the audience on an imaginative level: how can we support this attitude and induce a playful engagement online? How does this attitude relate to engagement?
• Can we challenge the profile of the information flaneur by introducing interactive experiences during the exploration? How cognitively heavy should they be?

5. Conclusions

Our work aimed at uncovering insights about what information people might find interesting during art exploration. We collected and analyzed 138 questions people wrote during encounters with artworks. Data were analyzed qualitatively through content analysis, and the existence of user groups was evaluated through statistical analysis of selected variables. Our results highlighted that people’s curiosities are highly contextualized, and that artworks are interpreted mainly as scenes. People are interested in meanings and symbols, less in atomic than complex information; they displayed use of fantasy and empathy, as well as an active interpretative role during the inquiry process.

No strong finding emerged from statistical analysis; we noticed that adults and kids likewise displayed an active role in the inquiry process, and that a previous knowledge of the painting is connected to more descriptive and atomic curiosities. Interesting differences were found when segmenting the questions according to single paintings, and to style. However, the findings are not supported by statistical significance and should be investigated in dedicated studies. Additionally, the chosen variable to assess the art expertise (frequency going to museums and art education) were not meaningful in highlighting differences of art knowledge among users.

Finally, we derived and discussed implications for design, and new questions inspired by the findings. Insights suggest design opportunities might lay in the interactive discovery of information, in storytelling-based descriptions, and in emotional connections with the characters and the artist.

We must address some limits of our research. Participants were not recruited from the audience of a real museum: They might be representative of people interested in art, but they do not represent online users. As mentioned, the results do not reflect interaction behaviors with real systems, thus preferences should be used as hints and tested with prototypes before being transferred into interaction requirements.

Despite this, the study was successful in testing the method, which proved to be a rich source of insights and creative design thoughts. People were not influenced by the use of a specific interface, and were not limited in expressing their information interest, as in other reviewed studies. It would be interesting to replicate the study with audiences of existing museums and to test the method within design teams for preliminary user research. What use of the insights would they make in their current practice?

Acknowledgment

This work was funded by EPSRC through the Media and Arts Technology Programme, an RCUK Doctoral Training Centre EP/L01632X/1.

References

Armitage, L., & Enser P.G.B.. (1997). “Analysis of user need in image archives.” Journal of information science. Sage Publications Sage CA: Thousand Oaks, CA, 287-99.

Axiell. (2017). “Museums accelerate implementation of digital strategies, making more content available online and on-site to improve visitor experiences.” Published June 28, 2017. Consulted January 10, 2018. Available http://www.axiell.com/news/museums-accelerate-implementation-digital-strategies-making-content-available-online-site-improve-visitor-experiences/

Braun, V., & V. Clarke. (2006). “Using thematic analysis in psychology.” Qualitative research in psychology 3.2, 77-101.

Choi, Y., & E. Rasmussen. (2003). “Searching for images: The analysis of users’ queries for image retrieval in American history.” Journal of the American Society for Information Science and Technology, 498-511.

Clough, P., T. Hill, M. Paramita, P. Goodale. (2017). “Europeana: What Users Search For and Why.” In International Conference on Theory and Practice of Digital Libraries. Springer, Cham, 207-19.

Coburn, J. (2016). “I don’t know what I’m looking for: Better understanding public usage and behaviours with Tyne & Wear Archives & Museums online collections.” MW2016: Museums and the Web 2016. Published January 29, 2016. Consulted January 10, 2018. Available https://mw2016.museumsandtheweb.com/paper/i-dont-know-what-im-looking-for-better-understanding-public-usage-and-behaviours-with-tyne-wear-archives-museums-online-collections/

cogapp. (2017). “Clyfford Still Museum. Online collection and slow looking.” Published September 20, 2017. Consulted January 10, 2018. Available https://www.cogapp.com/clyfford-still-museum-online-collection

Dörk, M., S. Carpendale, & W. Carey. (2011). “The information flaneur: A fresh look at information seeking.” In Proceedings of the 29th SIGCHI Conference on Human Factors in Computing Systems, 1215-24.

Elsweiler, D., M. Wilson, L. Kriegaard. (2011). “Understanding Casual-Leisure Information Behaviour.” New directions in information behaviour. Emerald Group Publishing Limited, 211-41.

Europeana. (2014). “Results of the Europeana user survey 2014.” Published June 29, 2014. Consulted January 10, 2018. Available http://pro.europeana.eu/blogpost/results-of-the-europeana-user-survey- 2014

Hale, M. (2016). “In Search of Art: A Log Analysis of the Ackland Art Museum’s Collection Search System.” Art Documentation: Journal of the Art Libraries Society of North America, 182-205.

Hearst, M. (2009). Search user interfaces. Cambridge University Press.

Hollink, L., A.T. Schreiber, B.J. Wielinga, & M.  Worring. (2004). “Classification of user image descriptions.” International Journal of Human-Computer Studies 61.5, 601-26.

INTK. (2013). “Visitor motivation in online museum audiences.” Published October 1, 2013. Consulted January 10, 2018. Available http://www.intk.com/en/ideas/visitor-motivation-in-online-museum-audiences

Jörgensen, C. (1998). “Attributes of images in describing tasks.” Information Processing & Management 34.2-3, 161-74.

Kahn, R. E., B. Wilson, A.L. Powell, A. Friedlander, & R.L. Larsen. (2005). “Funding for digital libraries research: Past and present.” D-Lib Magazine.

Kravchyna, V. (2004). “Information Needs of Art Museum Visitors: Real and Virtual.” Ph.D. Thesis. University of North Texas.

Loewenstein, G. (1994). “The psychology of curiosity: A review and reinterpretation.” Psychological bulletin 116.1, 75.

Mayr, E., F. Paolo, S. Miksch, G. Schreder, M. Smuc, F. Windhager. (2016). “Visualization of Cultural Heritage Data for Casual Users.” In Proceedings of the 1st Workshop for Visualization for the Digital Humanities. Baltimore, MD.

Mollica, J. (2017). “Send Me SFMOMA.” Published June 16, 2017. Consulted January 10, 2018. Available https://www.sfmoma.org/send-me-sfmoma/

Skov, M. (2009). “The Reinvented Museum: Exploring Information Seeking Behaviour in a Digital Museum Context.” Ph.D. Thesis. Royal School of Library and Information Science, University of Copenhagen.

Whitelaw, M. (2015). “Generous interfaces for digital cultural collections.”Digital Humanities Quarterly 9.1, 38.

Wilson, M. & D. Elsweiler. (2010). “Casual-leisure Searching: the Exploratory Search scenarios that break our current models.” In Proceedings of HCIR 2010, 28-31.