Tracing exploratory modes in digital collections of museum Web sites using reverse information architecture
First Monday

Tracing exploratory modes in digital collections of museum Web sites using reverse information architecture by Sarah Kreiseler, Viktoria Bruggemann, and Marian Dork

Museums are broadening their program beyond the physical institutions by providing digital collections online. In digital collections, objects are prepared and presented particularly for the Web and the ambition is to provide the entirety of a physical collection. To make these rich and comprehensive data sets accessible, an explore mode is increasingly offered. The present study considers this mode, first by making sense of the term “exploration” and suggesting four functional principles in support of exploration in digital collections — view, movement, contextualization, and participation. On this basis, we compare eight well-known museums with regard to the explore modes for their digital collections. We have devised a three-part methodology, reverse information architecture, to address the question: How is the function of exploration manifested in the structure and interface elements of digital collections? With this unique method we use the given content to investigate how far the four principles are implemented in explore modes of digital collections and, broadly said, how explorable they are. The introduced approach to studying digital collections could be opened up to other fields to analyze a variety of Web interfaces in general.


1. Introduction
2. Towards a more exploratory mode in information retrieval
3. Method of the study
4. Findings
5. Discussion
6. Conclusion



1. Introduction

The attractive and engaging presentation of the collection has become just as important for museums as the tasks of collecting and preserving the cultural objects. As the International Council of Museums (ICOM) defines it, a museum is an institution “which acquires, conserves, researches, communicates and exhibits the tangible and intangible heritage of humanity and its environment for the purposes of education, study and enjoyment” (ICOM, 2007). Additionally, the World Wide Web enables museums to increase their outreach and provide access beyond its physical manifestation. One particular possibility to use the Web as an extension to the physical museum is the offer of digital collections. Padilla-Meléndez and Águila-Obra (2013) showed that among the 40 most visited museums around the world, 87.5 percent provide access to an online database.

In digital collections, museums are presenting digital copies of their objects online. Instead of presenting them by reproducing the physical exhibition rooms (virtual museums), digital collections process the objects especially for the Web. Furthermore, they have the claim of providing public access to the entirety of the collection instead of presenting special virtual exhibitions with only a selection of digitized objects. Digital collections thus require a mode of access that shows the abundance of a collection. Therefore, museums have started to offer an “explore mode” (often as an addition to a “search mode”), which is meant to enable users to discover the collection without knowing details in advance (see Figure 1 for an example).

While there has been research on Web sites and social media activities of museums (Marty, 2008, 2007; Padilla-Meléndez and Águila-Obra, 2013; Pallud and Straub, 2014; Lin and Gregor, 2006), little has been said specifically about digital collections. They have become an important part of museums’ online activities but do not seem worth a detailed consideration of their functionality. Therefore, the aim of this study is to investigate digital collections of museums. Especially, we are interested in the explore mode as a way to discover the collections akin to a museum visit where one’s own paths through curated selections from the collection can be taken. With this focus, we pose the following questions: What functions support exploration in digital collections? How does this experience relate to interface elements and structure of digital collections?

In our study we consider eight well-known museums from all over the world. The museums have large physical collections and offer large digital collections as well. The second important selection criterion was that they provide a dedicated explore mode.

This paper introduces the method of reverse information architecture to analyze the Web site elements and to find out which structures may foster exploration in large collections. The overall information architecture of digital collections and the particular structure of three standard Web page types within them are the data source for our analysis. Before we explain the method of the analysis, the term of exploration will be introduced. The experience-oriented term of exploring is grasped in four interface functions supporting exploration. After describing the method and our results from the three different analyses, we conclude this paper with a discussion of the findings and future work.


More and more museum Web sites feature an exploratory mode of access into their digital collections as in this example from the Auckland Art Gallery
Figure 1: More and more museum Web sites feature an exploratory mode of access into their digital collections as in this example from the Auckland Art Gallery.




2. Towards a more exploratory mode in information retrieval

Graphical user interfaces (GUI) are the interfaces between computer processes and the user, showing interfaces on the desktop, programs, and the Web. In the development of GUIs for various information spaces, the search mode was and still often is omnipresent. The long history of search interfaces is transferred from physical working methods and as a result, the first GUIs were related to work and efficiency (Andersen and Pold, 2011; Johnson, 1999).

A similar approach came up in the digitization process of museum objects. Commonly, inventory catalogues were digitized and photographies of the objects were added. But this process was mainly carried out for internal use and with the intention to ease the management of entire collections. Search tools, as for instance the offer of different categories to search for, simplified the work for experts.

In the last decades, the dominant search mode especially in information retrieval was brought into question (Belkin, et al., 1982; Bates, 1989; Whitelaw, 2015, 2012) and as a result, more exploratory forms of access were proposed. Here, too, digital collections underwent a similar development. When previously internally used data sets are going online, it is advisable for museums to rethink the presentation and the access forms to their collections because of a greater variety of people who make use of these offers. Now the visitors not only include experts but also novices and intermediates (Marable, 2004). The explore mode was thus established in digital collections as an offer for users who want to discover the collection without having detailed knowledge of the structure, the collection, or museums in general. In the best way, the outcome would be a primarily autotelic experience which fosters a pleasant way to educate oneself. The term autotelic describes an activity that is “done not with the expectation of some future benefit, but simply because the doing itself is the reward” (Csikszentmihalyi, 1990).

It seems that museums have recognized the need for different approaches and that open exploration may function as a complement to focused searching. But what does exploration promise? Which concepts and studies have been introduced to make sense of exploration and how can these experiences be transferred into interface functions? With the next step we aim to develop a more graspable notion of exploration, especially in digital collections.

2.1. Exploration as an information experience

With the shift towards a more exploratory mode, the question of how to implement exploration into systems of information retrieval appeared (Marchionini, 2006). The following terms offer first answers to what exploration as an information experience could mean. They can be described as experience-based concepts that on a basis of playfulness, pleasure, and enjoyment strive towards new techniques in information retrieval. In the experience of exploration, the following ideas intertwine or can be enriched by additional aspects.


Exploring places for the first time or finding new facets of well-known items can be an exciting and a pleasurable experience. The concept of serendipity denotes this kind of joyful exploration. It is due to the fairy tale “The Three Princes of Serendip” making unintended discoveries “by accidents and sagacity” (Thudt, et al., 2012). Such coincidental discoveries can provoke curiosity and lead to a positive mood. Some personal characteristics increase these kinds of findings: people should be open-minded, bring observational skills, perseverance and the ability to connect apparently unrelated elements. In information retrieval, serendipity can be fostered through different access points that invite users to start browsing. The search process should not (only) depend on the user’s search queries, but “coincidental” starting points and paths should be offered.


In the year when the World Wide Web was invented, Bates (1989) proposed the notion of berrypicking to characterize an evolving information seeking process that involves multiple information sources, search queries, and navigation steps. In contrast to a reductionist concept of information retrieval during which a person asks one question and receives one answer, berrypicking denotes an iterative technique to find information in a gradual process. Different search options within a data set make the search process diverse and highly individual. An important information activity is the process of browsing, i.e., the exploration of an information space in order to understand its structure and contents (Chang and Rice, 1993). This means that access for information seekers should be given, but providing more than a classical search option by navigating between information sources in a step-by-step manner. This joint consideration of browsing and searching also underlines that a person may interact with information in varying levels of formality and directedness (Choo, et al., 2000).


Arguably berrypicking is connected with the experience of flow as the moment when people “are deeply involved in an enjoyable activity” for instance while “playing games, participating in sport, engaging in hobbies, or working” as well “while surfing the Net” (Pace, 2004). The flow experience is autotelic, i.e., the benefit lies in itself (Csikszentmihalyi, 1990). Within his study, Pace (2004) found Web-specific dimensions of flow: “enjoyment is a common factor in all flow experiences”, whereas using the Web “it appears to be linked to discovery — finding, learning or observing something for the first time.” Here, the terms of enjoyment and discovery are linked to each other (cf., Csikszentmihalyi, 1993). Furthermore, flow can engender the feeling of losing sense of time and space, which can be viewed as a form of telepresence (Pace, 2004).

While serendipity, berrypicking, and flow characterize highly interdependent qualities of exploration, we consider them to be sufficiently distinct to form the foundation of a design space for exploratory interfaces to digital collections.

2.2. Exploration as a function of collection interfaces

Graphical user interfaces offer a variety of possibilities to implement the mentioned experience-based qualities of exploration in the form of a digital collection. In the following, we define basic principles that integrate the experience of exploration into functional components of GUIs.

First of all, a variety of objects should be on view and secondly users have to get the ability to move through the collection. In addition to these principles we define two functions that enrich exploration: to contextualize the collection data and to participate in making sense of them. These four functional principles support an exploratory approach and users might learn for themselves by discovering the collection from an individual point of view, moving through them on own pathways, collecting information they are interested in, thereby being able to find hidden connections, and ideally also experience a sense of pleasure. While viewing and moving are essential aspects of any exploratory information practice, contextualization and participation are secondary activities in support of exploration that play a particular role in digital collections.


The first principle is to put information, and in our case study digitized museum objects, on display. Whitelaw introduces the term generous interfaces to refer “to be liberal in giving or sharing” by making “large abundant, ample” parts or whole collections accessible (Whitelaw, 2012). To implement broad access to a collection, Bates (2007) calls for “rich scenes, full of potential objects of interest” in online or digital collections. Generosity seems to be the precondition to show — and therefore put on display — the diversity of a collection. Interested users get more than one glimpse and an idea of what a collection holds. Such interfaces can also be described as “rich-prospect browsers” that provide comprehensive perspectives on entire collections (Ruecker, et al., 2011). In particular, overviews are often used as a starting point into the richness of a collection. The widely cited and reiterated information seeking mantra by Shneiderman (1996) advocates for ”overview first, zoom and filter, then details on demand.“ Meanwhile, Hornbæk and Hertzum (2011) criticize this mantra as not precise enough lacking a specific characterization of the functional and experiential qualities of overviews. In this sense, one can distinguish between the overview as an interface component, the activity of overviewing, and the experience of having a sense of overview.


Movement supports perspective changes, for instance between overviews and single detailed data elements. With the monadic exploration concept, movement is introduced as a key to understand interconnected information spaces by shifting between different perspectives of the collection (Dörk, et al., 2014). Movement encourages the finding of connections between entities and relations or differences can be found. Arguably a user can be seen as one monad moving through the richness of a collection by following relations (Dörk, et al., 2014).

This notion of information exploration as movement goes back to the concept of the information flaneur, who experiences an information space with a joyful and curious attitude (Dörk, et al., 2011). Commonly, a flaneur is guided by coincidences and encountering impressions instead of following an aim. Sauntering is characterized as a slow movement without having the approach of getting an encyclopedic completeness over a city, but rather a sense of the place (Keidel, 2005). With the transfer of the urban flaneur into the digital, Dörk, et al. (2011) introduce a “human-centered understanding of the search process.” The information flaneur moves through “informational landscapes” and “explores unfamiliar information using several tools and facets” (Dörk, et al., 2011). This definition suggests techniques to implement the function of movement into GUIs.

Moving around digital collections, users also need a sense of orientation to be able to move on to the next step or go back. The better users can orient themselves, the easier they can make findings, whether intended or not. Orienteering as a search technique was first brought up by Teevan, et al. (2004), showing that it consisted of small steps towards the needed information and that, concerning the users, it “decreased their cognitive load, allowed them to maintain a sense of location during their search, and gave them a better understanding of their search result.” It can thus be assumed that the practice of orienteering is likely being employed by visitors of digital collections.


The third function, contextualization, on one hand supports exploration but on the other hand is an addition to the exploration process. It is especially important and used in the presentation of museum objects. We understand contextualization as a translation of the German term “Vermittlung” that is commonly used when speaking about the presentation of museum objects in exhibitions. “Vermittlung” supports an understanding of giving visitors a chance to situate objects in time and space through additional information. In this sense, the single museum object cannot be presented on its own, but needs to be contextualized via textual annotation, multimedia content, or a guided tour so that visitors can grasp its significance within the collection. Furthermore, relations between single objects can be drawn to enrich them, for instance, by providing combined sets of objects. In the monadic exploration approach, discussed earlier, one principle emphasizes contextualization. “As each monad exerts its influence upon another, it is equally influenced by that other.” (Dörk, et al., 2014) Even if every entity stays for itself, it is influenced by others. In museums, connections are established by exhibition curators and collection experts. They classify, arrange, stage, and therefore contextualize single objects. This function is also important when introducing a digital collection to people not familiar with its significance and structure. Contextualizing and providing related objects may help to get an understanding of the whole collection and to overview the richness by getting hints and bridges from one object to another.


The fourth and last functional principle of exploration is participation. Since the term of the Web 2.0 arose in the early 2000s, the Web became more of a “social media” platform. It was new that users participate by sharing, commenting, or liking content. These new practices help users to get to “the right information using ‘community wisdom’ distilled from the actions of those who worked with this information earlier” (Brusilovsky, 2008). Dieberger, et al. (2000) introduced the term social navigation, referring to the experience of serendipity based on the traces left by others. On many Web sites “users are given the illusion of being the only person present” (Dieberger, et al., 2000). But supposedly, users are following traces that are “‘grown’ — or created dynamically — in a more organic, or bottom-up, fashion. In this way, social navigation is a closer reflection of what people actually do than it is a result of what designers think people should be doing” (Dieberger, et al., 2000). Participation changes an interface and “in this sense we can never fully grasp the interface as a form, but are compelled to pursue its various and ever-changing appearances” (Pold, 2005).

In the best way, this social transformation of a Web site resembles how a created space changes into a place when people are present (Harrison and Dourish, 1996). In this sense, Web sites can be seen as what Oldenburg (1997) describes as “third places”, which are open to the public, everyone is welcome, and a playful mood exists. Even if Oldenburg does not explicitly include exhibitions in his theory, we found striking similarities. During exhibition visits, people discover not only museum objects but the whole museum space with its paths, social traces, and the presence of other visitors. They gain a feeling of being a part of a community with the same interest. The design of digital collections could follow a similar approach by using social signals to aid orientation. It could put an end to fears of novice users not knowing the specific structure or functionality of a Web environment such as digital collections.



3. Method of the study

After outlining what exploration in information retrieval processes might mean, we are interested to examine to which degree the defined principles are already encountered in digital collections of museums. More specifically: How is the function of exploration manifested in the structure and interface elements of digital collections? To approach this question, we analyze the overall information architecture of the Web sites of existing digital collections and the particular elements on the respective Web pages. We distinguish between a Web site, meaning the entire structure of a digital collection, and a Web page as one particular page within this site.

The premise of this research is that the Web sites of digital collections are not mere marketing tools for the ‘real’ museum, but rather an exhibition form in its own right. This change of perspective calls for new methods of analyzing Web sites. Up to now, the content analysis of Web sites has not been transferred to digital collections and adjusted for their particular properties. In conventional content analyses, the navigation structure and the content of different Web sites are compared to each other (see Huizingh, 2000; Padilla-Meléndez and Águila-Obra, 2013), participants are asked to browse Web sites and Likert scales are evaluated (see Huizingh, 2000; Pallud and Straub, 2014) or experts assess the Web sites (see Lin and Gregor, 2006). In all analysis types, the content is not linked to the design and functionality of the Web sites, or they are merely rated from the user’s point of view.

With this research, we focus on visible elements and the structure of the digital collections and especially on the correlation between content, design, and function. Instead of examining the experiential qualities of exploration, the focus is on the actual information architecture of digital collections to analyze the potential for exploration. Based on the experiential and functional conceptualization of exploration, we devise a unique method to analyze the elements that are likely to foster it. We will thus assess the digital collections using above derived criteria to examine the structure and elements by relating them back to the conceptual layer. We call this method reverse information architecture to reflect that we worked reversely (concerning the composition of a Web site) to unfold the exploratory structures and design patterns of collection Web sites.

The most visited museums around the world offer a rich online collection and promise the ability to “explore” their collection, which attracted our attention. To establish a basis for fair comparison, we chose eight museums which have released a significant part of their collections online: The Auckland Art Gallery (NZ), Cooper Hewitt (USA), The Metropolitan Museum of Art (USA), Philadelphia Museum of Art (USA), Museo del Prado (Spain), Rijksmuseum (NL), Städel Museum (GER), and Tate (UK). All screenshots of the analyzed collections can be found here:


Three (main) page types of the Auckland Art Gallery
Figure 2: Three (main) page types of the Auckland Art Gallery.


3.1. Description of the analyses

During the first examination of the museum Web sites we recognized a common structure of their digital collections. Most featured three page types that can be consistently distinguished (see Figure 2). The A pages are the starting pages, the B pages consist of different intermediate pages (most commonly a page with search results, but also theme-arranged pages etc.), and the C pages are single-object pages. While A pages and B pages consist of lists or compilations, a C page displays one object from the collection in detail. These three page types need to be contrasted to analyze the structure of the digital collections and to compare similar pages to each other. Furthermore, pages that do not match the structure of the digital collection and lead on to other offers are defined as D pages; they are captured, but not examined further.

For all the analyses, we took screenshots of the different pages of the digital collections, using a resolution of 1,024x768 pixels. With a set width of 1,024 px, the length of the pages in the screenshots was not shortened and therefore differs. As the level of informational depth could differ between, for example, a masterpiece and a little known item, we always tried to find the most representative pages. To closely examine the content, design and functionality of these pages, we devised a three-part methodology.

3.1.1. Navigation analysis

If users shall be enabled to move through the collections, it is important to understand the navigation structure of the digital collections. Therefore, the links between different page types were recorded and examined in our first analysis. Is it, for example, possible to directly navigate from the home page (A) to a single object (C)? Or are users routed through intermediate pages (B)? Node-link diagrams represent the navigation structure and all possible navigation ways (see Figure 3).


The topological structures of two digital collections
Figure 3: The topological structures of two digital collections.


In a second step, the connections between nodes (pages) were weighted based on the relative screen space taken up by linked elements such as images or text. This information was assessed through a wireframing analysis, as discussed below. By identifying where the links lead to, the ratios of the elements to the complete Web page could be transferred into the navigation analysis, whereby the connecting lines in the graphic were displayed thicker or thinner according to the frequency of their links.

We considered screen space instead of the quantity of links. This is due to our impression that one might be more likely to click on an element that is larger than a smaller one. Also, links were only captured inside the digital collection, leaving global navigation elements or the interface of the Web browser out of consideration, to focus on the navigational structure inherent to digital collections.

3.1.2. Wireframe analysis

Besides the navigation structure across pages, we are interested in the functional patterns on individual pages. The wireframe analysis is focused on the Web pages by categorizing the individual page elements using a graphic program (Adobe Illustrator). In information architecture, wireframes serve as structural sketches of Web pages to devise and iterate their functional design. Our wireframe analysis can be seen as a form ‘reverse information architecture’ as the Web pages exist already and we are interested in the underlying functional structure. After annotating each relevant page element, the rectangles are assigned a distinct color to denote their functionality on the page (see Figure 4). Unlike conventional wireframes that are typically in grayscale, our wireframes use colors to distinguish between:

  • static (dark red) and linked images (orange),
  • static (dark blue) and linked text (light blue),
  • navigation elements like menus, headers or footers (yellow),
  • search box as a central element on all pages (green), and
  • interaction elements for social networks and multimedia (turquoise) and
  • multimedia elements (rose)


The opening page of the Auckland Art Gallery digital collection
Figure 4: The opening page of the Auckland Art Gallery’s digital collection with colored wireframes (left). Pure wireframes of the same page (right).


With this step of the analysis, we examine how much space is taken up by which types of elements on the Web pages. We compare the three most common Web page types of the digital collections: collection entry page (A), search results (B1), and object page (C). Using the manually created wireframes, the percentage values for each page element type are calculated by a simple script, which sets the size of the single elements in relation to the total size of the Web page. In a comparison graphic, the percentage of the space that the single elements fill on the different page types (A, B1, C) is illustrated comparing the digital collections of the eight museums to each other (see Figure 5). The height of each bar segment represents the relative space usage per element type, while the space on the Web pages that is not filled with elements is whitespace. The colors of the bars match the colors that were used for the wireframe analysis.


Evaluation graphic showing the percentage that elements take on the A-pages across all digital collections
Figure 5: Evaluation graphic showing the percentage that elements take on the A-pages across all digital collections.


3.1.3. Position analysis

The last analysis step is concerned with the vertical positioning of individual elements. In addition to the visual size distribution of the previous analysis, a prioritization of elements is also made in the form of the positioning on the page, in particular along the typically vertical scrolling direction. With this analysis we examine how the three page types (A, B1, and C) differ with respect to the vertical distribution of visual vs. textual and linked vs. not-linked elements. In Figure 6, a visualization shows this distribution aggregated across all eight institutions for each page type. The y-axis thereby shows the length of the Web pages and the position of the single elements on these. We set the maximum length to 4,000 px, because only some pages showed a longer length but were then not comparable to the other pages anymore and tended to be repetitive after a certain length. The x-axis adds up the relative space usage of the respective element types on all Web pages at this vertical position. With changing screen sizes there is no reliable statistic for the page fold, i.e., the position that requires a vertical scrolling operation to go beyond the top part of the page. However, a fold of 720 px of a typical laptop display was added in the visualization for a better sense of orientation.


The vertical distribution of text and image elements across the three page types of the eight museum Web sites under study
Figure 6: The vertical distribution of text and image elements across the three page types of the eight museum Web sites under study.




4. Findings

In the following we report the main findings that can be derived from the analyses of the reverse information architecture. The complete evaluation graphics can be found in the Appendix.

4.1. Navigation analysis

To exemplify what can be implied from the graphics, we start with a rather classical example of the navigation structure, describing our findings for the Auckland Art Gallery (see Figure 3). From the start page (A) it is possible to perform a search request, thus being forwarded to the search result page (B1). Also, single elements can be clicked on so that one is forwarded to the corresponding single object pages (C). The B2 page stands for a “curated” offer, where artists of the collection are presented on a thematic page. Both the B pages and the A page forward to C pages with most of their links. It is never possible to go back to the A page directly. Having reached a single object page, one is most likely navigated to other C pages, but generally there are few possibilities to move back to A or B pages. There is a limited amount of opportunities to explore the collection on one’s own paths, offering only the search-result list, one thematic page, and single-object pages. Having reached the latter, one has to stay in this detailed mode and there is a lack of navigational support to return to an overview via other pages.

A different type of navigation structure we observe is a circular movement in the navigational topology. It can be seen in the example of the Rijksmuseum (see Figure 7). At one glance, it is obvious how diverse the navigation structure is. The navigation diagram contains many paths and especially the B pages are interlinked to each other. The C page can be reached from most of the B pages; still, most of the movement is made circularly on the B pages. It is also noteworthy that, here as before, there are very few opportunities to return to the A page.

All of the eight analyzed digital collections can be classified within these two described typical structures: one-way or circular. For the Auckland Art Gallery, Metropolitan Museum of Art, Philadelphia Museum of Art, and Museo del Prado, we observe a one-way direction. In contrast, the Cooper Hewitt, Rijksmuseum and Stdel Museum exhibit a rather circular structure. The Tate’s navigation structure shows characteristics of both and can thus be seen as a hybrid. Here, the structure circulates around the B11 page (glossary) and the digital collection thus does not concentrate on one function or direction, but offers different paths and options.

4.2. Wireframe analysis

Comparing the A pages (see Figure 8), it is first of all striking that approximately two thirds of the space on most of the institutions’ start pages is used by images (red/static and orange/linked). Only the Cooper Hewitt and the Prado do not use as much images on this page; in comparison, they use much more static text (blue) and navigation elements (yellow). Search boxes (green) are offered almost everywhere, but do not take as much space. There are not many additional interaction possibilities (turquoise) on the A pages, except for the Rijksmuseum and the Städel Museum who offer few, for instance a heart icon to start a personal collection with favorites. The relative amount of whitespace differs extremely between the collections’ entry pages, ranging from 30 percent (Cooper Hewitt) to 70 percent (Städel Museum).

The B1 pages show slight changes in the distributions. Similar to the A pages, the linked images dominate, followed by navigation elements and static text. Noticeable is the Museo del Prado with navigation elements that take up two thirds of the overall space. This is due to a large footer that is used on this Web site which consists of more elements than usual (multimedia, links to exhibitions). The search box is still offered on most pages. All of the eight institutions offer few interaction possibilities. The whitespace is more balanced between the digital collections; it ranges between 30 percent and 50 percent.

On the C pages (single-object views), the whitespace takes an average of 50 percent. Instead of images, text and navigation elements dominate here. The text and images that are offered are mostly static, so that being forwarded to other pages is less likely. The interaction possibilities increase slightly in comparison to the B1 pages; here, the Rijksmuseum, Philadelphia Museum of Art, and Cooper Hewitt offer the most of them. The Rijksmuseum and Städel Museum need to be considered separately on this page type. On the Rijksmuseum’s C page, the screen space is filled with over 50 percent by mainly static images, while there are no navigation elements at all. However, the object image is zoomable providing access to high-resolution views on the object. The Städel Museum also lacks navigation elements, but instead of images it rather uses static text and 10 percent for multimedia content.

In summary, it can be said that moving from A to C pages, the elements become more static, i.e., less linked, across all the digital collections that we considered. Also, the amount of whitespace increases on the C pages to half of the screenspace. Generally, there is little space devoted to interaction possibilities, with the exception of a small increase on the C pages; also, multimedia content is hardly ever provided. Search boxes are offered on A and B pages, but almost disappear on the C pages.

4.3. Position analysis

On the A pages, it is first of all striking that many images appear above the fold (see Figure 9). Only after the fold, text elements increase. While images over all dominate on the A pages, the text elements seem to stay on a constant level after this high peak. The relation of linked and static elements furthermore shows that above the fold, most elements are not linked. Seeing that these elements are mostly images, it can be concluded that static images dominate above the fold. These elements can for instance be introductory images. After the fold, linked elements increase and prevail, assuming twice as much space as static elements.

The B1 page typically has a structure of thumbnail lists: images dominate, while text is put under the images and thus appears on an alternating basis with the images. This can especially be seen at the lower end of the graphic, where only two pages remain (Cooper Hewitt and MET) because of their extraordinary length. Again, the high peak of the linked elements on the B1 page is after the fold and linked elements now dominate more than on the A pages. Above the fold, linked and static elements seem balanced on the B1 pages, but static ones appear earlier at the start of the pages.

On the C pages featuring single objects, images are shown at the beginning. After the images (and the fold), text elements increase, showing mostly object information. Image elements then decrease immediately. One can conclude that the C pages are rather information-oriented because not only text dominates the pages after the fold, but also static elements. Only long after the fold, some linked elements are offered on the pages.

Comparing the position of the text and image elements from A to C pages, it is remarkable that images take most of the space above the fold. Mostly, the elements above the fold are also mostly static (except on B1 pages), with a maximum on the C pages.



5. Discussion

At the end of this paper, we want to draw conclusions from the overall impressions we got of the digital collections when analyzing them. We return to the four functions of exploration that were defined at the beginning of this paper, examining to which extent they can be found in the analyzed digital collections.

The first principle to implement exploration into interfaces was to view the richness of a collection. Our analysis has shown that images usually dominate the pages and therefore a view into a collection is certainly provided. Most evidently, this can be seen on the B1 pages (search result lists with thumbnails) that are offered by all of the institutions. Although search result lists constitute a rather classical design structure, users can at any rate choose from a variety of images here. But comparing the number of objects a collection offers to the images on these pages, one can say that the claim of making the richness accessible should be expanded to more unusual object views than thumbnail lists. The concept of generosity can mainly be found in the provided variety of images but not (yet) in overviews for instance by using information visualizations. Furthermore, perspective changes from overviews to details are typical from A to C pages, in a relatively abrupt manner, but returning to an overview after the immersion with one particular object is hardly possible due to the one-way direction from A to C.

By combining the results of the three analyses, a noticeable insight appears, referring to the second interface principle of movement. In all three analyses, the feeling of a dead end occurs and even in the circular type of the navigation structure, the C pages are almost isolated. With this result in mind, the question arises whether the trajectories of users in digital collections are planned or even curated because one or two ways mostly dominate in the navigation graphic. This implies that a sauntering movement is possible, however, it is at the same time considerable constrained. To find own ways through a collection, users would need a portion of coincidence, for example, by being able to make unintended findings via a “random object” button or by getting to the next object in an exceptional way, like the Städel Museum offers via keywords that link the objects. We also made the observation that many digital collections offer a well-considered start page (A) but then lose the inviting and exploratory qualities on the other pages. The navigation analysis suggests that users might only visit these pages once at the beginning and when they immerse deeper into the collection, there are few or no invitations to guide them back or further along to other objects. At the moment, A pages seem to be more important for the design concept (because the users are introduced to the collection there) than having a role in the overall structure and functionality of the whole digital collection.

Coming to the implementation of contextualization, it seems to be most important on C pages, where users can learn more about individual artworks and artifacts. The movements shown by the navigation analysis suggest that relations from a single object to other objects or related content is given, but since the wireframe analysis showed a decrease in linked material towards the C pages, only limited forwarding to related works or themes seems to be done. One could say that “Vermittlung” is only implemented in a basic form in digital collections, offering textual information to one object mostly.

Assuming that contextualizing elements on the C pages are either text elements or links to other objects, the position analysis shows that they are also mainly placed after the fold. Users have to move (scroll) down the page to get more information and pursue related themes and objects. Even so, it should be emphasized that some digital collections offer special features for contextualization. The MET provides a “Timeline of Art History” that locates single objects in the larger context of art history. The Philadelphia Museum of Art provides a unique view by introducing collectors and assigning them their objects of the collection as well as showing past exhibitions within the collectors’ collections. A last feature supporting contextualization are D pages, where special offers like articles, films, or exhibition views are provided. This page category can be seen within the Tate’s digital collection, where many links from the A page go to a variety of D pages. They offer videos, articles, a kids’ space as well as options to participate, for instance by creating a portfolio.

This leads to the last principle of exploration, suggesting that users can more personally and socially engage with the digital collection and participate by liking or sharing interesting findings or even creating personal collections. The wireframe analysis has shown that few such interaction elements are offered on the pages. One exception is the Rijksmuseum, which puts the “Rijksstudio” and its “user sets” on focus of the whole Web site concept, after a restructuring of their Web site in June 2016. Once registered, a Web site visitor can create their own sets by collecting single objects or details of them, add personal descriptions, like and share them, and browse through the sets of other people. Even in the search option, other user sets are searchable. Furthermore, Städel Museum, Tate, and Museo del Prado allow for the creation of personalized collections, but they appear more separated in the navigation structure and are furthermore not visible for unregistered users. Even though these sharing and customization options can support serendipitous findings, users may still have the feeling of browsing alone through the collection. The opportunities for more social navigation features could be expanded.

Last but not least we discuss the element of the search box. Its omnipresence attracted our attention during the research, but because it neither occupies much space in the wireframe analysis nor changes positions, its central role cannot be shown very well in our analyses. Even if a distinction between a search mode and an explore mode is usually made in digital collections, the explore mode typically has some form of search option. The search box is always shown above the fold and often positioned directly under the header. A second box often is part of the header to start a search across the whole museum’s Web site and not only within the digital collection. It is interesting to note how dominant this element still is and in combination with exploratory functions, it takes its own place within the experience of exploration, helping users by offering a well-known tool but leaving space for one’s own search terms.



6. Conclusion

We synthesized the experience of exploration with the concepts of serendipity, berrypicking, and flow and defined four principles that function as a support for exploration in digital collections — view, movement, contextualization, and participation. With the introduced method of reverse information architecture we examined the site structure and page elements of eight digital collections, to investigate how the claim of offering an exploratory experience is manifested in the actual collection interfaces of museums.

Since there are a lot of possibilities how content, design, and functionality can operate together, it was clear from the beginning that the outcome of our analysis could not be one “right” way of fostering exploration in digital collections. Instead, we have outlined some trends and recurring structures in the explore modes of museum collections. Especially the impression of following a one-way street or even an impasse was validated in the analyses. Considerable design work, it seems, is devoted to the start pages of collections, where users are rarely guided back to. Images of the digitized objects are often at the center of the digital collections, especially on the single-object pages, but the room for contextualization and participation that plays a big role in the physical spaces of museums could be expanded significantly in their digital counterparts. More generally, we often missed a well-considered concept concerning the navigation and functionality of the collection Web sites.

While we have found some outstanding exemplars of rich and inviting presentation of cultural artifacts within the digital collections, the overall impression is that a holistic concept is still missing. The experience of exploring is only partly encouraged, but at many stages could be expanded further. With the results, we hope to forward the discussion around digital strategies of museums and to inspire museum professionals working on digital collections.

In further steps, our reverse method could be extended by an experience-based analysis, for instance by an eye-tracking analysis of respondents visiting the digital collections. This evaluation would add a personal perspective on the Web sites by examining which functions are used more often and investigating whether they have a joyful and inspiring experience. It would provide evidence for the claim of exploration as a counterpoint to work-based search processes. The Web sites should also be reviewed by using a variety of devices including smartphones and tablets, and considering a diversity of users including children and citizens with disabilities.

Finally, the introduced method of reverse information architecture could be opened up as a method to analyze the cultural artifact of interfaces in general. The specialty of the method is that it works with the actual interfaces of digital collections and relates them to the design concepts that they are based on. Our hope is that this research can provide a useful foundation for understanding the structure of a digital information space to devise exploratory modes of interacting with them. End of article


About the authors

Since 2017, Sarah Kreiseler is part of the research program PriMus (Ph.D. in museums), a cooperation between Leuphana University of Lüneburg and in her case the Museum für Kunst und Gewerbe Hamburg. She holds a Master’s degree in European Media Studies (University of Potsdam). She is especially interested in exploring and analyzing interfaces as a new approach to study current expressions of culture and society.
E-mail: s [dot] kreiseler [at] gmx [dot] net

Viktoria Brüggemann holds a Bachelor’s degree in cultural work and arts management and is currently enrolled in the Master’s program “European Cultural History” at the European University Viadrina Frankfurt/Oder. Her emphasis is mainly on museum’s curatorial work and outreach, in the digital as well as in the physical realm.
E-mail: viktoria [dot] brueggemann [at] gmx [dot] de

Marian Dörk is a Research Professor for Information Visualization at the Urban Futures Institute for Applied Research of the University of Applied Sciences Potsdam. In his research he is particularly interested in the potential of visual interfaces to support new forms of data exploration. Since January 2015 he has been co-directing the Urban Complexity Lab, a transdisciplinary research space at the intersection between information visualization and urban transformation.
E-mail: doerk [at] fh-potsdam [dot] de



C.U. Andersen and S.B. Pold, 2011. “Introduction: Interface Criticism,” In: C.U. Andersen and S.B. Pold (editors). Interface Criticism: Aesthetics beyond buttons. Aarhus, Denmark: Aarhus University Press: pp. 7–18.

M.J. Bates, 2007. “What is browsing — really? A model drawing from behavioural science research,” Information Research, volume 12, number 4, at, accessed 6 June 2016.

M.J. Bates, 1989. “The design of browsing and berrypicking techniques for the online search interface,” Online Review, volume 13, number 5, pp. 407–424.
doi:, accessed 22 March 2017.

N.J. Belkin, R.N. Oddy, and H.M. Brooks, 1982. “ASK for information retrieval: Part I. Background and theory,” Journal of Documentation, volume 38, number 2, pp. 61–71.
doi:, accessed 22 March 2017.

P. Brusilovsky, 2008. “Social information access: The other side of the social Web,” In: V. Geffert, J. Karhumäki, A. Bertoni, B. Preneel, P. Návrat, and M. Bieliková (editors). SOFSEM 2008: Theory and Practice of Computer Science. Lecture Notes in Computer Science, volume 4910. Berlin: Springer-Verlag, pp. 5–22.
doi:, accessed 22 March 2017.

S.-J. Chang and R. Rice, 1993. “Browsing: A multidimensional framework,” Annual Review of Information Science and Technology, volume 28, pp. 231–276.

C.W. Choo, B. Detlor, and D. Turnbull, 2000. “Information seeking on the Web: An integrated model of browsing and searching,” First Monday, volume 5, number 2, at, accessed 22 March 2017.
doi:, accessed 22 March 2017.

M. Csikszentmihalyi, 1993. The evolving self: A psychology for the third millennium. New York: HarperCollins.

M. Csikszentmihalyi, 1990. Flow: The psychology of optimal experience. New York: HarperPerennial.

A. Dieberger, P. Dourish, K. Höök, P. Resnick, and A. Wexelblat, 2000. “Social navigation: Techniques for building more usable systems,” Interactions, volume 7, number 6, pp. 36–45.
doi:, accessed 22 March 2017.

M. Dörk, R. Comber, and M. Dade-Robertson, 2014. “Monadic exploration: Seeing the whole through its parts,” CHI ’14: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1,535–1,544.
doi:, accessed 22 March 2017.

M. Dörk, S. Carpendale, and C. Williamson, 2011. “The information flaneur: A fresh look at information seeking,” CHI ’11: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1,215–1,224.
doi:, accessed 22 March 2017.

S. Harrison and P. Dourish, 1996. “Re-place-ing space: The roles of space and place in collaborative systems,” CSCW ’96: Proceedings of the ACM Conference on Computer-Supported Cooperative Work, pp. 67–76.

K. Hornbæk and M. Hertzum, 2011. “The notion of overview in information visualization,” International Journal of Human-Computer Studies, volume 69, numbers 7–8, pp. 509–525.
doi:, accessed 22 March 2017.

E.K. Huizingh, 2000. “The content and design of Web sites: An empirical study,” Information & Management, volume 37, number 3, pp. 123–134.
doi:, accessed 22 March 2017.

International Council of Museums (ICOM), 2007. “Museum definition,” at, accessed 8 September 2016.

S. Johnson, 1999. Interface culture: How new technology transforms the way we create and communicate. New York: HarperEdge.

M. Keidel, 2005. Die Wiederkehr der Flaneure: Literarische Flanerie und flanierendes Denken zwischen Wahrnehmung und Reflexion. Würzburg: Königshausen und Neumann.

A.C.H. Lin and S. Gregor, 2006. “Designing websites for learning and enjoyment: A study of museum experiences,” International Review of Research in Open and Distributed Learning, volume 7, number 3, at, accessed 22 March 2017.

B. Marable, 2004. “Experience, learning, and research: Coordinating the multiple roles of online exhibitions,” Museums and the Web 2004 Conference, at, accessed 10 July 2016.

G. Marchionini, 2006. “Exploratory search: From finding to understanding,” Communications of the ACM, volume 49, number 4, pp. 41–46.
doi:, accessed 22 March 2017.

P.F. Marty, 2008. “Museum websites and museum visitors: Digital museum resources and their use,” Museum Management and Curatorship, volume 23, number 1, pp. 81–99.
doi:, accessed 22 March 2017.

P.F. Marty, 2007. “Museum websites and museum visitors: Before and after the museum visit,” Museum Management and Curatorship, volume 22, number 4, pp. 337–360.
doi:, accessed 22 March 2017.

R. Oldenburg, 1997. The great good place: Cafés, coffee shops, community centers, beauty parlors, general stores, bars, hangouts, and how they get you through the day. New York: Marlowe.

S. Pace, 2004. “A grounded theory of the flow experiences of Web users,” International Journal of Human-Computer Studies, volume 60, number 3, pp. 327–363.
doi:, accessed 22 March 2017.

A. Padilla-Meléndez and A.R. del Águila-Obra, 2013. “Web and social media usage by museums: Online value creation,” International Journal of Information Management, volume 33, number 5, pp. 892–898.
doi:, accessed 22 March 2017.

J. Pallud and D.W. Straub, 2014. “Effective website design for experience-influenced environments: The case of high culture museums,” Information & Management, volume 51, number 3, pp. 359–373.
doi:, accessed 22 March 2017.

S.B. Pold, 2005. “Interface realisms: The interface as aesthetic form,” Postmodern Culture, volume 15, number 2.
doi:, accessed 22 March 2017.

S. Ruecker, M. Radzikowska, and S. Sinclair, 2011. Visual interface design for digital cultural heritage: A guide to rich-prospect browsing. Farnham, Surrey, England: Ashgate.

B. Shneiderman, 1996. “The eyes have it: A task by data type taxonomy for information visualizations,” VL ’96: Proceedings of the 1996 IEEE Symposium on Visual Languages, pp. 336–343, at, accessed 22 March 2017.

J. Teevan, C. Alvarado, M. Ackerman, and D. Karger, 2004. “The perfect search engine is not enough: A study of orienteering behavior in directed search,” CHI ’04: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 415–422.
doi:, accessed 22 March 2017.

A. Thudt, U. Hinrichs, and S. Carpendale, 2012. “The bohemian bookshelf: supporting serendipitous book discoveries through information visualization,” CHI ’12: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1,461–1,>doi:, accessed 22 March 2017.

M. Whitelaw, 2015. “Generous interfaces for digital cultural collections,” digital humanities quarterly, volume 9, number 1, at, accessed 29 July 2016.

M. Whitelaw, 2012. “Towards generous interfaces for archival collections,” paper presented at the ICA Congress (Brisbane), at, accessed 29 July 2016.



All material used for the analysis is available on the appendix Web page of the paper:


Evaluation graphic of the navigation analysis
Figure 7: Evaluation graphic of the navigation analysis.



Evaluation graphic of the wireframe analysis
Figure 8: Evaluation graphic of the wireframe analysis.



Evaluation graphic of the position analysis
Figure 9: Evaluation graphic of the position analysis.



Editorial history

Received 22 September 2016; revised 21 March 2017; accepted 24 March 2017.

Creative Commons License
This paper is licensed under a Creative Commons Attribution 4.0 International License.

Tracing exploratory modes in digital collections of museum Web sites using reverse information architecture
by Sarah Kreiseler, Viktoria Brüggemann, and Marian Dörk.
First Monday, Volume 22, Number 4 - 3 April 2017

A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2017. ISSN 1396-0466.