Ebay, Incorporated

Paypal, Inc.

Shopping.com

John A Mount

Badrul M Sarwar

Amir Ashkenazi

Daniel Louis Klein

Teg Grenager

A method and a system to automatically segment text based on header tokens is described. A relevance value and an irrelevance value are determined for each token in a description, assuming no tokens are left out of computations. The irrelevance value is based on occurrences of a token in a sample set of descriptions. The relevance value is an estimated probability of relevance based on the header of the description being segmented.

Header-token driven automatic text segmentation

A method and a system to organize a data set into groups of data subsets in multiple passes using different parameters and to automatically name the groups is disclosed. For example, a data set is retrieved in accordance with a search query submitted by a user. The data set is organized into clusters based on a statistic(s) of the data set. The data set is then organized into groups of data subsets based on an attribute(s) indicated by the data set. Each of the groups are automatically named based on a property shared by data units of the group. The name(s) of a group may be mined from the data units of the group, retrieved from a structure that maps to attribute values indicated by the data units of the group, etc.

Multi-pass data organization and automatic naming

Systems and methods for using automated translation and other statistical methods to convert a classifier in one language to another language are provided. In one embodiment of the invention, a method preferably includes marking target language examples of passages of text in order to obtain an initial classifier in the target language, re-classifying a plurality of target language examples in the initial classifier, and then questioning&#x2014;i.e., determining the validity&#x2014;of the marking used to obtain the initial classifier. Preferably, the questioning is based on the re-classifying. Then, preferably following the questioning, the method isolates a high-quality set of target examples based on the results of the questioning. Finally, the method uses the high-quality set of target examples to prepare a high-quality classifier in the target language.

Growing community of inventors

San Francisco, CA, United States of America

John A Mount