Apparatus and methods for accessing a collection of content portions

• US 7,028,053 B2
• Filed: 01/16/2003
• Issued: 04/11/2006
• Est. Priority Date: 01/16/2003
• Status: Expired due to Fees

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Chi, Ed H., Pirolli, Peter L.
• Primary Examiner(s): Kindred, Alford W.
• Application Number:
  US10/248,408
• Publication Number:
  US 20040143580A1
• Time in Patent Office: 1,181 Days
• Field of Search: 707 1- 4, 707/6, 707100-102, 707/200, 707/104.1, 707/500.1, 707/501.1, 715/500, 715/700
• US Class Current: 1/1
• CPC Class Codes: G06F 16/34 Browsing; Visualisation therefor
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/958 Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing
• CPC Class Codes: Y10S 707/99948 Application of database or data structure, e.g. distributed, multimedia, or image

Abstract

Techniques for browsing through a large collection of content portions uses a scatter and gather approach where a collection of content portions is clustered into one or more clusters using multi-modal data modeling. Additionally, as part of the multi-modal data modeling, “proximal cues” surrounding links or connections or surrounding “image links” in a content portion are used to quickly identify the user'"'"'s information needs. Thus, taking into account proximal cues during multi-modal data modeling improves the scattering and the gathering process, as well as personalizes the scattering and the gathering process to most effectively cluster content portions of interest to the user.

Claims

1. A computer implemented method for accessing a collection of content portions, comprising:

(a) determining multi-modal information for at least one content portion in the collection of content portions;

(b) determining proximal information for the at least one content portion in the collection of content portions, the proximal information including at least one proximate cue that surrounds a link and/or connection, the proximate cue including a word, text, image and/or sound;

(c) generating a plurality of centers for clustering content portions of the collection of content portions;

(d) clustering each content portion in the collection of content portions to a center of the plurality of centers having similar multi-modal information and proximal information;

(e) retaining content portions of at least one selected cluster;

(f) purging content portions of each non-selected cluster; and

(g) wherein, when at least two clusters are selected, merging the selected clusters.

2. The method of claim 1, further comprising:

determining at least one user path in the collection of content portions;

determining at least one content portion associated with the user path; and

using the content portion associated with the user path when clustering each content portion in the collection of content portions to the center having similar multi-modal information and proximal information.

3. The method of claim 1, further comprising:

generating for each content portion, a vector based on the multi-modal information and the proximal information of the content portion; and

using the vectors as basis for determining similarity between content portions.

4. The method of claim 1, further comprising generating a summary for at least one generated cluster.

5. The method of claim 1, further comprising generating proximal information based on an image in a content portion.

6. The method of claim 1, further comprises repeating the steps of (c)–

(g).

7. A system computer implemented usable to access a collection of content portions, comprising:

a multi-modal determining circuit, routine or application usable to determine multi-modal information and proximal information for at least one content portion in a collection of content portions, the proximal information includes at least one proximate cue that surrounds a link and/or connection, the proximate cue includes a word, text, image and/or sound; and

a clustering circuit, routine or application usable to generate a plurality of centers, and to cluster each content portion in the collection of content portions to one of the plurality of centers using the multi-modal information and proximal information of that content portion,

wherein the clustering circuit, routine or application retains content portions of at least one selected cluster and purges content portions of each non-selected cluster and merges selected clusters when at least two clusters are selected.

8. The system of claim 7, further comprising an interface usable to display at least one cluster generated by the clustering circuit, routine or application.

9. The system of claim 7, further comprising:

a user path determining circuit, routine or application that determines at least one user path in the collection of content portions; and

a content portion determining circuit, routine or application that determines at least one content portion associated with the user path;

wherein the clustering circuit, routine or application receives the at least one determined content portion from the content determining circuit, routine or application.

10. The system of claim 7, further comprising:

a multi-modal vector construction circuit, routine or application usable to construct, for each content portion, a vector based on the multi-modal information and the proximal information of that content portion; and

a vector similarity determining circuit, routine or application that determines a similarity between content portions based on the constructed vectors.

11. The system of claim 7, further comprising a summary circuit, routine or application usable to generate a summary for at least one generated cluster.

12. The system of claim 7, wherein the clustering circuit repeats generating the plurality of centers, clustering each portion in the collection of content portions to one of the plurality of centers using the multi-module information and proximal information of that content portion, retaining content portion of at least one selected cluster and purging content portion of each non-selected cluster, and merging selected clusters when at least two clusters are selected.