The Phoebe A. Hearst Museum of Anthropology (PAHMA) is the oldest and largest anthropological collection in the western U.S., with some 3.8 million objects spanning nearly 2 million years of human culture. But with just 4000 square feet of exhibit space, only a tiny fraction of the Museum's collections are available for public viewing. Anthropologists and educators around the world want access to the collections for research and teaching, and the Museum is eager to expose the depth and breadth of their collections to the public. For Michael Black, the Museum's Head of Research & Information Systems, the obvious solution was a web-based collections browser. Unfortunately, the Museum lacked the financial resources, staff, and technical expertise to develop such a system.
In fall 2007, a group of graduate students in the UC Berkeley School of Information (I School) were looking for a museum partner for their Master's final project. They wanted to apply Natural Language Processing (NLP) techniques — specifically, the kind that are currently driving the booming enterprise search sector — to the cultural heritage sector. The students brought technical expertise in NLP, information architecture, database design, and web technologies, balanced with expertise in user-centered design, needs assessment, and information visualization. An IST contact working with the Berkeley Natural History Museums (BNHM) consortium connected PAHMA and the I School students, and the Delphi project was born.
The Delphi project team spent a feverish six months interviewing Museum staff, researchers, and museum website users, developing the semantic search toolkit, and designing and building the web-based collections browser. Delphi features support searching, browsing, and viewing information about objects, creating and viewing sets of objects, and annotating objects in the collections. The Delphi faceted browser (Figure 1.) provides a simple way to navigate through museum collections, and helps visitors discover concepts and relationships between objects that may not otherwise be obvious. The concept listing on the left is derived automatically from an ontology that organizes concepts associated with the collections into distinct facets such as location, culture, and materials. Delphi's semantic toolkit supports text analysis for ontology development, as well as the text mining required to generate the semantic search index. Associated services support the collections browser, and a flexible presentation architecture can be easily customized to match the style and layout of a given institution's website.
Figure 1. Delphi faceted browser.
A private beta of the initial version allowed Museum staff to review the site — this was the first public exposure of their entire collection online. It also allowed Native American Tribal representatives to review the collections and provide input on the presentation of culturally sensitive artifacts.
At this point, IST–Data Services (IST–DS) picked up the project and coordinated a production deployment based upon IST's WebFarm and Database services. One of the Delphi technical team (Patrick Schmitz) joined IST–DS, and the Museum contracted with another team member (Jon Lesser, now at the Center for History and New Media at George Mason University) to continue development. A second I School team conducted a usability study during the beta, providing additional feedback. In spring 2008, a revised website with improved user interface, richer collections information, and a visual redesign was deployed on the IST infrastructure and released to the public.
Among the key users are Native Americans whose cultures are represented in the collections. To date, tribal feedback has been positive and many tribal representatives are excited to have online access to the Museum's collections. Delphi has had a particularly profound impact on the more remote tribes whose members cannot easily visit the physical museum, but want to — and now can — browse the Museum's collections online from any computer including those in tribal offices, cultural heritage centers, and local libraries. Such access greatly enhances ongoing partnerships between PAHMA and tribes in collection research, exhibitions, artifact loans, and repatriation claims.
Schmitz and Black presented the Delphi toolkit and the PAHMA deployment experience at the Museums and the Web 2008 conference, and this fall, Black will present Delphi to a European conference for archives, libraries, and museums. In addition to several UC Berkeley museums exploring a Delphi deployment, the conference exposure has generated interest in the open-source toolkit from museums around the world.
Later this fall, PAHMA will publicize Delphi and introduce a wide audience to the site in conjunction with a capital campaign to raise funds for expanded facilities. The Museum will seek feedback from educators, students, researchers, donors, and the general public around the world who will benefit from the new website and the broad exposure to world culture it provides.
Delphi development continues, and will include integration with CollectionSpace, a project to develop an open-source collections management system in which IST–DS and the Museum are also involved.