More about LIO and Other Attributes Used in ImageSnippets™

Help Topics

ImageSnippets™ uses a very small ontology, called 'lio' (lightweight image ontology) - described below - to relate the image and regions in it to other data with triple-tags. We also use a subset of attributes from more commonly recognized, pre-existing multimedia resource schemas in our Details tab. These fields are also imported from and embedded in the image header (when the image is uploaded and saved in IS) for greater functionality with a variety of image applications. Some of these fields also make use of the look-up function to allow the selection of URI-defined entities for the object value. More explanation regarding our intentions with these fields and their suggested use is described on the Details, Keywords and Description page.

From the Dublin Core Schema:

dc:title - Title
dc:subject - Subject

dc:creator - Creator
(the creator field allows a look-up
of entities in subscribed datasets)

dc:rights - copyright notice


xmpRights:usageTerms - for copyright terms
iptc4xmpcore:CiEmailWork - Creator e-mail(s)

iptc4xmpcore:CiUrlWork - Creator website(s)
note: The URL of the html+RDFa created by IS is exported in this field to add an additional reference point back to the metadata for applications that recognize the use of this field.

iptc4xmpcore:sceneCode - IPTC Scene Code

XMP specifies the use of photoshop
for the following elements:

photoshop:DateCreated, photoshop:City
photoshop:State, photoshop:Country

rdfs:comment is used for the contents of the description tab

The Materials, Styles and Techniques options also found on the Details tab use URI-defined entities for the object values and a look-up function is provided to add more entities than are displayed. These properties were modeled after similar properties in the Visual Resources Association (VRA) schema.

We are committed to the recognition and re-use of existing metadata standards while also creating relations that satisfy our intended uses for the interface within the linked data world. We have deliberately used a subset of these attributes for the beta release of ImageSnippets™. Customized versions of the application can use any combination of and/or full specifications of any schema(s) and/or associated controlled vocabularies. Additionally, advanced users are free to add their own namespaces (click on the namespaces link in a metadata editor window) and use their own properties (such as foaf:depicts and others) for triple-tags in the editor.

The following concepts and terms are used in lio:
Images and works.

The subject of the lio: relations is conventionally described as the image, but when the image is a direct rendering of an artwork, such as a thumbnail of the Mona Lisa, the logical subject is understood to be that artwork. In general, the subject is always the work, but for most images other than images of artwork, the image itself is the work. (This would be problematic if the photographic image is both a depiction of an artwork and is also a significant work in its own right, but such cases are very rare.)

Class values and representing typicality

Several of the lio: relations are allowed to take values which are classes rather than individuals, with the meaning that the actual value is some 'typical' member of the class. For example, one may assert that part of an image lio:looksLike, which appears to not make sense (how can something look like the class of kitchens?) but which does under this convention. This simplifies the annotations by avoiding a large number of triples involving blank nodes which assert that some anonymous thing is in the class.


This relates an image to the main thing that the image is a picture of. In most cases this is the obvious visible subject or topic of the image. The thing depicted can be anything: a person, an object, a place, or even an event such as a race or a football match. There can be several things depicted: a photo of a group of people might depict each of them. Some images might not depict anything.

lio:depicts is a superproperty of the inverse of foaf:depiction, which is almost exclusively used for 'mug shots' of people. It is broader in its range than most published relations concerned with depiction.


Something which is visible in the image, but only incidentally. Use this for objects that one would not describe the image as being a picture of, even though they are visible in it. Again, the thing shown can be anything, and there may be several of them. Some images might not show anything.


Used to indicate something in the image which visually resembles something, without actually being a picture of that thing. The canonical example is a face in the clouds, but this can also be used describe iconic shapes (a heart, a star). The shape can be an image artifact such as lens flare, or an accidental image alignment, or something visible in the depicted scene. The 'class as typical instance' device can be used with this property.


Used to describe an emotional or visual impression communicated by an image. The value might be sadness, joy, etc., or movement, stability, restlessness, etc.. The object URIs for this property are usually found in DBpedia or Yago.


Used to identify the general location or event where the image was created, or the circumstances surrounding its creation, especially when this is not immediately obvious from the image itself. This could be a general description (a kitchen) or a particular (1976 Grand Prix). The 'class as typical instance' device can be used with this property.


Used to identify something which is used in the image in an artistic or artificial way. As with looksLIke, this may be an aspect of the image itself (a colored shape) or of the scene depicted (the edge of a building viewed in extreme perspective.) The 'class as typical instance' device can be used with this property.


Used to identify noteworthy elements of design, composition and principles of art: line, color, texture, etc.. The 'class as typical instance' device is usually required, as the relevant parts or aspects of the image almost never have URIs assigned to them.


Used to attach image tags within RDFa. Value is a single tag, not a list or sequence of tags.


Used to attach information about the sublocation where the image was created which may be incidental to a place, event or setting observed in a work. This could refer to any geographical level or be less precisely individuated than a traditional geographical hierarchy.


Used to attach information about the materials used in a work. Intended to be synchronized with the Visual Resources Association vocabulary.


Used to attach information about the style of a work. Intended to be synchronized with the Visual Resources Association vocabulary.


Used to attach information about the techniques used in a work. Intended to be synchronized with the Visual Resources Association vocabulary.