Interactive segment extraction in computer-human interactive learning

Abstract

A collection of data that is extremely large can be difficult to search and / or analyze. Relevance may be dramatically improved by automatically classifying queries and web pages in useful categories, and using these classification scores as relevance features. A thorough approach may require building a large number of classifiers, corresponding to the various types of information, activities, and products. Creation of classifiers and schematizers is provided on large data sets. Exercising the classifiers and schematizers on hundreds of millions of items may expose value that is inherent to the data by adding usable meta-data. Some aspects include active labeling exploration, automatic regularization and cold start, scaling with the number of items and the number of classifiers, active featuring, and segmentation and schematization.

Description

Background technique [0001] Collections of very large data can be difficult to search and / or analyze. For example, in the case of the web, a large portion of the data is unstructured and the values ​​are locked within the data itself. It is not enough to store the web page of the service provider. In order for the information to be useful, the information needs to be understood. A string of numbers that could be a model number, bank account, or phone number depending on the context. For example, in the context of a snowboard product, the string "length: 170, 175, 180cm" refers to 3 different snowboard lengths, not 1700 kilometers of skis. Interpreting data incorrectly can yield useless information. [0002] As an example, if a user enters the two words "mtor" and "stock" into an Internet search engine, and the results consist primarily of web pages related to the drug mTor, the search engine fails to recognize the search as a stock quote query. As another example, if a us...

AI Technical Summary

Problems solved by technology

Although Internet search engines often do a reasonable job of targeting title queries and documents, accuracy quickly drops to low levels because the information is not automatically understood by the search engine

Images