Computer-Implemented System And Method For Providing Visual Suggestions For Cluster Classification

• US 20140059048A1
• Filed: 10/28/2013
• Published: 02/27/2014
• Est. Priority Date: 07/28/2009
• Status: Active Grant

• Current Assignee: Nuix North America Inc. (Nuix Ltd.)
• Original Assignee: FTI Consulting Incorporated
• Inventors: Knight, William C., Nussbaum, Nicholas I.
• Granted Patent: US 9,336,303 B2
• Time in Patent Office: Days
• US Class Current: 707/737
• CPC Class Codes: G06F 16/287 Visualization; Browsing
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 16/358 Browsing; Visualisation therefor
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 16/954 Navigation, e.g. using categorised browsing
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 5/047 Pattern matching networks; Rete networks
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

An embodiment provides a computer-implemented system and method for providing visual suggestions for cluster classification. One or more clusters comprising uncoded documents from a set are obtained. A different set of reference documents that are each classified with a code is designated. A cluster center in one of the clusters is identified. The cluster center is compared to one or more of the reference documents. Those of the reference documents that are similar to the cluster are identified based on the comparison. The classification codes of each of the similar reference documents are visually represented as a suggestion for assigning one of the classification codes to the cluster.

Claims

1. A computer-implemented system for providing visual suggestions for cluster classification, comprising:

a cluster module to obtain one or more clusters comprising uncoded documents from a set;

a designation module to designate a different set of reference documents that are each classified with a code;

a cluster center module to identify a cluster center in one of the clusters and to compare the cluster center to one or more of the reference documents;

a similarity module to identify those of the reference documents that are similar to the cluster based on the comparison; and

a representation module to visually represent the classification codes of each of the similar reference documents as a suggestion for assigning one of the classification codes to the cluster.

2. A system according to claim 1, further comprising:

a display module to display a visual representation of the cluster together with the visual representations of the classification codes of the similar reference documents.

3. A system according to claim 1, wherein the visual representation of the each classification codes comprises at least one of a symbol, shape, and color different from the visual representations of the remaining classification codes.

4. A system according to claim 1, further comprising:

a spine module to determine a similarity between the clusters and to organize the clusters along one or more spines based on the similarity, each of the spines comprising a vector;

a spine vector module to identify cluster centers for all of the clusters along one of the spines and to generate a score vector for the spine based on the cluster centers;

a reference document vector module to generate score vectors for the reference documents based on content of the reference documents;

a comparison module to compare the score vector for the spine to the score vectors for the reference documents and to identify those of the reference documents that are similar to the spine based on the comparison of the spine score vector to the score vectors of the reference documents;

a spine classification module to visually represent the classification codes of each of the reference documents similar to the spine as a suggestion for assigning one of the classification codes to the spine.

5. A system according to claim 4, further comprising at least one of:

a spine neighborhood module to generate a neighborhood comprising a predetermined number of the reference concepts most similar to the spine based on the comparison of the spine score vector to the score vectors of the reference documents and to assigning one of the classification codes to the spine based on the classification codes of the reference documents in the neighborhood; and

a cluster neighborhood module to generate a neighborhood for the cluster comprising a predetermined number of the reference concepts most similar to the cluster based on the comparison of the cluster center to the reference documents and to assign one of the classification codes to the cluster based on the classification codes of the reference documents in the neighborhood for the cluster.

6. A system according to claim 1, further comprising:

a list module to represent one or more of the uncoded documents from the set in a list displayed on a display;

a selection module to receive a user selection of one of the uncoded documents in the displayed list;

a comparison module to compare the selected uncoded document to the reference documents and to identify reference documents that are similar to the selected uncoded document based on the comparison of the selected uncoded document to the reference documents; and

a display module to represent the reference documents similar to the selected uncoded document and the classification codes for each of the reference documents similar to the selected uncoded documents on the display together with the displayed list.

7. A system according to claim 1, further comprising:

a document neighborhood module to compare one of the uncoded documents in one of the clusters to the reference documents and to generate a neighborhood comprising a predetermined number of the reference documents that are most similar to the uncoded document based on the comparison of the uncoded document to the reference documents; and

a suggestion module to suggest one of the classification codes for assignment to the uncoded document based on the classification codes of the reference concepts in the neighborhood.

8. A system according to claim 7, wherein the predetermined number comprises a sum of predetermined numbers of reference documents for each of the classification codes.

9. A system according to claim 7, further comprising:

a distance module to determine a distance between the uncoded document and the reference documents in the neighborhood; and

a generation module to generate the suggested classification code for the uncoded document, comprising at least one of;

identifying the reference document with the closest distance to the uncoded document and to assign the classification code of the reference document with the closest distance as the suggested classification code for the uncoded document;

a maximum count module to calculate an average of the distances between the uncoded document and the reference documents associated with each of the classification codes and to assign the classification code with the closest average distance as the suggested classification code of the uncoded document; and

a weighted maximum count module to count the reference documents in the neighborhood for each of the classification codes, to weigh each count based on the distance between the reference documents included in the count and the uncoded document, and to assign the classification code with the highest weighted count as the suggested classification code of the uncoded document.

10. A system according to claim 7, further comprising:

a confidence level module to provide a confidence level for the suggested classification code.

11. A computer-implemented method for providing visual suggestions for cluster classification, comprising:

obtaining one or more clusters comprising uncoded documents from a set;

designating a different set of reference documents that are each classified with a code;

identifying a cluster center in one of the clusters and comparing the cluster center to one or more of the reference documents;

identifying those of the reference documents that are similar to the cluster based on the comparison; and

visually representing the classification codes of each of the similar reference documents as a suggestion for assigning one of the classification codes to the cluster.

12. A method according to claim 11, further comprising:

displaying a visual representation of the cluster together with the visual representations of the classification codes of the similar reference documents.

13. A method according to claim 11, wherein the visual representation of the each classification codes comprises at least one of a symbol, shape, and color different from the visual representations of the remaining classification codes.

14. A method according to claim 11, further comprising:

determining a similarity between the clusters and organizing the clusters along one or more spines based on the similarity, each of the spines comprising a vector;

identifying cluster centers for all of the clusters along one of the spines and generating a score vector for the spine based on the cluster centers;

generating score vectors for the reference documents based on content of the reference documents;

comparing the score vector for the spine to the score vectors for the reference documents and identifying those of the reference documents that are similar to the spine based on the comparison of the spine score vector to the score vectors of the reference documents;

visually representing the classification codes of each of the reference documents similar to the spine as a suggestion for assigning one of the classification codes to the spine.

15. A method according to claim 14, further comprising at least one of:

generating a neighborhood comprising a predetermined number of the reference concepts most similar to the spine based on the comparison of the spine score vector to the score vectors of the reference documents and assigning one of the classification codes to the spine based on the classification codes of the reference documents in the neighborhood; and

generating a neighborhood for the cluster comprising a predetermined number of the reference concepts most similar to the cluster based on the comparison of the cluster center to the reference documents and assigning one of the classification codes to the cluster based on the classification codes of the reference documents in the neighborhood for the cluster.

16. A method according to claim 11, further comprising:

representing one or more of the uncoded documents from the set in a list displayed on a display;

receiving a user selection of one of the uncoded documents in the displayed list;

comparing the selected uncoded document to the reference documents and identifying reference documents that are similar to the selected uncoded document based on the comparison of the selected uncoded document to the reference documents; and

representing the reference documents similar to the selected uncoded document and the classification codes for each of the reference documents similar to the selected uncoded documents on the display together with the displayed list.

17. A method according to claim 11, further comprising:

comparing one of the uncoded documents in one of the clusters to the reference documents and generating a neighborhood comprising a predetermined number of the reference documents that are most similar to the uncoded document based on the comparison of the uncoded document to the reference documents; and

suggesting one of the classification codes for assignment to the uncoded document based on the classification codes of the reference concepts in the neighborhood.

18. A method according to claim 17, wherein the predetermined number comprises a sum of predetermined numbers of reference documents for each of the classification codes.

19. A method according to claim 17, further comprising:

determining a distance between the uncoded document and the reference documents in the neighborhood; and

generating the suggested classification code for the uncoded document, comprising at least one of;

identifying the reference document with the closest distance to the uncoded document and assigning the classification code of the reference document with the closest distance as the suggested classification code for the uncoded document;

calculating an average of the distances between the uncoded document and the reference documents associated with each of the classification codes and assigning the classification code with the closest average distance as the suggested classification code of the uncoded document; and

counting the reference documents in the neighborhood for each of the classification codes, weighing each count based on the distance between the reference documents with the classification code and the uncoded document, and assigning the classification code with the highest weighted count as the classification code of the uncoded document.

20. A method according to claim 17, further comprising:

providing a confidence level with the suggested classification code.