Why Images?
As JSTOR's Mission & Goals suggest, we strive to
meet many objectives and to satisfy the needs of all participants. The original
concept for JSTOR was to convert the back issues of paper journals into electronic
formats that would allow for savings in space (and in capital costs associated with
that space) while simultaneously improving access to the journal content. Thus, it
is equally as important for JSTOR to be providing faithful replications of the
original print journals as it is for it to be providing access to the archive,
since the electronic version is to be used as a substitute for the print
version.
One important technological decision that JSTOR made was to deliver the content
of the archive as images. We decided to combine the advantages of page images with
a searchable text index, and JSTOR stores the data in both forms. JSTOR delivers
scanned page images to its users, while using the raw text files (created using
Optical Character Recognition (OCR) software) behind the images for search
purposes.
Benefits of Images:
- Faithful Replication: If, in keeping with our mission to
function as a trusted archive, JSTOR is to serve as a substitute for the journal
volumes on the shelves, it must offer an electronic version that is a faithful
replication of the original. An image-based approach ensures the integrity of the
materials in the archive, while also retaining the appearance and "look and feel"
of the journal in its original presentation. This is central to our mission and a
key basis upon which JSTOR was founded.
- Representation of Non-Text Content: Whether they appear as
photographs, charts, tables, or special characters and formulae, certain
components of articles generally cannot be displayed with 100% accuracy using
text-based methods available to standard web browsers.
- Accuracy of Images: Page images are 100% accurate. JSTOR
creates a text index for search purposes as part of its production process
through the use of OCR software. Our scanning vendor conducted a series of
reviews of OCR samples on a variety of materials and found a 97% average accuracy
rate (on uncorrected text). In JSTOR, some journals will have OCR accuracy rates
as high as 99.95%. But, although our OCR is accurate for search purposes, it is
unacceptable for display, owing to typographical, word order, formatting, and
other elements that are not accurately represented. The appearance of
typographical and other errors could undermine the perception of quality that
publishers have worked long and hard to establish and that users of all kinds
expect. Indeed, while users in the visually impaired and learning disabled
communities might prefer text, displaying our OCR would not offer a product and
experience that is equivalent to what users in the non-visually impaired and
non-learning disabled communities encounter. Appropriate assistive technology
designed specifically for the visually impaired and learning disabled communities
can offer far better accuracy than JSTOR can were we to display the OCR’d
text we have created for search purposes.
The importance to libraries and publishers, as well as to the fulfillment of our
not-for-profit mission, of faithful replications of journals, the ability to
display non-textual material accurately, and the problems associated with
displaying the OCR text we have created for search purposes are the primary
motivations for JSTOR's use of images as the mechanism for delivery of journal
articles.
We are aware that our image based approach causes certain difficulties for users
who are visually impaired or learning disabled and use assistive technologies to
access material on the Internet. JSTOR now offers options to help alleviate some of
these difficulties. For more information, please see JSTOR and Accessibility.
