Amazon Technologies, Inc.

Adam Callahan Sanders

Thomas Jay Hoover

Stefan Schroedl

Srikanth Thirumalai

Scott Allen Mongrain

Logan Luyet Dillard

David Erdmann

Rahul H Bhagat

Grant Michael Emery

Max Lee Kanter

Shashank Gupta

Brian Cameros

Benjamin Thomas Cohen

Liezel Marie Barraca

Michael David Quinn

Yongwen Liang

Eli Glen Rosofsky

Yong Gang Cao

Laxmikant A Agrawal

Disclosed are various embodiments for automating the generation of a visualization region for use in visually conveying concepts associated with products or services in an electronic marketplace. In response to a search query, a plurality of concepts that may beneficial in further refining the search query may be determined and may be represented by a plurality of representative items available in the electronic marketplace. Visual identifiers associated with the representative items may be used in the generation of a visualization region for use in further refining the search query to a particular concept.

Visualization region search refinement

Various embodiments of a system and method for highly scalable data clustering are described. Embodiments may include generating contexts of information from item-level data for multiple items; each context may include tokens that each represents an aggregate characteristic of items associated with that context. Embodiments may also include comparing the tokens of different contexts to determine measures of similarity between the contexts. Embodiments may also include grouping at least some of the contexts into clusters with other contexts based on the determined measures of similarity. Embodiments may include, subsequent to detecting a first context and a second context as being members of a common cluster, correcting item level-data of an item associated with the second context based on item-level data of an item associated with the first context.

System and method for highly scalable data clustering

Method and systems personalizing the search results presented to a user in an electronic marketplace. A search query is received and a set of search results related to the search query are obtained. Individual entries of the search results are associated with a category type. The search results are sorted in a number of category groups according to the category type. An entry from each category group is selected and scores are determined for the selected entries. A highest scored entry is assigned to a highest available position in a resultant search result set. The scores for the entries that remain unassigned are increased. For the category group corresponding to the assigned entry, another entry is selected and scored. A highest entry score is assigned the next highest position. The process may continue recursively until all, or a threshold number, of entries have been assigned to the resultant list.

Increasing the relevancy of search results across categories

The present technology may identify item category affinities by identifying a plurality of classifications of an item. An accuracy of the plurality of classifications relative to one another for the item may be identified. A category affinity of the item may be determined based on the accuracy of the plurality of classifications relative to one another.

Video classification

A technology is described for making a recommendation for an item. An example method may include identifying a user interest in an item and identifying individuals that may have an interest in the item. The individuals may be presented to a user via a recommendation interface control that allows a name of the item to be added to a recommendation list for an individual selected from the individuals. An individual selected from the individual list to receive the recommendation may be identified and the name of the item may be added to the recommendation list for the individual.

Item recommendation

Method and systems personalizing the search results presented to a user in an electronic marketplace. Data for a search query in an electronic marketplace is received. The query is executed and search results are received. Scores are calculated each item in the search result based at least in part on at least one of global navigational data and personal navigational data. Each item is allocated a position in an ordered search result list based at least in part on the scores. When at least some subset of the items in the search result list have been scored and allocated a position in the ordered search result list, the ordered search result list may be presented to the user.

A method is described for making a recommendation for an item. The method may include the operation of presenting a recommendation interface control to a user on a client device after the user has consumed an item. Using the recommendation interface control, the user then selects an individual and adds the name of the item the user consumed to the selected individual's recommendation list.

Disclosed are various embodiments for providing a product comparison application utilizing crowdsourcing. A network page is rendered providing a grid comprising a plurality of products, product attributes, and/or attribute data. Additions and/or modifications of attributes, products, and/or attribute data are assigned a score to determine which addition or modification to display and/or associate with a product catalog. The grid provides a view for comparison of products.

Growing community of inventors

Seattle, WA, United States of America

Adam Callahan Sanders