Ink Model URI Scheme

The Ink Model specification defines a way to use Uniform Resource Identifiers (URI) to reference nodes of the InkModel's primary ink-tree or views by specifying a scheme in accordance with the guidelines in RFC 3986.

URI Scheme

The specification uses the following terms as they are defined in RFC 3986:

• scheme
• authority
• path
• segment
• unreserved characters
• pchar
• pct-encoded characters
• fragment

The Ink Model specification provides a specific URI scheme for identification of InkModel nodes and other elements, using the following grammar:

• URI = "uim:" [ //authority ] [path] [#fragment]
• authority = skipped
• path = 1 * ( "/" segment )
• segment = 1 * ( pchar )
• fragment = optional

Remarks:

The URI's authority component is skipped. All path definitions in the table below are considered path-rootless according to RFC 3986.

The path component may begin with a segment containing a percent-encoded URI (defined in RFC 3986, Section 2.1: https://tools.ietf.org/html/rfc3986#section-2.1) that uniquely identifies an Ink Model. In case the InkModel's identification segment is skipped, the URI is used for local node referencing on Ink Model level and the URI is treated as local URI.

Important note: Implementations must encode URIs of entities, which belong to the serialized model, only in their local form (using local URIs).

Ink Model

Path: <pct-encoded InkModel URI>/node/<node-id>

Fragment: n/a

Description: URI of a node, contained within an Ink Model. The <pct-encoded InkModel URI> segment contains a percent-encoded URI of an InkModel. This segment is optional. The <node-id> segment contains an > > InkNode identifier, encoded as a GUID string value.

Example: Suppose we have an InkNode with GUID 5454710e-608d-4eee-9552-680f3349ab3e, contained within the main ink tree or a view of an InkModel with URI https://example.com/inkmodel/9b67babb-a2a3-491b-96ce-71c820ce0c64, (provided by the provider https://example.com/).

URI of the InkNode in the InkModel's scope (local ID): uim:node/5454710e-608d-4eee-9552-680f3349ab3e

URI of the InkNode (globally unique): uim:https%3A%2F%2Fexample.com%2Finkmodel%2F9b67babb-a2a3-491b-96ce-71c820ce0c64/node/5454710e-608d-4eee-9552-680f3349ab3e

Named entities

Path: <pct-encoded InkModel URI/>ne/<ne-id>

Fragment: n/a

Description: URI of a named entity definition, usually used to group knowledge about a specific InkModel node.

The <pct-encoded InkModel URI> segment contains a percent-encoded URI of an InkModel. This segment is optional. The <ne-id> segment contains an identifier of a named entity definition.

MD5-hash based Unique Identifier Generation Algorithm

The described algorithm allows generation of unique identifiers based on a tag (encoded as string value) and a collection of components.

The supported component types are:

• Integer number
• Floating-point number
• String
• List of properties defined as key-value pairs; the key and value of a pair are considered to be of type string.

The algorithm is expressed in pseudocode below. The described method takes a tag value and a list of components as arguments and generates a unique MD5 hash as an 8-byte array.

METHOD generateMD5Hash(tag, components): RETURNS byte[8]

    SET buffer = new StringBuffer()

    buffer.write(tag)
    buffer.write("\n")

    FOREACH component OF components
        IF (component is defined)
            buffer.write(component.convertToString())
        buffer.write("\n")

    RETURN MD5_HASH(buffer.toString())

The convertToString() method used in the pseudocode snippet above should be implemented as follows per each component type:

• Integer number - convertToString() returns the string representation of the integer number

• Floating-point number - convertToString() returns the string representation of the floating-point number using fixed-point number representation with 4 digits after the decimal point

• String - convertToString() returns the string value

• List of properties - convertToString() pseudocode implementation:

METHOD convertToString(properties): RETURNS string

    SET buffer = new StringBuffer()

    properties = SORT(properties) // sort the properties list by ascending alphabetical name order

    FOREACH pair OF properties
        buffer.write(pair.name)
        buffer.write("\n")
        buffer.write(pair.value)
        buffer.write("\n")

    RETURN buffer.toString()

Vocabulary

Sensor Channel Types

The table below describes a list of common sensor channel types, defined as URIs, which are used to distinguish between different sensor characteristics.

URI	Description
will://input/3.0/channel/X	Sensor channel storing the x-coordinates on a surface or within 3D space.
will://input/3.0/channel/Y	Sensor channel storing the y-coordinates on a surface or within 3D space.
will://input/3.0/channel/Z	Sensor channel storing the y-coordinates within 3D space.
will://input/3.0/channel/Timestamp	Sensor channel storing the time difference to the previous sample.
will://input/3.0/channel/Pressure	Sensor channel storing the pressure applied on the pen tip.
will://input/3.0/channel/RadiusX	Sensor channel storing the x-radius of a touch event.
will://input/3.0/channel/RadiusY	Sensor channel storing the y-radius of a touch event.
will://input/3.0/channel/Azimuth	Sensor channel storing the azimuth angle of a pen device.
will://input/3.0/channel/Altitude	Sensor channel storing the altitude angle of a pen device.
will://input/3.0/channel/Rotation	Sensor channel storing the barrel rotation of the pen device.

Core Elements

The table below describes a list of WILL Core element types, defined as URIs.

URI	Description
will://core/3.0/core/Path	Ink stroke
will://core/3.0/core/Sensordata	Sensor data
will://core/3.0/core/Brush	Brush

Common InkModel Views

The table below describes a list of known InkModel views.

URI	Description
will://views/3.0/HWR	Currently provided by the Semantic Ink Services.
will://views/3.0/NER	Currently provided by the Semantic Ink Services.

Remark This section is still in experimental state and the listed definitions may change.

Semantic

The table below describes a list of types and entities, used to store ontological knowledge definitions into the ink model's knowledge graph.

URI	Description
will://semantic/3.0/hasNamedEntityDefinition	Predicate the defines a dedicated entity storing named entity recognition (NER) results.
will://semantic/3.0/hasSource	Predicate for the source of the knowledge graph.
will://semantic/3.0/hasUri	Predicate for the URI within the knowledge graph.
will://semantic/3.0/hasAbstract	Predicate for a short abstract to describe the entity.
will://semantic/3.0/hasImage	Predicate for the image describing the semantic entity.
will://semantic/3.0/hasThumb	Predicate for a thumbnail to the concept type.
will://semantic/3.0/hasArticle	Predicate for a URL pointing to article about a semantic entity.
will://semantic/3.0/hasType	Predicate for describing the type of the concept.
will://semantic/3.0/hasCategory	Predicate for the category of the semantic entity.
will://semantic/3.0/hasLabel	Predicate for label of the semantic entity.
will://semantic/3.0/is	Predicate for best math from handwriting recogntion.
will://semantic/3.0/hasAlt	Predicate for alternative recognition results.
will://semantic/3.0/hasBBox	Predicate for bounding box of the recognized word.
will://semantic/3.0/hasRelevantConcept	Predicate for relevant concept type if there are multiple types found by NER.

Remark This section is still in experimental state and the listed definitions may change.

Content Segmentation

The table below describes a list of types and entities used to describe an ink model's structure in terms of content segmentation.

URI	Description
will://segmentation/3.0/TextRegion	Defining a larger region with text.
will://segmentation/3.0/Paragraph	Groups text lines in a paragraph.
will://segmentation/3.0/TextLine	Groups words in a text line.
will://segmentation/3.0/Word	Groups ink strokes in a word.
will://segmentation/3.0/Phrase	Groups words in a logical in a unit.
will://segmentation/3.0/LineBreak	Defining a linebreak.
will://segmentation/3.0/MathBlock	Grouping ink strokes to a math formula.
will://segmentation/3.0/Shape	Grouping ink strokes to a shape.

Remark This section is still in experimental state and the listed definitions may change.

Shapes

The table below describes a list of types and entities in terms of shape recognition and definition.

URI	Description
will://shapes/3.0/Rectangle	Groups ink strokes as a rectangle.
will://shapes/3.0/Circle	Groups ink strokes as a circle.
will://shapes/3.0/Ellipse	Groups ink strokes as a ellispe.
will://shapes/3.0/Triangle	Groups ink strokes as a triangle.
will://shapes/3.0/Doodle	Groups ink strokes as a doodle.
will://shapes/3.0/Arc	Groups ink strokes as an arc.
will://shapes/3.0/Line	Groups ink strokes as a line.
will://shapes/3.0/Polygon	Groups ink strokes as a polygon.
will://shapes/3.0/Parallelogram	Groups ink strokes as a parallelogram.
will://shapes/3.0/Rhombus	Groups ink strokes as a rhombus.
will://shapes/3.0/attr/points	Attribute describing a list of points.
will://shapes/3.0/attr/pt	Attribute describing a single point.
will://shapes/3.0/attr/pt1	Attribute descibing the first point.
will://shapes/3.0/attr/pt2	Attribute descibing the second point.
will://shapes/3.0/attr/pt3	Attribute descibing the third point.
will://shapes/3.0/attr/radius	Attribute describing the radius.
will://shapes/3.0/attr/radiusX	Attribute describing the radius x.
will://shapes/3.0/attr/radiusY	Attribute describing the radius y.
will://shapes/3.0/attr/rect	Attribute describing a rectangle.
will://shapes/3.0/attr/centerPoint	Attribute describing the center point.
will://shapes/3.0/attr/angle	Attribute describing an angle.
will://shapes/3.0/attr/phiAngle	Attribute describing a phi angle.
will://shapes/3.0/attr/startAngle	Attribute describing a start angle.
will://shapes/3.0/attr/sweepAngle	Attribute describing a sweep angle.

Remark This section is still in experimental state and the listed definitions may change.