Three-dimensional display of document set

• US 7,555,496 B1
• Filed: 09/26/2006
• Issued: 06/30/2009
• Est. Priority Date: 08/12/1996
• Status: Expired due to Fees

• Current Assignee: Battelle Memorial Institute
• Original Assignee: Battelle Memorial Institute
• Inventors: Pottier, Marc C., Lantrip, David B., Pennock, Kelly A., Wise, James A., Schur, Anne, Thomas, James J., York, Jeremy
• Primary Examiner(s): MIZRAHI, DIANE D
• Application Number:
  US11/535,360
• Time in Patent Office: 1,008 Days
• Field of Search: 707/2, 707/3, 707/10, 707100-102, 382/255, 382/305, 715/209, 715/260, 706/46
• US Class Current: 1/1
• CPC Class Codes: G06F 16/34 Browsing; Visualisation therefor
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

A method for spatializing text content for enhanced visual browsing and analysis. The invention is applied to large text document corpora such as digital libraries, regulations and procedures, archived reports, and the like. The text content from these sources may be transformed to a spatial representation that preserves informational characteristics from the documents. The three-dimensional representation may then be visually browsed and analyzed in ways that avoid language processing and that reduce the analysts'"'"' effort.

Claims

1. A method for visually representing on a display information from a plurality of semantic objects, comprising the steps of:

Defining at least one topic or topic attribute;

Forming clusters, using a computer processor, from the plurality of semantic objects, wherein a cluster defines a group of semantic objects arranged by their respective relative significance to the at least one topic or topic attribute;

Quantitatively characterizing respective clusters;

Defining, for a plurality of semantic objects, a quantitative difference of the respective semantic object from the quantitatively characterized respective cluster; and

Outputting for the display representations of at least one topic or topic attribute and the plurality of semantic objects into a representational space, wherein the representational space has a lower dimensionality associated with each of the semantic objects, an arrangement of representations is influenced by a semantic relation of a semantic object and the at least one topic or topic attribute;

creating a plurality of topic or topic attribute layers, each of the layers corresponding to a topic or topic attribute applied to each respective cluster, and identifying a coordinate location within the representational space for each semantic object associated with each layer; and

defining an amplitude surface within the representational space associated with each layer by applying a smoothing function to the coordinate location representations of the semantic objects, and superimposing upon one another all of the amplitude surfaces of the respective layers.

2. A method for visually representing on a display information from a plurality of semantic objects, comprising the steps of:

Defining at least one topic or topic attribute;

Forming clusters, using a computer processor, from the plurality of semantic objects, wherein a cluster defines a group of semantic objects arranged by their respective relative significance to the at least one topic or topic attribute;

Quantitatively characterizing respective clusters;

Defining, for a plurality of semantic objects, a quantitative difference of the respective semantic object from the quantitatively characterized respective cluster; and

Outputting for the display representations of at least one topic or topic attribute and the plurality of semantic objects into a representational space, wherein the representational space has a lower dimensionality associated with each of the semantic objects, an arrangement of representations is influenced by a semantic relation of a semantic object and the at least one topic or topic attribute;

wherein said outputting step seeks to preserve a distance between pairs of semantic objects in said representational space based on the defined quantitative difference of the respective semantic object from the quantitatively characterized respective cluster.

3. A method for visually representing on a display information from a plurality of semantic objects, comprising the steps of:

Defining at least one topic or topic attribute;

Forming clusters, using a computer processor, from the plurality of semantic objects, wherein a cluster defines a group of semantic objects arranged by their respective relative significance to the at least one topic or topic attribute;

Quantitatively characterizing respective clusters;

Defining, for a plurality of semantic objects, a quantitative difference of the respective semantic object from the quantitatively characterized respective cluster; and

Outputting for the display representations of at least one topic or topic attribute and the plurality of semantic objects into a representational space, wherein the representational space has a lower dimensionality associated with each of the semantic objects, an arrangement of representations is influenced by a semantic relation of a semantic object and the at least one topic or topic attribute;

wherein the clustering step comprises producing a partition set on the plurality of semantic objects, by applying a clustering algorithm with primary emphasis on k-means and complete linkage hierarchical clustering to define a cluster centroid, the cluster centroid representing the quantitative characterization of a respective cluster.

4. The method according to claim 3, wherein the cluster centroid is created by a process comprising:

(i) selecting a number of seeds;

(ii) placing said seeds in a hyperspace by sampling regions to ensure a specified distribution of seeds;

(iii) identifying non-overlapping hyperspheres for each cluster and assigning each semantic object to each respective cluster based on which hypersphere the respective object is located;

(iv) calculating a centroid coordinate, representing the center of the mass for each cluster; and

(v) repeating steps (iii) and (iv) until centroid movement is less than a specified threshold.

5. A method for visually representing on a display information from a plurality of semantic objects, comprising the steps of:

Defining at least one topic or topic attribute;

Forming clusters, using a computer processor, from the plurality of semantic objects, wherein a cluster defines a group of semantic objects arranged by their respective relative significance to the at least one topic or topic attribute;

Quantitatively characterizing respective clusters;

Defining, for a plurality of semantic objects, a quantitative difference of the respective semantic object from the quantitatively characterized respective cluster; and

Outputting for the display representations of at least one topic or topic attribute and the plurality of semantic objects into a representational space, wherein the representational space has a lower dimensionality associated with each of the semantic objects, an arrangement of representations is influenced by a semantic relation of a semantic object and the at least one topic or topic attribute;

wherein each of the quantitative characterization of a respective e cluster represents a cluster centroid, the outputting step comprising the steps of applying a Multi-dimensional Scaling Algorithm to cluster centroid coordinates in hyperspace, producing a vector for each semantic object with distance measures from the semantic object to each cluster centroid, and constructing an operator matrix and multiplying the operator matrix by the vector to produce coordinates for each semantic object.

6. A method for visually representing on a display information from a plurality of semantic objects, comprising the steps of:

Defining at least one topic or topic attribute;

Forming clusters, using a computer processor, from the plurality of semantic objects, wherein a cluster defines a group of semantic objects arranged by their respective relative significance to the at least one topic or topic attribute;

Quantitatively characterizing respective clusters;

Defining, for a plurality of semantic objects, a quantitative difference of the respective semantic object from the quantitatively characterized respective cluster; and

Outputting for the display representations of at least one topic or topic attribute and the plurality of semantic objects into a representational space, wherein the representational space has a lower dimensionality associated with each of the semantic objects, an arrangement of representations is influenced by a semantic relation of a semantic object and the at least one topic or topic attribute;

wherein each of the quantitative characterization of a respective e cluster represents a cluster centroid, the outputting step comprising the steps of applying an Anchored Least Stress Algorithm to cluster centroid coordinates in hyperspace, producing a vector for each semantic object with distance measures from the semantic object to each cluster centroid, and constructing an operator matrix and multiplying the operator matrix by the vector to produce coordinates for each semantic object.

7. A method for visually representing on a display information from a plurality of semantic objects, comprising the steps of:

Defining at least one topic or topic attribute;

Forming, using a computer processor, clusters from the plurality of semantic objects, wherein a cluster defines a group of semantic objects arranged by their respective relative significance to the at least one topic or topic attribute;

Quantitatively characterizing respective clusters;

Defining, for a plurality of semantic objects, a quantitative difference of the respective semantic object from the quantitatively characterized respective cluster; and

Outputting for the display representations of at least one topic or topic attribute and the plurality of semantic objects into a representational space, wherein the representational space has a lower dimensionality associated with each of the semantic objects, an arrangement of representations is influenced by a semantic relation of a semantic object and the at least one topic or topic attribute;

wherein the semantic objects have a native high dimensional space, and a distance between semantic objects in the representational space is non-linearly related to a distance between semantic objects in the native high dimensional space.