Executive Summary of Findings

Introduction

For several hundred years, botanists have been collecting, documenting, and archiving plants. These collections, housed in herbaria, provide specimens and data for studies of plant taxonomy and evolution. They also represent a critical material history of ecological state and environmental change through time. Over the past 10-20 years, Herbarium collections have increasingly been digitized and made available online. The associated search and discovery interfaces are effective for the scientific community, but assume the user has specialized knowledge of plants and knows what they are looking for. As such, these digitized herbarium collections remain inaccessible to large portions of the population.

The Herbarium User Experience (HerbUX) project aims to design an interface to be used in classrooms, museums, and other public spaces, that increases access to these critical collections for non-expert audiences such as students and members of the public. It will be easy to use, encourage non-directed exploratory browsing, directly support pedagogical methodologies and learning outcomes, and be aesthetically engaging. Our project builds upon previous experiments in interface design and is based on input from the user groups described below. In this first prototype development stage of the project, we provide interface mockups to support future proposals that will include funding for programmers and UX design professionals to build the interface itself. We did not set out to evaluate or critique existing interfaces. Rather, our intention was to be guided by best UX-design principles and input from potential users to begin development of a new interface that increases accessibility to the rapidly expanding database of digitized herbarium specimens.

Identifying user needs: meetings with experts

We convened three online conferences with professionals from stakeholder groups to identify specific interface design and accessibility needs for different target audiences. Each meeting was recorded and transcribed, and we subsequently performed qualitative analysis to identify themes and patterns from each session. Participants were asked to discuss a particular scenario related to the use of digitized natural history collections, and then to outline key features of an interface that would be needed. In each case, we outlined the scenario as a way to elicit key interface features (a “wish list”) from the participants during small group breakout discussions. The most important of these features (i.e., those that were mentioned the most often) were incorporated into our prototype design.

The three expert groups were:

Summary of themes by meeting

These three expert groups represent users with strongly contrasting needs, in particular relating to how the users interact with the interface directly, and how the information is used once it has been extracted from the database. For example:

• Instructors developing natural history collections-based activities often require easily accessible, pre-filtered and curated datasets. Students are more or less a “captive audience” that may spend extended periods on the interface and may return multiple times.
• We expect that museum goers will spend a short amount of time browsing the interface without a particular question/goal in mind, and may never return. An engaging “hook” to grab attention is needed, and outcomes/learning should be clear very quickly to maintain users’ interest.
• Herbarium professionals access collections records daily. Engagement is not a concern but ease-of-use and full data accessibility are essential. They may need to generate a visual overview of their collections for outreach purposes or when reporting to administrators or for generating grant reports.

Major themes

Access and engagement were strong and consistent themes across all expert groups. Participants consistently mentioned their wish to make interfaces to plant specimen collections accessible to and usable by non-specialist audiences, and to find ways of making the collections compelling and interesting so that those audiences would choose to engage with the data. Participants suggested a variety of detailed ideas to promote access and engagement. From this, we identified 4 major themes:

Place specimens in a broader context

This fits with the concept of the “extended specimen” in which specimen data are enriched through a multilayered network of connections to other datasets and disciplines. The authors of the extended specimen concept centered on extensions into scientific domains–for example, a specimen could be “extended” to include data about morphology, ecological context, DNA sequences, conservation status, etc. However, our participants also advocated for connecting data in more humanistic domains, such as historical context and human culture associated with specific plants or plant species. Most cited the importance of introducing humanistic context as a way of reaching out and engaging audiences, providing a “jumping off point” for learning more about historical context, connections to studies of social history and anthropology, the “plants-people-places” concept (CITATION) and for documenting historical expeditions. Giving more context to the specimens outside of the scientific realm helps foster general interest in plants, science, and conservation. Metadata on specimen labels provides a trove of potentially interesting information that could be the basis of these connections.

Specific suggestions included:

Proposed extended specimen data sources

• Tying in the ecology and natural history of different species, e.g. information on pollinators, habitat, images of live plants, plant conservation status (rare, invasive, etc.)
• Annotating specimens to increase utilization of specimen images and data in other disciplines (e.g., flower color)
• Including pictures of live plants of the same species as the specimens; also ecological context, elevation, range and range shifts over time
• Placing specimens in a humanistic/social context by linking to resources such as DBPedia.
• Prioritizing aspects that allow audiences to connect to their everyday lives: medicine, food, geography, historical events, art, cultural uses, land ownership.
• Mixing science and humanities: the history of science, tracking collection biases, the personalities and populations that engage in science, medicine, and tie-ins with colonialism.

Presentation and visualization of extended data

• Collector and collection narratives (e.g. Storymaps)
• Highlighting the history and collections of individual botanists/collectors, e.g. by mapping all of the specimens collected by one person
• Visualizing the extended specimen as an interconnected web with the plant specimens at the center, radiating out to other data: ethnobotany, art, literature, history, geography, etc.

Design search and browse functions (“discovery”) in a way that engages user interests

Visually compelling and accessible landing and search pages are essential to grab and hold users’ attention. This is especially relevant for members of the general public, who (in contrast to herbarium workers, scientists, and educators and their students) may have no specific reason to engage with the data portal. Image-based browsing and searching can be used as a “hook” to engage these users.

Meeting participants also noted that the general public rarely searches by taxonomic designation or scientific name, both of which are the most common search approaches currently employed. The entry point into the database – either by searching or browsing – should not require specialized botanical or scientific knowledge, and should encourage serendipitous discovery. Users should be able to search by facets and terms that matter to them: for example, several participants suggested the possibility of searching by color, although this particular aspect may be difficult given the discoloration of dried specimens.

Specific suggestions included:

Entry point/landing page
Landing page:

• why are you here – just browsing, or with a specific question? Different answers take you down a different paths
• present curated categories based on general topics of public interest for browsing
• generate a random stack of specimens to browse through without any input at all

Searching, browsing, and filtering

• Get away from the search box: easy filtering using icons, map polygons, general categories e.g., growth form
• Place- or map-based searches
• Google-type search that brings up the most likely item first and then possible related items
• Endless scroll, driven by adjectives
• allowing users to search for and view collections by expeditions

Use a familiar, physical metaphor to generate and arrange subcollections

To maximize utility and engagement, users must be able to easily handle and work with search outputs. Meeting participants emphasized the need for digital tools that mimic the handling of specimens in real-life, based on a visualization using specimen images on a “digital workbench”. Such needs for “active sorting” included subsetting a full collection and/or grouping certain specimens, as well as downstream uses of curated subcollections. This came up most frequently in the Educators and Herbarium Professionals groups, as museum visitors tend to have a one-time interaction with the data.

Many people talked in slightly different terms about being able to sort into “piles” or “stacks” on a “table” or “bench”. Some didn’t refer to the digital workbench metaphor but mentioned that they would like to be able to create arbitrary (or user-defined) collections that wouldn’t necessarily be possible just through metadata filtering (for an example, see the Smithsonian Learning Lab). These “piles” should be shareable (for collaboration), and saveable (e.g. for classes).

“Digital workbench” suggested features

• drag-and-drop of specimen images for manual sorting
• support selecting the “best exemplar” within a pile
• ability to compare among piles.
• batch annotations
• metadata export
• saving and sharing of curated subcollections

Other suggestions for search output visualization

• A cloud of thumbnails that moves and shifts and pulses as filters are applied. These should be filters that use imagery representing how non-specialists see plants (see Barnes Foundation search interface)
• Flip book

Prioritize visual aesthetics in the design of interfaces and presentations

Meeting participants overwhelmingly emphasized the importance of visual aesthetics. Generally, this means using images preferentially over text to attract and engage users, especially non-expert audiences. High-quality images of herbarium specimens, which are generally included as part of digitized specimen records, are an obvious source of compelling visuals. Too often, however, the specimen images are small, or hidden, or overwhelmed by the text of the accompanying metadata. For experts, much of the value of these collections lies in the metadata, and this natural focus gets transferred to the design of the interfaces. But data in a spreadsheet is a hard sell for non-experts.

Along with bringing the specimen image to the fore, participants also noted the importance of metadata visualization to illustrate the scope of a collection or sub-collection. This could be carried out in a number of different ways, highlighting different aspects of the metadata, but one of the most common visualizations suggested was a map to highlight geographic distributions.