Hewlett-packard Development Company, L.P.

Ildus Ahmadullin

Niranjan Damera Venkata

Jan Philip Allebach

Niranjan Damera-Venkata

Example embodiments relate to automated document composition using quality distributions. In example embodiments, each image in a document is processed by determining quality scores that describe image distortion for potential aspect ratios, determining height deviations for the aspect ratios, and generating quality distributions using the height deviations and quality scores. At this stage, a page template is applied to a subset of elements from the document to render a page of the document, and a page score is determined for the page based on, at least, the quality distribution of an image, where the page template is adjusted to include a selected aspect ratio that maximizes the page score of the page.

Quality distributions for automated document

A probabilistic content layout model generates pages. Each of a number of compositions flows over multiple of the pages, and one or more of the pages each has multiple breakpoints. Each breakpoint is defined in relation to a given composition, such that the breakpoint breaks the given composition on the page that includes the breakpoint and such that the given composition continues on, a subsequent page.

Hierarchical probabilistic document model based document composition

In one implementation, a document aesthetics evaluation system accesses a plurality of content layout descriptors associated with a document template and a plurality of aesthetics measures. Each aesthetics measure from the plurality of aesthetics measures being assigned by a designer to a content layout descriptor from the plurality of content layout descriptors. The document aesthetics evaluation system also defines a content layout model for the document template based on the plurality of content layout descriptors and the plurality of aesthetics measures.

Document aesthetics evaluation

A method of presenting a plurality of content includes receiving measurements of similarity distances between content of the plurality of content; mapping the content to a two-dimensional surface based on the similarity distances; and partitioning the content of the two-dimensional surface into k levels of a similarity pyramid, wherein each of the k levels includes 4^k&#x2212;1 cells, and wherein each content of the plurality of content is separately mapped to one of the cells on one of the k levels of the similarity pyramid.

Growing community of inventors

Palo Alto, CA, United States of America

Ildus Ahmadullin