DHQ: Digital Humanities Quarterly
Preview
2017
Volume 11 Number 4
Preview  |  XML |  Discuss ( Comments )

Supporting the Exploration of Online Cultural Heritage Collections: The Case of the Dutch Folktale Database

Iwe Everhardus Christiaan Muiser <e_dot_c_dot_muiser_at_gmail_dot_com>, University of Twente, Enschede / Meertens Institute, Amsterdam
Mariët Theune <m_dot_theune_at_utwente_dot_nl>, University of Twente, Enschede
Djoerd Hiemstra <d_dot_hiemstra_at_utwente_dot_nl>, University of Twente, Enschede
Theo Meder <theo_dot_meder_at_meertens_dot_knaw_dot_nl>, Meertens Institute, Amsterdam / University of Groningen, Groningen

Abstract

This paper demonstrates the use of a user-centred design approach for the development of generous interfaces/rich prospect browsers for an online cultural heritage collection, determining its primary user groups and designing different browsing tools to cater to their specific needs. We set out to solve a set of problems faced by many online cultural heritage collections. These problems are lack of accessibility, limited functionalities to explore the collection through browsing, and risk of less known content being overlooked. The object of our study is the Dutch Folktale Database, an online collection of tens of thousands of folktales from the Netherlands. Although this collection was designed as a research commodity for folktale experts, its primary user group consists of casual users from the general public. We present the new interfaces we developed to facilitate browsing and exploration of the collection by both folktale experts and casual users. We focus on the user-centred design approach we adopted to develop interfaces that would fit the users' needs and preferences.

Introduction

Over the past few decades, an ever-growing amount of cultural heritage collections has been digitised and made available online. In principle, the materials in these collections are now available to larger audiences than ever before. In practice, the accessibility of many online collections is limited by the absence of easy, user-friendly ways to explore the collection and get an overview of their contents [Ruecker et al. 2011] [Whitelaw 2015]. This may result in users accessing only materials they already know or expect to be in the collection, while unknown items remain hidden. Finally, accessibility for the general public will be reduced if a collection is presented mainly from an expert perspective [Trant 2006].
Whitelaw proposes to improve accessibility of online collections by designing so-called generous interfaces, "rich, browsable interfaces that reveal the scale and complexity of digital heritage collections" by allowing the whole content of a database to be accessed from a single starting point [Whitelaw 2015]. This is a similar solution to the rich-prospect browsers proposed by Ruecker et al., which show "a visual representation of every item in a given collection, combined with tools for manipulating the display"  [Ruecker et al. 2011].
In this paper we demonstrate the use of a user-centred design approach for the development of generous interfaces/rich-prospect browsers for an online cultural heritage collection, by determining its primary user groups and designing different browsing tools to cater to their specific needs [Stephenson 1999]. In doing so, we address the following problems faced by many online cultural heritage collections:
  • Lack of accessibility, in particular for casual users with no existing knowledge of the collection.
  • Limited functionalities to explore the collection through browsing.
By improving accessibility and offering rich browsing possibilities, we also intend to reduce the risk of less-known content being overlooked.
The object of our study is the Dutch Folktale Database, an online collection of tens of thousands of folktales from the Netherlands (www.verhalenbank.nl). Although this collection was designed as a research commodity for folktale experts, in practice its primary user group consists of casual users: members of the general public who access the collection infrequently and mostly for personal use.
Below, we start off by describing the content and use of the Dutch Folktale Database. Then we present the new tools and interfaces we developed to facilitate browsing and exploration of the collection by both folktale experts and casual users. We focus on the user-centred design approach we adopted to develop interfaces that would fit the casual users' needs and preferences.

About the Dutch Folktale Database

Folktales have been an important aspect of culture for as long as we can trace back in history. Jokes, legends, fairy tales, etcetera have been conceived, told and retold over time. For considerably less time, they have also been documented in manuscripts, documents, books, and catalogues. In 1994 a database containing folktales collected in The Netherlands was established at the Meertens Institute in Amsterdam and was baptised the Dutch Folktale Database. It has been extended with new folktales and subcollections ever since. The first version was a local database, only available to researchers within the institute, but in 2004 the database became available online. At the moment of writing the Dutch Folktale Database contains roughly 44.200 folktales from all over the Netherlands, in several languages (mostly Dutch and Frisian) as well as numerous dialects and historical versions of Dutch. They have been collected by 1477 collectors and cover almost a thousand years, with the oldest folktales dating back to the 12th Century. The folktales have been annotated with an abundant amount of metadata related to the story content such as summary, keywords, motifs and folktale type, as well as metadata related to the storytelling situation such as geolocation, narrator, collector, corpus, and source. These rich annotations make the database very useful as a resource for research, on top of its evident archival function.
The Dutch Folktale Database is a good illustration of the accessibility problems of online cultural heritage collections sketched above. Its original search options demanded a fair amount of knowledge about the contents of the collection, and were not suitable for casual users. Stories of interest could remain hidden if a specific search was not done correctly. Even though browsing is a preferred information seeking method for people working with folktales [La Barre and Tilley 2012], possibilities for browsing the collection were very limited. At the same time, the size, variability and growth of content made it difficult for its users to have an overview of the collection.
In the project FACT (Folktales As Classifiable Texts), which forms the context for the work presented in this paper, methods for automatic folktale annotation have been developed that are expected to further speed up the growth of the database [Nguyen et al. 2012] [Nguyen et al. 2013] [Trieschnigg et al. 2012] [Trieschnigg et al. 2013b]. In addition, the metadata of the folktales already present in the database have been cleaned up and standardised [Muiser et al. 2012], improving the possibilities to search, categorise and compare them. Some faulty or missing data however, cannot be corrected or retrieved. Folktales may have been narrated by an anonymous person, or annotated without a date or location. This can make them difficult to access, as they may be pushed to the back, or even left out, of search results. By providing users with the opportunity to browse and explore the collection from many different angles, we hope to enable more "accidental" discovery of also these hidden gems.

Use of the Dutch Folktale Database

One of the basic principles of user-centred design is to start by understanding the users [Gould and Lewis 1985] [Stephenson 1999] [Preece et al. 2015]. Therefore, before deciding on the direction to take for the development of new interfaces for the Dutch Folktale Database, an analysis of how it is being used and who is using it was necessary.
As a first step, we conducted a short poll among the website visitors of the Dutch Folktale Database, asking them for which purpose they wanted to use the information from the database. The seven answer options can be seen in Table 1. The poll was similar to a survey on an earlier version of the website [Trieschnigg et al. 2013a], except that we added one answer option ("For a school assignment") because this purpose was often entered in the "other" field of the earlier survey. In our poll, users could choose only one answer whereas in the survey of Trieschnigg et al., more than one answer could be selected.
The poll ran for around 3 weeks in the spring of 2015 and was filled in by 226 users of the Dutch Folktale Database. In Table 1 we can see that most people visited the website for personal use (73.5%). This number has increased since the earlier poll, of which the results are also shown in Table 1 for comparison. A possible explanation is that since the beginning of 2013, the Dutch Folktale Database uses a new content management system called Omeka (http://omeka.org/). Omeka makes the website rank higher in the results of search engines, thereby likely attracting more casual visitors.
Use of the Dutch Folktale Database 2015
(n=226, single choice per user)
2012/2013 [Trieschnigg et al. 2013a]
(n=88, multiple choices per user)
Personal use 166 (73.5%) 56 (63.3%)
School assignment 16 (7.1%) -
Storytelling 14 (6.2%) 26 (29.5%)
Education 9 (4.0%) 12 (13.6%)
Other 9 (4.0%) 11 (12.5%)
Scholarly use 8 (3.5%) 15 (17.0%)
Journalism 4 (1.8%) 3 (3.4%)
Table 1. 
Responses of visitors of the Dutch Folktale Database website to the question "What do you want to use the information from the Folktale Database for?"
To get more information on who the users are, where they come from and how they use the Dutch Folktale Database, we used Google Analytics (https://analytics.google.com/) to collect visitor data between Feb-June 2015. Google Analytics is a platform that provides a large amount of statistics about websites. It allows for a detailed analysis of the behaviour of the users, including click-through paths.
We found that during the investigated period, the Dutch Folktale Database received about 350-400 individual users per day, in about 400-450 sessions. They were quite evenly distributed over five age groups (18-24, 25-34, 35-44, 45-54, 55-64, 65+) with percentages between 15% and 20%. Almost nine out of ten visitors (88%) had never visited the website before. The remaining 12% were "tagged" by Google Analytics as a returning visitor, meaning a visitor who started a new session after having closed all pages of the website. Whether users return can give us an indication of whether the user experience on a site is adequate. The amount of time a user spends on the site is also an indicator [Khoo et al. 2008]. We found that the average session duration was only 1:20 minutes, which seems rather short and suggests that there was insufficient incentive for users to remain on the site and explore further. Users visited an average of 2.03 pages per session. This is an important metric, as it indicates how willing users were to explore more folktales. The low number may have been due to the lack of an easy way for users to explore the collection.
About 80% of the visitors arrived at the Dutch Folktale Database through a general web search engine (e.g., Google). This would usually lead them directly to the full text of a folktale. Therefore, one explanation for the short time users spent on the site may be that many users immediately found what they were looking for. However, we would like to tempt these users to explore other folktales as well.
Only 11% of all visitors came to the website by entering the URL directly into their browsers; the remaining 9% arrived through portals such as pagina.nl and the main website of the Meertens Institute. The group who entered the URL directly is likely to consist of experts on folktale research or other professional users, as those familiar with the website are likely to know its URL and use this as their entry point. These results are in line with the findings from our 2015 poll, in which only 15.5% of the visitors indicated that they used the database for professional purposes (storytelling, education, scholarly use, journalism). Of these user groups, the storytellers and scholars (9.8%) are most likely to make relatively frequent use of the Folktale Database.
Summarizing, the results from our survey and Google Analytics show that although one of the main purposes of the Dutch Folktale Database is to support humanities research, the largest user group consists of people who stumble upon the database through search engines, and for whom this is their first visit. These casual users are an important target group of the Dutch Folktale Database, as we aim to optimise its social impact and its utility for general audiences. Unfortunately, the analysis revealed that on average the users did not stay long on the website and did not visit many pages. This may have been caused by the original lack of browsing facilities for the Dutch Folktale Database, which we discuss next.

Accessing the Dutch Folktale Database

At the time of the analyses reported in the previous section, the possibilities for searching and browsing the Dutch Folktale Database were limited to using a simple Google-style search box located in the top right corner of each page, or browsing through an alphabetical list of lexicon entries linked to stories. After entering a search word or phrase in the search box, the user was provided with a list of search results. There were no options for filtering the results; the list could only be changed by changing the search phrase or by using the "advanced search" option, which confronts the user with a large list of search fields corresponding to folktale metadata, see Figure 1. However, as we will discuss in more detail below, most of these advanced search fields are only meaningful or relevant for folktale experts, not for casual users. Moreover, the desired values needed to be typed manually into these search fields, meaning the users needed to know their possible values to successfully use them in a search.
Figure 1. 
A part of the advanced search interface as it looked in 2015. In its current version, a few fields such as Taal (language) and Bron (source type) offer controlled choice options and some other fields give suggestions when a search query is entered.
To get more insight into what the users were looking for in the collection, Trieschnigg et al. analysed the search queries made between April 2010 and Jan 2012 [Trieschnigg et al. 2013a]. They found that among the most frequent searches in both the simple and the advanced search interfaces were queries for particular folktale subgenres, e.g., traditional legend or urban legend. The fifth most frequent query in the simple search box was the empty query. This indicates that many visitors do not have a specific information need, but simply want to browse the collection. Unfortunately, these attempts at browsing by means of an empty query would not yield any results at all.
These findings by Trieschnigg et al. suggest that both casual and expert users would be served well by offering them the possibility to browse through collections of items of a certain type in addition to searching directly for specific items. However, experts and casual users may prefer different browsing methods. For example, expert users have been found to like more complex graph-based search approaches, whereas casual users prefer simple search methods such as a controlled natural language approach, which experts find too restrictive [Elbedweihy et al. 2012].
An approach that seems suitable for meeting the information needs and search preferences of both casual and advanced users is faceted search [English 2002]. This approach uses the metadata of a collection to visualise the search space and allows the users to refine, change or expand their searches based on selections of these metadata. We applied different variations of this approach in a number of visualisation tools and browsing interfaces for both scholarly and casual users, as described in the following sections.

Visualisation Tools for Folktale Researchers

In this section we present two visualisation tools we developed to support experts (i.e., folktale researchers) in their exploration of the Dutch Folktale Database and to help them discover new connections between folktales, based on the metadata in the collection. The tools we developed provide dynamic visualisations such as plotting search results on a map, timeline or network graph. Such visualisations could lead to new insights into existing data and could be used to verify assumptions that have been based on qualitative research. By making these tools available to folktale researchers we aim to inspire alternative ways of investigation that do not rely on prior familiarity with the database, but support overview and exploration instead.
Two dimensions of folktales, geographic location and date, were obvious choices for visualisation. These dimensions are of interest to researchers because of their relationship with the historical and geographical environment in which the tales came into existence [Abello et al. 2012]. Both location and date can be used to provide a bird’s eye view of the collection, or a subset of the collection. Another dimension that is of interest to folktale researchers is the connection between folktales [Tehrani 2013]. These connections are loosely defined, because they can be based on many different aspects of the folktales and their metadata.
All visualisation tools we describe below have been developed using existing web technologies such as the Solr search engine (http://lucene.apache.org/solr/), the web scripting languages PHP(http://www.php.net/) and javascript (https://developer.mozilla.org/en-US/docs/Web/JavaScript) and D3.js (https://d3js.org/), a JavaScript library for manipulating documents based on data that offers powerful visualization components. The use of these technologies makes it possible for the tools to be accessed via a web browser. This way it is not necessary for users to install any software or download data.

Map Tool for Folktale Researchers

The map tool supports selecting subsets of folktale data based on search queries. It visualises the folktales on a map in combination with a timeline (see Figure 2). This dynamic visualisation utilises a faceted navigation technique [English 2002], which can be used to obtain subsets of folktales meeting specific selection criteria and narrow down on folktales of interest. The facets are shown as pie charts on the right side of the interface (see Figure 2) and can be clicked to update the results on the map. The facets are updated as well to show the contents of the subset. The visualisation starts with a full dataset overview from which cross sections can be made. At the moment, the tool only visualises the locations where the folktales were narrated and not, for example, the locations mentioned in the folktales. Still, the completeness of overview of the collection and therefore the accessibility of all corners of the collection is apparent.
The interface relies on shapes and colours to represent several properties of the data. A circle on the map represents a location where at least one folktale was narrated. The size of a circle represents the number of tales in a location. Transparency can be set for the circles to create an effect similar to a heat map, which can give a clear picture of the concentration of folktales in certain areas. To make individual circles more noticeable, they can all be made the same size. This way, small circles will not be hidden behind large ones. Folktales that have not been annotated with a location are currently shown in the North Sea; this is a temporary solution to keep them from being overlooked.
Geographical mapping as a principle can serve as a starting point for international cooperation between folklore researchers by linking their national collections, as advocated by Meder [Meder 2010]. Due to various factors such as ownership, differences in metadata and opinions about content, it is very hard to establish a single international database of folktales. It seems more feasible to let folktale databases exist in their own context, and promote the "donation" of content to a central system. A logical assumption that can be made about folktales is that, with the exception of Internet based tales, they were recorded at a certain time and place. This means that all folktale collections are expected to contain at least time and place as metadata, which makes a map-based interface very suitable as a collective interface.
Figure 2. 
The Dutch Folktale Map Tool. Here, all oral folktales with the tag "spokerij" (translated "haunting") are shown on the map.

Network Tool for Folktale Researchers

Similarity between folktales is of great interest to researchers who are searching for the origins, kinship, variability and stability of tales [Tehrani 2013]. However, most empirical investigations of narrative similarity have only been carried out on a small scale so far [Nguyen et al. 2014]. In our visualisation tool, the similarity of large numbers of folktales can be determined based on different configurations of metadata such as folktale type, keywords, named entities, word count, genre, etcetera. This comparison yields a score that can be used to determine their degree of similarity. In line with Abello et al., our system visualises the similarity between tales as a dynamic network graph (see Figure 3) [Abello et al. 2012].
The tool can be used from different starting points. A single folktale can be loaded to which neighbours, folktales with similar metadata, can be added. These neighbours can subsequently be interconnected and added upon as well. This way, an entire network can be built up from a single item. Another starting point is to do a keyword search to obtain multiple folktales as nodes. These can then be interconnected and added upon as well.
Once a network has been built, it is possible to start exploring grouped nodes and links by clicking on them. There is also a possibility to select multiple nodes to show the composition of this subset, using the same pie chart shaped facets as are available in the map tool. By doing so, the user is invited to find new perspectives to compare folktale stories. The tool gives the user the freedom to hypothesise and test which metadata fields are valuable to compare in order to gain new insights into the relations between folktales, and how these relations evolved over place and time. It could be used to verify existing theories and classifications that have been based on qualitative research. For instance, when connecting folktale nodes purely based on keywords, we found that clusters are formed that show good overlap with the frequently used Aarne-Thompson-Uther (ATU) folktale type-index [Uther 2004]. In other tests with the network tool, users have already reported finding unexpected connections between tales. This is because the tool connects texts outside traditional classification systems that experts commonly use, such as ATU. These early findings suggest that in the future, the tool may open up new avenues for folktale research that go beyond what was previously possible.
Figure 3. 
The Dutch Folktale Network Tool.
A simplified network visualisation has been developed based on the network software. It is included on each folktale page on the Dutch Folktale Database website, see Figure 4. Around the folktale of interest a "flower" of similar documents appears when such documents are available. Users can inspect these similar documents by clicking on them. We believe that both experts and casual users will appreciate this addition, as Elbedweihy et al. found that both groups like their search results to be augmented with related information [Elbedweihy et al. 2012].
Figure 4. 
Simple network visualisation of folktales, incorporated in each folktale webpage.
Both the expert Map tool and the expert Network tool are still under development. We expect their current interfaces to be somewhat hard to understand, because the focus thus far has not been on user friendliness but on functionality. Thorough user evaluations with expert users still need to be carried out to investigate the usability of both interfaces, similar to what was done with the interfaces we designed for casual users, as discussed in the next sections.

Search and Exploration Interfaces for Casual Users

Folktale researchers only form a fraction of the users of the Dutch Folktale Database. As shown by our user polls, the large majority of its visitors are interested in the collection for personal use. To accommodate this user group, we adopted a user-centred approach to design two search and browsing interfaces for casual users. We built these interfaces using the same software as used for the expert visualisation tools discussed in the previous sections, but in their design we focused on simplicity and usability. Unlike folktale researchers, who are strongly motivated to use the collection and are less likely to be discouraged by usability problems, casual users will quickly abandon the site if their experience with it is not good. Our analysis of the use of the Dutch Folktale Database made it clear that visitors need to be encouraged to spend more time on the site, discover more content and come back for more.

Faceted Search Interface for Casual Users

The main access point to the Dutch Folktale Database used to be a simple search box. As noted by Whitelaw, the trouble with this is that it requires a query from the user, even though many users are not seeking specific information [Whitelaw 2015]. Our goal was therefore to design a new search interface that would allow the users to make use of metadata to search the collection, like in the advanced search (see Figure 1), but with an easy-to-use faceted search interface instead of a list of search fields. Such an interface should provide the users with a good overview of the collection and of the different dimensions it can be searched or browsed on. We expected this type of interaction with the collection to be intuitive, as most users will already be familiar with faceted search interfaces from popular web shops such as Amazon.
Our first step was to investigate which of the folktales’ metadata were of interest to casual users, as opposed to domain experts. To this end we recruited a test group of 9 potential users and presented them with a list of metadata that could potentially serve as facets. We asked the participants to indicate for each potential facet whether they would want it to be included in the search interface. The ages of the participants (4 male, 5 female) ranged from 21 to 78 years, with an average age of 43 years. This resembles the age distribution of the visitors of the Dutch Folktale Database. Of the participants, two were scouting leaders and one was a teacher on a primary school; these are the kinds of casual users that would occasionally visit the Dutch Folktale Database looking for stories to tell by the campfire or at school. None of the participants were familiar with the Dutch Folktale Database.
The list of facets a majority of the users showed interest in is shown in Table 2. Examples of potential facets that did not make the selection are "motif", a term from folktale research that was not familiar to the participants, and "literary", a field with a yes or no value indicating whether a collection item was a literary text or not.
Facet (example value) Percentage of votes
keywords (wolf, grandmother) 100%
subgenre (fairytale) 89%
language (standard Dutch) 89%
source type (book) 89%
creator/teller (G. de Boer) 89%
title (Little Red Riding Hood) 78%
date (1971-09-28) 67%
source (oral) 67%
place of telling (Rotterdam) 56%
collector (C. Bakker) 56%
Table 2. 
Metadata selected as search facets by a majority of participants (n=9).
The next step was the design of the search interface. We created several alternative designs and presented them to six of the participants who had judged the potential facets. In our design choices we mainly focused on various options to present the facets, e.g., listing selected facets next to each other or below each other; showing selected facets together with non-selected facets, or in a separate section; showing all possible values of each facet or only the most frequent ones, etcetera. Based on the users’ preferences, we created the final design shown in Figure 5. Selected facets are shown below the main search box, but also among the non-selected facets in the menu on the left. The list of shown values for each facet is limited to a maximum of the eight most frequent ones, but users can fold open the list for more values. After each value, the number of matching items in the current result list is shown. When no search terms have been entered and no facets have been selected, the numbers for all items in the database are shown, giving the user an overview of the entire collection. The new search interface was implemented using an Omeka plugin called SolrSearch, which implements SOLR to become the primary search engine on Omeka-based sites. SolrSearch was made by researchers of the ScholarsLab of the University of Virginia (https://github.com/scholarslab/SolrSearch).
Figure 5. 
The final design of the faceted search interface. The implementation of this final design can be found at http://www.verhalenbank.nl/solr-search.
We tested the usability of the new search interface using a Dutch translation of the Standard Usability Test (SUS) [Brooke 1996]. In this test, participants indicate their level of agreement with ten statements on a Likert scale ranging from 1 to 5. Example statements include, "I found the system unnecessarily complex" and "I thought the system was easy to use". The new design was tested by nine participants recruited from a public library. We expected library visitors to be more likely than the average person to have an interest in folktales, and thus to more be representative of our target group of causal users. To compare the usability of the new search interface with that of the old version, we also tested the old search interface. The test of the old version involved a total of 14 participants: five library visitors, eight bachelor students from our university (study programme: Creative Technology), and one actual visitor of the Dutch Folktale Database who reacted to a call for participants on the website. None of the participants had used either of the tested interfaces before.
Participants were given a brief introduction to the Dutch Folktale Database, and were then asked to do a search using an initial query of their own choice. They were encouraged to explore the results page at their leisure, after which they filled in the SUS questionnaire.
The results of the SUS test are calculated as described by Brooke [Brooke 1996]. A total SUS score of 68 (out of 100) is seen as average. In our test, the old search interface received a SUS score of 60.89, which is below average (Figure 6: SUS score V1). The new faceted design received a SUS score of 81.39, which is well above average (Figure 6: SUS score V2). We can conclude that the new design is a marked improvement over the old search interface. However, a few critical remarks are in order. First, those who tested the new search interface were more evenly distributed in age than those testing the old version, which did not include the age groups 25-34 and 35-44. Second, the students who tested the old search interface might have been more critical of it due to their academic background, and less interested in the contents of the Folktale Database. This may have resulted in a lower SUS score for the old search interface than would have been given by users who were more representative of the target group. Still, Figure 6 shows similar trend lines for both interfaces, so we feel it is safe to say that the overall usability of the new search interface is higher than that of the old version.
Figure 6. 
SUS scores per age group. V1 is the original search interface; V2 is the new faceted search.

Map-based Interface for Casual Users

In a short poll on the website of the Dutch Folktale Database (Spring 2015), 152 out of 240 respondents (63%) said they would like to make use of an interactive map in the future. Therefore we created a second interface, based on the map tool for folktale researchers that was described earlier in this paper. We used the same technology as in the expert map tool, but designed a new interface more suitable for casual users.
Given the complexity and size of the data that can be portrayed with maps, especially interactive ones, there is a strong need for simplicity in design so that the interface remains clear and overseeable for users [Harrower and Fabrikant 2008] [Jones et al. 2009]. We established the following requirements for the map-based user interface, based on the user interface guidelines specified by Brown [Brown 1999]:
  • Minimalism: avoiding unnecessary elements so that the interface is easy to grasp and the content is emphasised.
  • Intuitiveness: the interface should feel natural and take minimal effort to learn.
  • Exploration: the design should encourage exploration of both the stories in the database and of the interface itself.
  • Simplicity: it should take the users a minimal amount of time and steps to find the stories they are interested in.
  • Appeal: the interface should be visually appealing, encouraging users to use it and explore it further.
These requirements can be contradicting; for instance, a minimalistic interface might emphasise the database content well, but could come at the cost of being unintuitive. It is hence very important to find the right balance between the requirements. For this reason, we carried out numerous informal user tests with different interface designs.
The main concept was to provide a map showing folktales in the form of bubbles (similar to the map tool for folktale researchers) and only a few minimal "filter buttons" corresponding to facets that can be used to select subsets from the collection. Figure 7 shows the initial design of the filter buttons. After clicking a filter button, the values of the facet are folded out for selection or deselection by the user. The folktales matching the user’s selection are shown on the map. This way, the users get to see only the stories they are interested in, while still being invited to explore the other folktales in the database.
Figure 7. 
first design of filter buttons for folktale subgenre, source type, language, keywords and collectors.
We included the following filters: folktale subgenre (e.g., legends, jokes), source type (e.g., book, Internet), language/dialects (e.g., standard Dutch, Frisian), keywords (e.g., death, man), and tale collectors. We selected this last category instead of creators/tellers, because users looked more often for collectors than creators in the transaction logs analysed by [Trieschnigg et al. 2013a]. We made a conscious choice to strongly limit the number of facets, preferring minimalism and simplicity of the interface over completeness.
Based on informal feedback of several potential users on our initial design ideas, we created a first mock-up of the interface (Figures 7 and 8) with a few interaction possibilities, but no actual functionality. In a small-scale user test, four people (age range 20-50, two male, two female) solved a few simple tasks with the mock-up. This test showed that some of the icons we used for the buttons were unclear. However, the biggest problem with the interface design turned out to be the visualisation of multiple folktales at the same location on the map. In the mock-up we used a bookmark metaphor for this, using tabs to indicate multiple folktales per location (Figure 8). The test showed that users did not notice these tabs very quickly, and when they did, it was not clear to them that the tabs correlated with multiple folktales. In our final interface design, shown in Figure 9, we therefore used a fundamentally different solution, indicated by (4) in the figure. The folktales are now shown as individual bubbles, placed in a circle around their location. The subgenre of each folktale is indicated by an icon within its bubble. The same icons are also used in the fold-out menu of the subgenre filter button. This provides users with an instant indication of the kinds of stories that are available per location. If the number of folktales per location exceeds the number that can be presented in the circle, a "More" button is included in the circle. When clicked, it shows the next set of folktales at that location. This new circle includes a back button to return to the previous set, and another "More" button if even more folktales are available.
Figure 8. 
Tab-based design for navigating multiple folktales at the same location, showing a fictional example. The summary of the currently selected folktale is shown, with at the bottom a link to the full story in the Dutch Folktale Database and a link to other folktales collected by the same collector.
In the final design we changed the icons of most filter button icons to show a clearer link with the depicted facet; see (1) in Figure 9. At the top left of the interface (2) we added a switch between the map interface (Verhalenkaart in Dutch) and the regular interface of the Dutch Folktale Database (Verhalenbank). Since we expected users to have an interest in folktales at specific locations, e.g., local legends, a location search feature was included. This allows users to instantly go to a specific place on the map by typing in a location. We also included a "Locate me" button for directly zooming in on the user’s current location to show folktales in their vicinity (3). Finally, we customised the map, using a muted colour scheme to avoid distracting the users, and showing only those details which are important for finding folktales (e.g., including town names but leaving out the names of highways).
Figure 9. 
Final design of the map-based interface for casual users. The implementation of this design can be found at http://www.verhalenbank.nl/visuals/map.
To determine the usability of the final design, we created a prototype with limited functionality for a final evaluation experiment. The prototype did not have access to the actual folktale database, but contained all essential elements needed to test the interface and its functionality. The evaluation was carried out with 22 participants, 19 of whom carried out the test online. The remaining three interacted with the prototype in the presence of an observer. Participants varied in terms of their gender (59% male, 41% female), work or study (ranging from healthcare to ICT), and age (between 17 and 50 years, with an average age of 24). Most had received higher education. Only one of them had used the Dutch Folktale Database before.
Participants were given three tasks that explored all the major elements and core functionality of the map-based interface, including using multiple filter options and navigating between several folktales at a single location. After performing the tasks they were asked to fill in the Usefulness, Satisfaction, and Ease of use (USE) questionnaire [Lund 2001], which requires the users to indicate their level of agreement with a number of statements concerning ease of use, ease of learning and satisfaction on 7-point Likert scale. We also included a few interface-specific statements on the effectiveness of the filter options and the navigation between multiple folktales. Finally, we asked them to provide feedback on the interface. The average results of the questionnaire are given in Table 3. The participants were positive on all USE dimensions as well as on the interface-specific statements.
USE questionnaire Avg. Agreement St. Dev.
Ease of Use (11 statements) 5.6 1.3
Ease of Learning (4 statements) 5.9 1.2
Satisfaction (7 statements) 5.3 1.4
Additional statements
I can easily use the filters. 5.9 0.7
I understood what the filter buttons were for. 6.0 1.0
It was easy to navigate between multiple stories per location. 5.8 1.3
Table 3. 
Evaluation results for the map-based interface prototype. Agreement was rated between 1 (strongly disagree) and 7 (strongly agree).
The participants commented that the interface was user-friendly, clear, easy to learn and tranquil. Negative comments mostly pertained to limitations of the interactive mock-up. Other comments included suggestions such as adding hover tooltips and adding extra animations to provide feedback. The three participants we observed while they interacted with the prototype quickly started using the filter buttons, understanding their purpose within a few seconds. No navigation problems were observed; the participants seemed to grasp the concept of multiple folktales around a location quickly and well.
The results indicate that we succeeded in designing an interface for casual users that is minimalistic, simple, appealing and intuitive. To which extent the interface encourages exploration could not be tested given the limited functionality of the prototype. Investigating this requires a longer-term study using the fully functional interface, which has only recently been integrated with the Dutch Folktale Database.

Conclusions and Future Work

In this paper we have described our methodology for designing 'generous interfaces' [Whitelaw 2015] and rich-prospect browsers [Ruecker et al. 2011] for an online collection of folktales, serving two main types of users with different knowledge and different information needs: (1) researchers (expert users), who are interested in patterns and relations between folktales, and who generally have a good understanding of what they are looking for, and (2) casual users who, without browsing possibilities mostly tend to search on genres or well-known fairy tales; see [Trieschnigg et al. 2013a].
In our design process for casual users we followed a user-centred design methodology, of which the main steps can be summarized as follows. The first step was to employ web analytics and surveys to find out who used the collection, how they used it and why. This helped us to obtain insights into the different user groups. Next, the most relevant dimensions for searching and browsing the collection were determined in consultation with representatives of the main user groups. We carried out usability tests of candidate designs using paper prototypes and mock-ups. Based on the results, the final interface designs were established, technically implemented and integrated in the website of our folktale collection
A similar methodology could be applied to other projects aimed at developing user interfaces for digital cultural heritage collections, and we hope our case study can serve as an inspirational example. We believe that taking a user-centred approach is essential for the design of interfaces that are truly "generous" and accessible for different kinds of users. This holds in particular for casual users: usability is an important requirement for them, but their needs may be easily overlooked in the case of collections that, like the Dutch Folktale Database, were originally set up with scholarly use in mind. Not all similar projects need to follow exactly the same steps as ours; however, we believe that (1) getting to know the users and their needs and (2) testing interface design choices with representative users before implementing them are essential steps in any interface design process. Even when done on a small scale and with limited resources, these steps can provide valuable insights.
Our project resulted in several flexible exploration tools and interfaces that exploit the rich metadata in the Dutch Folktale Database for use in faceted search and browsing. The interfaces provide a bird’s-eye view of the collection, allowing both casual users and experts to find useful perspectives on the data. We expect this will encourage casual users to discover more of our national cultural heritage, and help researchers to answer both old and new research questions on a larger scale than was previously possible. In the last decades a number of national folktale database initiatives have been launched in different countries. Our tools can form the basis for an online platform for future cooperation and international data sharing between these folktale databases, making these treasure troves of oral culture accessible to scholars and non-scholars alike.
The newly developed interfaces have been integrated into the website of the Dutch Folktale Database. The next step is to determine whether this indeed has the intended effect of retaining casual users for a longer period of time, encouraging them to explore more folktales, and having them return more often to the website. To this end, a new round of user data collection needs to be done using Google Analytics, and the results need to be compared to our earlier findings. However, this remains as future work. Also future work is an analysis to investigate whether the new interfaces indeed lead to be more unique content being accessed. A comparison of folktale documents by a combination of keywords, subgenre, subject, motifs etcetera showed that about 5% of all documents have no direct neighbours in the collection. These folktales will not show up in the network visualisation currently employed on the website, but they can be exposed via the faceted browsing and map-based interfaces. It remains to be seen whether this is sufficient, or if more needs to be done to facilitate discovery of these items.

Acknowledgements

We would like to thank everybody who participated in our polls and experiments. We also thank the reviewers for their useful suggestions. This research was supported by the Folktales as Classifiable Texts (FACT) project, part of the CATCH programme funded by the Netherlands Organisation for Scientific Research (NWO).

Works Cited

Abello et al. 2012 Abello, J., Broadwell, P. and Tangherlini, T. (2012). "Computational Folkloristics." Communications of the ACM, 55(7), pp. 60–70.
Brooke 1996 Brooke, J. (1996). "SUS: A 'Quick and Dirty' Usability Scale." In: P. Jordan, B. Thomas, B. Weerdmeester, and A. McClelland, eds., Usability Evaluation in Industry. London: Taylor and Francis.
Brown 1999 Brown, C. (1999). Human-Computer Interface Design Guidelines. Exeter: Intellect Books.
Elbedweihy et al. 2012  Elbedweihy, K. Wrigley, S. and Ciravegna, F. (2012). "Evaluating Semantic Search Query Approaches with Expert and Casual Users." In: The Semantic Web (ISWC 2012), Berlin Heidelberg: Springer Verlag, pp. 274-286.
English 2002 English, J., Hearst, M., Sinha, R., Swearingen, K. and Yee, K. (2002). "Flexible Search and Navigation Using Faceted Metadata." In: Proceedings of the 25th Conference on Research and Development in Information Retrieval (SIGIR 2002), New York: ACM, pp. 11–15.
Gould and Lewis 1985 Gould, J. and Lewis, C. (1985). "Designing for Usability: Key Principles and What Designers Think." Communications of the ACM 28(3), pp. 300-311.
Harrower and Fabrikant 2008 Harrower, M. and Fabrikant, S. (2008). "The Role of Map Animation for Geographic Visualization." In M. Dodge, M. McDerby and M. Turner (eds.): Geographic Visualization: Concepts, Tools and Applications, John Wiley & Sons, Ltd, Chichester, UK, (2008): pp. 49-65.
Jones et al. 2009 Jones, C., Haklay, M., Griffiths, S. and Vaughan, L. (2009). "A Less-is-more Approach to Geovisualization -- Enhancing Knowledge Construction across Multidisciplinary Teams." International Journal of Geographical Information Science, 23(8), pp. 1077-1093.
Khoo et al. 2008 Khoo, M., Pagano, J., Washington, A., Recker, M., Palmer, B. and Donahue, R. (2008). "Using Web Metrics to Analyze Digital Libraries." In: Proceedings of the 8th ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL '08), New York: ACM, pp. 375-384.
La Barre and Tilley 2012 K. La Barre, K. and Tilley, C. (2012). "The Elusive Tale: Leveraging the Study of Information Seeking and Knowledge Organization to Improve Access to and Discovery of Folktales." Journal of the Association for Information Science and Technology 63(4), pp. 687-701.
Lund 2001 Lund, A. (2001). "Measuring Usability with the USE Questionnaire." Usability Interface 8(2), pp. 3-6.
Meder 2010 Meder, T. (2010). "From a Dutch Folktale Database towards an International Folktale Database." Fabula 51(1-2), pp. 6-22.
Muiser et al. 2012 Muiser, I. Theune, M. and T. Meder, T. (2012). "Cleaning up and Standardizing a Folktale Corpus for Humanities Research." In: Proceedings of the Second Workshop on Annotation of Corpora for Research in the Humanities 2012 (ACRH-2), Lisbon: Edições Colibri, pp. 63-74.
Nguyen et al. 2012 Nguyen, D., Trieschnigg, D., Meder, T. and M. Theune, M. (2012). "Automatic Classification of Folk Narrative Genres." In: LThist 2012: First International Workshop on Language Technology for Historical Text(s), Vienna: ÖGAI, pp. 378–382.
Nguyen et al. 2013 Nguyen, D., Trieschnigg, D., and Theune, M. (2013). "Folktale Classification Using Learning to Rank." In: Proceedings of the 35th European Conference on Information Retrieval (ECIR 2013), New York: ACM, pp. 195-206.
Nguyen et al. 2014 Nguyen, D., Trieschnigg, D. and M. Theune, M. (20104). "Using Crowdsourcing to Investigate Perception of Narrative Similarity." In: Proceedings of the 23rd Conference on Information and Knowledge Management (CIKM 2014), New York: ACM, pp. 321-330.
Preece et al. 2015 Preece, E., Sharp, H, and Rogers, Y. (2015). Interaction Design. 4th ed. Chichester: John Wiley & Sons.
Ruecker et al. 2011 Ruecker, S. Radzikowska, M. and Sinclair, S. (2011). Visual Interface Design for Digital Cultural Heritage: A Guide to Rich-prospect Browsing. Farnham, Surrey: Ashgate Publishing.
Stephenson 1999 C. Stephenson (1999). "Recent Developments in Cultural Heritage Image Databases: Directions for User-Centred Design." Library Trends, 48(2), pp. 410-37.
Tehrani 2013 Tehrani, J. (2013). "The Phylogeny of Little Red Riding Hood." PloS one 8(11):e78871.
Trant 2006 Trant, J. (2006). "Exploring the Potential for Social Tagging and Folksonomy in Art Museums: Proof of Concept." New Review of Hypermedia and Multimedia, 12(1), pp. 83-105.
Trieschnigg et al. 2012 Trieschnigg, D., Hiemstra, D., Theune, M., de Jong, F. and T. Meder, T (2012). "An Exploration of Language Identification Techniques for the Dutch Folktale Database." In: Proceedings of the Adaptation of Language Resources and Tools for Processing Cultural Heritage workshop at LREC 2012. Istanbul: ELRA, pp. 47-51.
Trieschnigg et al. 2013a Trieschnigg, D., Nguyen, D. and Meder, T. (2013). "In Search of Cinderella: A Transaction Log Analysis of Folktale Searchers." In: Proceedings of the First ACM SIGIR Workshop on the Exploration, Navigation and Retrieval of Information in Cultural Heritage (ENRICH 2013).
Trieschnigg et al. 2013b Trieschnigg, D., Nguyen, D. and Theune, M. (2013). "Learning to Extract Folktale Keywords." In: Proceedings of the 7th Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities (LaTeCH 2013), Stroudsburg: Association for Computational Linguistics, pp. 65-73.
Uther 2004 Uther, H.J. (2004). The Types of International Folktales: A Classification and Bibliography Based on the System of Antti Aarne and Stith Thompson. Vols 1-3. Helsinki: Suomalainen Tiedeakatemia.
Whitelaw 2015 Whitelaw, M. (2015). "Generous Interfaces for Digital Cultural Collections." Digital Humanities Quarterly 9(1). http://www.digitalhumanities.org/dhq/vol/9/1/000205/000205.html