System and method for automatic generation of presentations based on agenda

• US 20080040340A1
• Filed: 01/08/2007
• Published: 02/14/2008
• Est. Priority Date: 03/31/2004
• Status: Active Grant

• Current Assignee: Satyam Computer Services Limited (Tech Mahindra Limited)
• Original Assignee: Satyam Computer Services Limited (Tech Mahindra Limited)
• Inventors: Rao, Korrapati, Amur, Ravi, Varadarajan, Sridhar
• Granted Patent: US 8,689,097 B2
• Time in Patent Office: Days
• US Class Current: 707/5
• CPC Class Codes: G06F 40/103 Formatting, i.e. changing of presentation of documents automatic justification G06F40/189; automatic line break hyphenation G06F40/191
• CPC Class Codes: G06F 40/186 Templates

Abstract

In enterprises, presentations are designed and made to visiting clients on a routine basis and it is expected that such presentations be effective as much of business depends on right kind of information in right depth and volume made to right audience. The typically followed manual process depends on the capabilities of a person to create such presentations. It is required to obtain as little information as possible to generate presentations as much automatically as possible. The automatic generation of presentations is based on (a) agenda; (b) audience profile; and (c) templates. The system analyzes an agenda and audience profile to obtain adequate information to identify the most appropriate templates and uses the same to generate one or more draft presentation versions that best match the requirements of the agenda. Further, the invention also provides for an automatic generation of (a) slide annotations; (b) presentation theme; and (c) templates.

Claims

1. A system for automatic generation of presentations and templates based on annotated templates and slides, said system comprising:

an agenda analysis element for processing an agenda to extract a plurality of terms;

an audience analysis element for obtaining a plurality of factors about a plurality of persons;

an agenda theme generation element for processing an agenda to extract a plurality of terms related to the theme of said agenda;

an agenda item generation element for processing an agenda to extract a plurality of terms related to an item of said agenda;

a template selection element for selecting a template based on an item of an agenda;

a slide matching element for comparing a slide with respect to a slide of a template;

a slide selection element for selecting a slide based on a slide of a template;

a draft presentation generation element for determining a plurality of slides that form part of a presentation;

a draft presentation selection element for selecting a presentation for an item of an agenda;

term-pair matching element for comparing two terms;

a term-set matching element for matching two term sets;

a key term identification element for determining a plurality of key terms from a plurality of terms;

a slide annotation element for annotating a slide based on a text content of said slide;

a presentation theme generation element for determining a theme of a presentation based on annotations of a plurality of slides of said presentation; and

a template generation element for generating a plurality of templates from a plurality of annotated presentations.

2. The system of claim 1, wherein said agenda analysis element comprises means for determining terms from an agenda, wherein said determining comprises:

extracting a plurality of terms from said agenda based on a pre-defined dictionary of terms; and

computing frequency of each of said plurality of terms.

3. The system of claim 1, wherein said audience analysis element comprises means for determining a plurality of factors about a term from the profile of a plurality of persons part of an audience, wherein said determining comprises:

determining designation of each of said persons;

determining area of work of each of said persons;

determining division of work of each of said persons;

determining dF of said plurality of factors based on a pre-defined lookup table;

determining pF of said plurality of factors based on a pre-defined lookup table; and

determining kF of said plurality of factors based on a pre-defined lookup table.

4. The system of claim 1, wherein said agenda theme generation element comprises means for determining a plurality of terms part of a theme of an agenda, wherein said determining comprises:

determining a plurality of ag-terms of said theme based on said agenda analysis element;

determining a plurality of au-terms based on said audience analysis element; and

resolving each of said plurality ag-terms with respect to each of said plurality of au-terms.

5. The system of claim 1, wherein said agenda item generation element comprises means for determining a plurality of terms part of an item of an agenda, wherein said determining comprises:

determining a plurality of ag-terms of said item based on said agenda analysis element;

determining a plurality of au-terms based said audience analysis element; and

resolving each of said plurality of ag-terms with respect to each of said plurality of au-terms.

6. The system of claim 1, wherein said template selection element comprises means for selecting a template for an item of an agenda from a database containing a plurality of templates, wherein said selecting comprises:

determining a plurality of theme-terms related to a theme of said agenda;

determining a plurality of item-terms related to said item;

matching said plurality of theme-terms and said plurality of item-terms with template-theme of each of said plurality of templates; and

identifying a best matching template based on said matching.

7. The system of claim 1, wherein said slide matching element comprises means for matching a slide from a database containing a plurality of slides based on a template-slide of a template, wherein said matching comprises:

matching TAnnotation of said slide with TAnnotation of said template-slide and computing a match score (M2), wherein said TAnnotation of said slide is a set of plurality of terms and said TAnnotation of said template-slide is a set of plurality of terms;

matching GAnnotation of said slide with GAnnotation of said template-slide and computing a match score (M3), wherein said GAnnotation of said slide is a set of plurality of terms and said GAnnotation of said template-slide is a set of plurality of terms; and

computing SlideMatchScore as a sum of M2 and M3.

8. The system of claim 1, wherein said slide selection element comprises means for selecting a slide from a database containing a plurality of slides based on a template-slide of a template, wherein said selecting comprises:

matching said template-slide with each of said plurality of slides;

computing SlideMatchScore of each of said plurality of slides based on said slide matching element; and

selecting one of said plurality of slides based on best SlideMatchScore.

9. The system of claim 1, wherein said draft presentation generation element comprises means for determining a plurality of slides that form a draft presentation (P) for an item of an agenda based on a template, wherein said determining comprises:

generating each of said plurality of slides based on said slide selection element;

counting the number (N1) of slides in said plurality of slides;

determining the required number (N2) of slides based on presentation duration in said item of said agenda;

checking if N1>

N2 and removing of slides from P based on identified optional slides, identified similar slides, and identified slides each with a low slide match score;

checking if N1<

N2 and adding slides to P based on identified optional slides and identified similar slides; and

evaluating P (pMatch) based on slide match score of each of said plurality of slides.

10. The system of claim 1, wherein said draft presentation selection element comprises means for selecting a presentation for an item of an agenda, wherein said selecting comprises:

generating a plurality templates for said item based on said template selection element;

generating a plurality of presentations based on said plurality of templates and said draft presentation generation element;

determining pMatch score of each of said plurality of plurality of presentations; and

selecting one of said plurality of presentations based on best pMatch score.

11. The system of claim 1, wherein said term-pair matching element comprises means for matching Term1(fF1, dF1, pF1, kF1) and Term2(fF2, dF2, pF2, kF2), wherein said matching comprises:

comparing Term1 and Term2 based on equality, and assigning C11 a value of 0;

comparing Term1 and Term2 based on a hierarchical distance (C12) between Term1 node and Term2 node with respect to a hierarchical tree of terms;

comparing Term1 and Term2 based on a network distance (C13) between Term1 node and Term2 node with respect to a network of terms;

computing C1 as minimum of C1, C12, and C13;

computing C2 as |fF1−

fF2|;

computing C3 as |dF1−

dF2|;

computing C4 based on a table lookup using pF1 and pF2;

computing C5 based on a table lookup using kF1 and kF2; and

computing MatchScore as W1*C1+W2*C2+W3*C3+W4*C4+W5*C5.

12. The system of claim 1, wherein said term-set matching element comprises means for matching Term-Set1 and Term-Set2, wherein Term-Set1 is a set of plurality of set1 terms and Term-Set2 is a set of plurality of set2 terms, wherein said matching comprises:

computing minimum (minMatchScore) of a plurality of match scores, wherein said plurality of match scores is based on said term-pair matching element between a term of said set1 term and each term of said set2 term; and

computing match score MS as sum of minMatchScore associated with each term of said set1 term.

13. The system of claim 1, wherein said key term identification element comprises means for determining a plurality of key terms from a plurality of terms, wherein said identification comprises:

computing distance between a pair of terms in said plurality of terms based on said term-pair matching element;

finding a plurality of significant terms from said plurality of terms based on sum of minimum distance between each of said plurality of terms and a term of said plurality of significant terms wherein said distance measurement is based on said term-pair matching element; and

making said plurality of significant terms a part of said plurality of key terms.

14. The system of claim 1, wherein said slide annotation element comprises means for annotating a slide containing a text portion with respect to a dictionary to determine an annotation, TAnnotation, associated with said slide, wherein said annotating comprises:

analyzing said text portion based on said dictionary to extract a set Tst={T1, T2, . . . , Tn} of terms;

computing distance Dij as <

Dij , Dij2>

between term Ti in Tst and term Tj in Tst, wherein said computing comprises;

comparing Term1 and Term2 based on equality, and assigning C11 a value of 0;

comparing Term1 and Term2 based on a hierarchical distance (C12) between Term1 node and Term2 node with respect to a hierarchical tree of terms;

comparing Term1 and Term2 based on a network distance (C13) between Term1 node and Term2 node with respect to a network of terms;

computing Dij1 as minimum of C11, C12, and C13; and

computing Dij2 as the distance between Ti and Tj as per said text portion;

clustering Tst into {C1, C2, . . . , Ci, . . . , Ck} based on Dij;

checking size of Ci with respect to a pre-defined threshold;

selecting Ci cluster representative as At based on Dij between cluster elements and based on said key term identification element;

making At a part of TAnnotation; and

computing a plurality of factors associated with a term Tim part of At, wherein said computing comprises;

computing fF of said plurality of factors based on the number of occurrences of Tim in Tst;

computing dF of said plurality of factors based on size of a cluster containing Tim;

computing pF based on D factor value, S factor value, and L factor value associated with each term in a cluster containing Tim; and

computing kF based on T factor value, R factor value, and M factor value associated with each term in a cluster containing Tim.

15. The system of claim 14, wherein said slide annotation element further comprises means for annotating a slide containing a text content embedded in a graphics portion with respect to a dictionary to determine an annotation, GAnnotation, associated with said slide, wherein said annotating comprises:

analyzing said text content based on said dictionary to extract a set Tst={T1, T2, . . . , Tn} of terms;

computing distance Dij as <

Dij1, Dij2>

between term Ti in Tst and term Tj in Tst, wherein said computing comprises;

comparing Term1 and Term2 based on equality, and assigning C11 a value of 0;

comparing Term1 and Term2 based on a hierarchical distance (C12) between Term1 node and Term2 node with respect to a hierarchical tree of terms;

comparing Term1 and Term2 based on a network distance (C13) between Term1 node and Term2 node with respect to a network of terms;

computing Dij1 as minimum of C11, C12, and C13; and

computing Dij2 as the distance between Ti and Tj as per said text portion;

clustering Tst into {C1, C2, . . . , Ci, . . . , Ck} based on Dij;

checking of size Ci with respect to a pre-defined threshold;

selecting Ci cluster representative as At based on Dij between cluster elements and based on said key term identification element;

making At a part of GAnnotation; and

computing a plurality of factors associated with a term Tim part of At, wherein said computing comprises;

computing fF of said plurality of factors based on the number of occurrences of Tim in Tst;

computing dF of said plurality of factors based on size of a cluster containing Tim;

computing pF based on D factor value, S factor value, and L factor value associated with each term in a cluster containing Tim; and

computing kF based on T factor value, R factor value, and M factor value associated with each term in a cluster containing Tim.

16. The system of claim 1, wherein said presentation theme generation element comprises means for determining a theme of a presentation containing a plurality of slides, wherein said determining comprises:

determining TAnnotation and GAnnotation associated with each of said plurality of slides;

combining a plurality of annotations associated with said plurality of slides to determine TAGAs;

clustering TAGAs based on said term-pair matching element into a plurality of clusters;

finding a small number of clusters from said plurality of clusters such that sum of size of said small number of clusters is close to size of TAGAs;

determining a cluster representative of each of said small number of clusters based on said key term identification element; and

making said cluster representative of each of said small number of clusters a part of said theme.

17. The system of claim 1, wherein said template generation element comprises means for determining a plurality of templates based on a plurality of annotated presentations, wherein said determining comprises:

clustering said plurality annotated presentations into {C1, C2, . . . , Ci, . . . , Ck} based on a plurality of presentation themes associated with said plurality of annotated presentations;

checking size of Ci with respect to a pre-defined threshold;

determining a plurality of cluster presentations part of Ci;

finding a cluster representative of Ci based on a plurality of cluster presentation themes associated with said plurality of cluster presentations and based on said key term identification element;

determining an anchor template based on distance measure between each of said plurality of cluster presentations and said cluster representative;

making said cluster representative as template theme of said anchor template;

selecting a plurality of anchor slides from said plurality of cluster presentations with respect to a slide of said anchor template based on a plurality of annotations associated with said plurality of cluster presentations and a plurality of annotations associated with said slide of said anchor template; and

making said plurality of anchor slides a part of said anchor template.