Ebay, Incorporated

Yahoo! Inc.

Michael Mathieson

Parashuram Kulkarni

Jon Degenhardt

Abdelhalim Abbas

Jingzhou Hua

Ramakrishnan Natarajan

Sanjay Pundlkrao Ghatare

Muhammad Faisal Rehman

Deepa Joshi

Vipul C Dalal

Sameep Navin Solanki

Kalpana Ravinarayanan

Ted Tao Yuan

Michael Smith

Michael J Wilkens

Various embodiments include systems and methods for search result ranking using machine learning. A goal model can be created using machine learning. Responsive to a search query, a plurality of data factors can be inputted into the goal model to create a model output. Search results can be presented to a user based on the model output.

Search result ranking using machine learning

A system and method for assessing excessive accessory listings in search results includes a processor-implemented textual mining module that parses a data field of a document and generates at least one token from the data field. A processor-implemented scoring module calculates a score for the at least one token, with the at least one token score representing a likelihood that the at least one token belongs to one of two binary classifications. The processor-implemented scoring module also calculates a score for the document based on the at least one token score, with the document score representing a probability of the document being in one of the two binary classifications. A processor-implemented decision tree module inputs the document score and document attribute values into a decision tree and generates an output representing a refined score based on the document score and at least one of the document attribute values.

Methods and systems for assessing excessive accessory listings in search results

Described herein are methods and systems for promoting item listings that satisfy a query based on the item listings being assigned to certain categories that have, based on historical click data, exhibited high demand characteristics for the query. Consistent with some embodiments, a certain number of leaf-level categories are identified based on demand data for those categories, and the item listings assigned to those categories are promoted through a weighting factor derived in part based on the click probability score associated with the category. In some embodiments, certain sub-categories may be selected when the demand associated with the child categories of the sub-category is well balanced.

Method and system for ranking search results based on categories

A search query that includes a first spelling of a term is received. A second spelling of the term is generated. A first set of items that are associated with the first spelling of the term is accessed, and a second set of items that are associated with the second spelling of the term is accessed. A comparison is performed between the first set to the second set. A determination is made as to how to respond to the search query based, at least in part, on the comparison. The response may include a suggested search query that includes the second spelling of the term. The second spelling may be a suggested spelling correction or an alternative spelling. The response may include search results that are relevant to a search query that includes the first spelling of the term, the second spelling, or both the first and second spelling.

Discovering alternative spellings through co-occurrence

Growing community of inventors

Sunnyvale, CA, United States of America

Michael Mathieson