oa Mining flikr for museum feedback: Case study on the Qatar Islamic Museum of Art

Museums are increasingly placing digital images of portions of their collections online. Weblogs of visits to these online 'digital museums' can then analysed to give the collection curators a sense of what portions of the digital collections are attracting the greatest degree of interest (as coarsely measured by 'hits' on images). Some sites support commenting by viewers, and these comments also can be useful in gaining insight into the interests of the viewers [1] and can be 'mined' to suggest additional useful metadata for the collection [2]. However, these digital museums generally include only a small fraction of the items on display in the physical museum. We propose a method for uncovering visitor interest across the entire physical museum's display, by 'mining' a public repository of personal photographs for photos taken during visits to the museum. The visitors can be assumed to take photos of those museum exhibits that has the highest personal impact on them—for example, if they find the item to be interesting, shocking, entertaining, or so forth. In this case study we examine photos of visits to the Museum of Islamic Art in Doha, as uploaded to Flickr. Specifically, we select a sample of 500 photos tagged with "Qatar Islamic Museum" (the keywords that appear to be most common for this museum). From that sample we manually categorized the images to be either photos of the museum itself, or photos of item(s) in the collection. For each of these base categories, we further downloaded and analysed the metadata applied by the photographer to the image, Comments that other Flickr users added, and the number of times that the image was Favorited by Flickr users. The photographs of collection items were further manually manually grouped and characterized by item type (eg, Ceramics, Tapestries, etc.). Where possible, multiple images of the same object were identified. These results of the manual analysis support an insight into the visitor experience to the Museum of Islamic Art. We also present a prototype of a 'control panel' for the Museum that will allow the Museum's curators to automate the process of downloading photos of the Museum and its collection, together with the Flickr metadata accessible through the Flickr API (Comments, Favorite counts, etc). This data is then displayed in an Excel spreadsheet together with the original photo, and macros support the curator in analysing the data. We argue that this system will support the curators in seeing the Museum from a visitor's point of view—to gain insights into what they enjoyed, valued, and connected with most strongly. [1] Cunningham, S.J., and Mahoui, M. 2013. Interacting with and through a digital library collection: Commenting behaviour in Flickr's The Commons. Proceedings of JCDL 2013, 23 - 26. [2] Momeni, El., Tao, K., Haslhofer, B., and Houben, G-J. 2013. Identification of useful user comments in social media: A case study on Flickr Commons. Proceedings of JCDL 2013, 4 - 12.