Amazon Technologies, Inc.

Dongmei Jia

Sherif M Yacoub

Bernhard Wolkerstorfer

Parth Gupta

Roberto Fernandez Galan

Nicholas Alan Tostenrude

Dan Bu

Haode Yang

Stephen Kang

Gerald J Strode

Techniques are generally described for modeling seasonal relevance for online search. A first query or item comprising text is received. A language model may be used to generate a plurality of token embeddings representing the text. In some cases, the first neural network may predict a seasonal relevance vector for the first query or item based on the plurality of token embeddings, the seasonal relevance vector being predicted based at least in part on a similarity between the text of the first query or item and at least one other query/item for which monthly concentration data is available. The seasonal relevance vector, including a respective seasonal relevance score for each month of a year, may be output.

Modeling seasonal relevance for online search

Techniques are described for identifying potentially interesting portions of a content item to be provided as a preview of the content item for prospective purchasers, based on previously collected data associated with the content item. Portions of a content item may be identified as potentially interesting based on a number of annotations (e.g., highlights, bookmarks, notes, and shares) previously made by viewers of a digital version of the content item. Potentially interesting portions may also include portions which prior viewers spent more time viewing, portions related to identified interests of the potential buyer, portions that are identified as separable for particular categories of content, and/or portions that have been previously identified as associated with elements of the content such as character, plot, and/or keywords.

Previewing content based on detected customer activities

Growing community of inventors

Bellevue, WA, United States of America

Dongmei Jia