The Metropolitan Flunks Subject Search

General Description

The digital image collection available on the website of The Metropolitan Museum of Art in New York is a mixed bag.  I refer, of course, not to the contents of the collection, but to the display and searchability of the subject matter of the works in the collection.

The Met’s description of the digital collection  is as follows:

The Collection Database is a searchable database of artworks and related materials from the permanent collection of the Metropolitan Museum of Art. An individual database record includes information about an object as well as images, when available. The Museum’s curators have selected several works of note within the collection Database as Highlights of the collection. Due to the extremely large number of objects in the Museum’s permanent collection, not all artworks are currently available in the collection Database. Furthermore, information contained in the database records is, in some cases, incomplete, and all information is subject to change according to ongoing research and new acquisitions.

This statement acknowledges that not all objects in the permanent collection are contained in the database, and that those which are do not always have associated images, BUT it would have been nice for the Met to give the user some sense of proportionality.  According to the site there are 128,347 items in the online collection.  But I can’t tell form this what portion of the permanent collection this represents,  The only piece of further information I can easily find is that the Highlights contain only 1407 objects. Equally important I can’t tell how many of  the 128,347 objects have associated images.  I can say, unfortunately, that it appears that the vast majority of the items don’t.  One can browse the works to see page after page of thumbnails stating “image not available”   What a drag. It is not a bad thing that so many objects lack images; better to have access to the text record than to have nothing at all.  It is a major oversight, however, that the interface does not allow one to filter out those objects without images. Met, add a simple filter to allow the user to browse only the objects with images.  That would make the experience significantly for many visitors.   It should be noted that the Met does offer a Note to the Reader which “explains selected terms and practices mentioned elsewhere on this website.”  This short section is certainly helpful to the uninitiated.  The most interesting tidbit here was that “Some photographs of works of art require photograph copyrights. These are indicated as follows: Photograph © 1972″  This is information that I have not previously seen elsewhere.   Where did it appear?

Textual Information

As for the text record itself, the Met is generous with this information which includes:

Gallery Label
Notes
Provenance
Exhibitiom History
References

From many examples I looked at this information was quite thorough and very interesting.  Far more information is provided about the works than is ever found on the curatorial tags adjacent to works on display in the galleries. While I have no way of being certain, it seems that the full record in the Met’s own content management system is provided to the the visitor.  This a huge step toward making the works available in a meaningful way, and in this respect I believe the Met is ahead of the game in sharing information. The inclusion of this text, being the only searchable information, also adds measurably to the overall usefulness of the search functions, which is where my real interest lies.  The usefulness of this text for subject matter searches is still very limited as these were not written with that purpose in mind.  But depending on the detail in the gallery label and notes, it offers some chance of finding images by subject matter.  But, of course, you have to choose the right word; there is no controlled vocabulary.  And typically only the primary objects in the picture are mentioned.  More on this, my real subject, in a bit.

An aside: I hope some one is working on making this textual record available whenever the object is displayed.  That is a perfeect opportunity for the non invasive use of RFID technology.  Why not put the associated digital record of a work on an RFID tag, attached that tag to the object, and put display screens for each object right next to them when they are on display.  All that information could travel with the object and be displayed wherever it travelled.  No doubt someone is toying with this idea on a grand scale and I simply haven’t come across it yet.

No doubt the administration office of institutions are nixing such a project for cost reasons in this dreadful economic time.  But one must dream about possibilities.  Why should ready access to this information be available online, but not when when is viewing the object itself?

Browsing

The browsing experience is quite simple and flexible.  One can use the list view where each row shows the Image Thumbnail, the Title of Work or Type of Object, Artist or Maker. Date, and Accession Number.  Clicking at the top of each column reorders the contents of the column either alphabetically or numerically.  This is nice, but relatively useless for quick searching when one cannot also jump through the Alphabet by letter, or through the numbers in some chunked way.  Not when there are hundreds of pages.  Isn’t this obvious?  How hard is it to think these things through, guys?  The image to the left is a small part of the list page, which contains 20 images, in this case  for the Robert Lehman Collection.

In the Image View, shown on the right for a page of the Cloisters Collection, thumbnails appear in an array of 20 per page without any text.  Clicking on an image brings up the full record of the work.   It is perfectly satisfactory.  It could be enhanced,  though, by allowing the user the option as to how many thumbnails should appear per page.  Many folks prefer to scroll through a long pages than to click thorough shorter pages.  Just visit any decent online clothing store to learn how to do this.

There is another significant omission in the browse (and search) functions. As noted before one can tell by browsing that the vast majority of the 128,00 item in the collection don’t have any image; but there is no way to find out how many high resolution images there are.  And  there is no way to limit ones search to these images only.   That should be a must.

Image View

It is at the individual item view that I became supremely despondent.  The main item page still contains only the the small thumbnail and you must click to “enlarge” it.  But be careful how you take that term.  For the Met it means  going from an image of about 300 px wide to one that is  500  px wide.

Not much of an increase, and one that hardly takes advantage of the resolution of today’s monitor screen sizes, not to mention even larger display formats.  On the left is the largest view of Pierre Bonnard’s The Terrace at Vernonnet, a mere 500 px wide.   Images at this size merely provide an impression of the work.

The Met tries to compensate for this by allowing, for some unknown number of works (could these be the “Highlights?”), access to a fairly high resolution image (Great), but only through a similarly paltry sized window.   The image on the right is the highest resolution of the Bonnard on the left.  It shows a tiny portion of the image, the top of the wine bottle on the table and the area to the right of it.   A rough estimate of the full high resolution image is about 5000 px wide.  It is hardly satisfying to get a close peek at a work when the window is so small and it is so difficult to move around in the image.  To click and drag inside that narrow window is quite frustrating.  Overshoot your tug and nothing happens.  Even when you do it correctly there is no way to judge the movement necessary to get to where you want to be in the image.    But this functionality is of limited use, so it doesn’t matter a lot.  The only good thing I can say is that the detail captured in these high resolution images is quite amazing and give you a view of the works not otherwise available, even in a on site visit.  I’d note though that at the highest enlargement some of the images seem a bit pixilatted.  This serves no purpose.  Now if only you could see them in a reasonable size.  Why, I ask, can’t we have access to the 5000 px image, and be limited only by the size of our own display?

Subject Search

The Digital Image Collection offers a basic simple keyword search that presumably searches all the associated text.  There is also an Advanced Search which lets you to search the entire site or to limit the search to one of the curatorial divisions :

• American Decorative Arts
• American Paintings and Sculpture
• Ancient Near Eastern Art
• Arms and Armor
• Arts of Africa, Oceania, and the Americas
• Asian Art
• The Cloisters
• The Costume Institute
• Drawings and Prints
• Egyptian Art
• European Paintings
• European Sculpture and Decorative Arts
• Greek and Roman Art
• Islamic Art
• The Robert Lehman Collection
• The Libraries
• Medieval Art
• Modern Art
• Photographs
• Antonio Ratti Textile Center
• Provenance Research Project

On another page a brief description of the scope of each of the curatorial divisions is provided.  You can further refine an Advanced Search by entering a search term for one or more of the following subjects:

• Artist or Maker
• Title of Work or Type of Object
• Country or Culture of Origin
• Medium or Materials
• Date of Creation by year or date range
• Credit Line or Reproduction Rights
• Description
• Provenance
• Inscription or Marks
• Accession Number
• Keyword

But sad to say there is no real subject matter search, even of a limited sort, via a hierarchical tree.  The best one can do is to search within the Title or the Description in the hope that the subject you are looking for appears in one of those two field.

I conducted my standard test, a search for “rose.” A simple keyword search of the entire digital collection gave 1,400 results.  Using the advance search function “rose”  yielded 1660 results.   These searches are helpfully broken down by the curatorial area in which they are found.  This confirmed what I found by browsing the results: the vast majority of these were from the the costume collection (e.g. 1046 out of 1660.)   While many of those in the costume collection referred to the use of a rose motif in the item, many referred to the color of the garment or accessory.  Among the other images a deeper look revealed that the image came up because it was on rose paper, or used rose ink, or made of rose gold, or was the gift or work of someone named Rose (first or last name).  In a few cases there seems to be no reason that the item was retrieved in this search at all.  Objects with a rose as (or part of) the subject matter really came up only when “rose” was in the title of the work. The only time I found rose in the description referring to the flower rose was in a work titled Maria Luisa of Parma, the reference was actually  to a different work by the same artist, and so was inapropos.  Interestingly, the fairly detailed description for this painting was unusual in containing references to many of the objects in the painting.

Parker discusses the furniture and site in detail. He notes that the armchair corresponds closely to a chair belonging to the MMA, part of a set acquired by the sitter’s mother in Paris in about 1749, which he tentatively attributes to the Parisian chairmaker Nicolas Quinibert Foliot. He identifies the rug as a Ushak carpet, noting that they were imported in quantity from Turkey to Europe in the 18th century. The console table he calls French, either made in Paris or produced by a member of the small colony of French woodworkers in Parma. Observing that after 1753 the court of Parma had snuffboxes supplied directly from Paris, he identifies the example held by the Princess as “almost certainly of French manufacture.”

Someone looking for images of snuffboxes should actually find the one in this painting.  Were that more descriptions contained this level of detail.  at least until images are properly cataloged for subject matter or tagged by the public.

But in general the descriptions do not provide mush subject matter detail as this example from one of Giotto’s paintings shows:

Giotto was the most famous artist of his day, praised by Dante and considered by his successors to have revived the art of painting after centuries of decline. This panel is part of a series of scenes from the life of Christ, of which six others are known. They may come from one of the four altarpieces by Giotto recorded in the Franciscan church of Santa Croce in Florence. The way in which the figures move freely within a carefully described space is close to that of frescoes painted by the artist in about 1320 in the same church.

Notes: This panel is from a series of scenes from the life of Christ to which six others belong: the Presentation in the Temple (Gardner Museum, Boston), the Last Supper, the Crucifixion, and the Descent into Limbo (all, Alte Pinakothek, Munich), the Entombment (Villa I Tatti, Florence), and Pentecost (National Gallery, London).  According to Christiansen’s reconstruction [see Ref. 1982], the panels were arranged horizontally as an independent altarpiece or the predella of an altarpiece. Opinions differ regarding the original location. Stylistic similarities to the frescoes in the Bardi and Peruzzi chapels may suggest that the series comes from one of the four altarpieces by Giotto recorded in the Franciscan church of Santa Croce in Florence [see Refs. Longhi 1930 and Tartuferi 2000]. The series has also been identified with a work of Giotto mentioned by Vasari in Sansepolcro [see Refs. Vasari 1568, Davies 1951, and Bologna 1969]. Gordon, however, proposes a Riminese provenance based in part on the combination of the Nativity and the Epiphany in the MMA picture [see Ref. 1989].The back of the panel was gessoed and painted with an imitation-porphyry decoration that has been obliterated by graffiti, some in a fourteenth- or early fifteenth-century hand [see Ref. Christiansen (1982].”

Given the common use of the rose in the imagery of art through the ages, these results tell me that the descriptions of works in the Met’s collection are poor substitutes for real subject matter classification, and of little help in a subject matter search.

On this score the Met gets an F.

Strengths

• Easy browsing features.  List view and Image View.  Five  columns that can be used for ordering.
• The grey and soft green background are good choices for viewing
• “Add to Own Gallery” feature
• Automatic breakdown of search results by Curatorial Department

Weaknesses

• Small image size
• Limited view of high resolution images
• Few  items in the collection have images (or high resolution image)
• No way to search for just those work with images
• At highest enlargement some images seem a bit pixilatted

Tagged as: controlled vocabulary, Giotto, Image search, Metropolitan Museum of Art, New York City, RFID, rose, Subject search