Granted Patent B2

US 12,608,357 · App. 18/412,686 · Granted Apr 21, 2026

Inventors: Roni Ben-Aharon (Tel Aviv, IL); Nadav Abrahami (Tel Aviv, IL); Uri Dromy (Tel Aviv, IL); Barak Igal (Kfar Yehoshua, IL); Daphna Ofek (Ginaton, IL)

Assignee: Wix.com Ltd.

G06F16/2246G06F3/00G06F3/0481G06F3/0486G06F8/24G06F16/9577G06F16/958G06F40/106G06F40/143G06F40/186G06F16/88G06F40/154G06F40/166

Abstract

Claims (43)

1 . A visual design system (VDS), the VDS comprising:

a processor;

a database to store at least one layout and an associated layout signature wherein said associated layout signature represents a hierarchical composition of semantic component types of its components;

a unit running on said processor determine a component set signature for an existing layout, to compare said component set signature with at least one associated layout signature of at least one layout from said database to find a set of candidate layouts for said existing layout;

wherein said unit comprises:

a signature extractor to extract said component set signature from said existing layout by performing at least one of: semantic type mapping of components of said existing layout, dividing sematic types according to visual properties of said components and generating signature elements according to component semantic types;

a page analyzer to analyze at least said semantic types and to generate said component set signature, said page analyzer comprising at least a layout splitter to divide said existing layout according to at least one of: geometrical considerations, semantic links and dynamic layout anchors;

a layout rater to determine a score for a candidate layout as a function of at least a layout rater learning system, trained on layouts selected and applied by users of said VDS, layouts tested by users of said VDS and discarded, feedback from users of said VDS and actual page viewing/rating information of implemented layouts, wherein said layout rater recalculates said score according to said learning system; and

a ranker to rank candidate layouts according to said qulity score to produce a final set of candidate layouts;

which are visually diverse and semantically similar to said existing layout; and

an experiment system to create and run at least one experiment using at least one of: A/B and multivariate testing on said final set of candidate layouts to gather information on user preferred layouts to present to said user; and

an adapter and applier to adapt at least attributes, component types and content of said existing layout to a user selected preferred quality layout.

2 . The system according to claim 1 and wherein said experiment system comprises:

an experiment manger to coordinate said at least one experiment;

an experiment creator to receive said final set of candidate layouts and to present said final set of candidate layouts to said user for selection as part of said at least one experiment; and

an experiment analyzer to analyze user selected layouts to provide said information on user preferred layouts for said VDS.

3 . The system according to claim 1 wherein said at least one experiment is according to at least one of: a visual quality of said user selected layouts, an explicit user metric, an implicit user metric and a system use metric.

4 . The system according to claim 1 and wherein said page analyzer comprises:

at least one of:

a component splitter to split at least one component of said existing layout according to the content of said at least one component;

a component filter to filter out unsuitable components for said signature extractor; a component merger to unite at least two components for layout processing and signature extraction by said signature extractor; and

a container handler to reduce the amount of different component set signatures for said existing layout by performing at least one of: using hierarchical signatures, container flattening, container flattening and reconstruction, single component container replacement; dominant type selection and recursive implementation.

5 . The system according to claim 1 and wherein said VDS is a website building system.

6 . A method for a visual design system (VDS), the method comprising:

storing in a database, at least one layout and an associated layout signature wherein said associated layout signature represents a hierarchical composition of semantic component types of its components

determining a component set signature for an existing layout wherein said determining a component set signature comprises: extracting said component set signature from said existing layout by performing at least one of: semantic type mapping of components of said existing layout, dividing sematic types according to visual properties of said components and generating signature elements according to component semantic types;

analyzing at least said semantic types and generating said component set signature, said analyzing comprising at least dividing said existing layout according to at least one of: geometrical considerations, semantic links and dynamic layout anchors; comparing said component set signature with at least one associated layout signature of at least one layout from said database to find a set of candidate layouts for said existing layout;

determining a score for a candidate layout as a function of at least a layout rater learning system trained on layouts selected and applied by users of said VDS, layouts tested by users of said VDS and discarded, feedback from users of said VDS and actual page viewing/rating information of implemented layouts, wherein said layout rater recalculates said score according to said learning system; and

ranking candidate layouts according to said score to produce a final set of candidate layouts which are visually diverse and semantically similar to said existing layout;

creating and running the results of at least one experiment using at least one of: A/B and multivariate testing on said final set of candidate layouts to gather information on user preferred layouts to present to said user; and

adapting said existing layout to a user selected preferred layout.

7 . The method according to claim 6 wherein said running at least one of: A/B and multivariate testing comprises:

coordinating said at least one experiment;

receiving said final set of candidate layouts and presenting said final set of candidate layouts to said user for selection as part of said at least one experiment; and

analyzing user selected layouts to provide said information on preferred layouts for said VDS.

8 . The method according to claim 6 wherein said at least one experiment is according to at least one of: a visual quality of said user selected layouts, an explicit user metric, an implicit user metric and a system use metric.

9 . The method according to claim 6 and wherein said analyzing at least said semantic types comprises:

at least one of:

splitting at least one component of said existing layout according to the content of said at least one component;

filtering out unsuitable components for said extracting; and

uniting at least two components for layout processing and signature extraction by said extracting; and

reducing the amount of different component set signatures for said existing layout by performing at least one of: using hierarchical signatures, container flattening, container flattening and reconstruction, single component container replacement; dominant type selection and recursive implementation.

10 . The method according to claim 6 wherein said VDS is a website building system.

Assignments (2)

Continuity (20)

References Cited (73)