Cost-benefit approach to automatically composing answers to questions by extracting information from large unstructured corpora

A computer implemented method to normalize a database, comprising the following computer executable acts:

• automatically forming a set of queries from a question posed by a user;
• and
• performing a cost-benefit analysis on the set of queries to generate a query subset, the cost benefit analysis enables users to assess or select various parameters that influence a utility model, the utility model processes a user setting for a cost, the utility model allows a user to assess a parameter v, indicating a dollar value of receiving a correct answer to a question, and where a parameter c represents a cost of each query rewrite submitted to a search engine.

A computer implemented question-answering system, comprising the following computer executable components:

• a rewriting component that receives a user query and automatically formulates a set of queries;
• and
• a cost-benefit component to reduce the set of queries based upon an analysis of expected gains in accuracy of an answer in view of associated costs for additional queries, the cost-benefit component enables users to assess or select various parameters that influence a utility model, the utility model processes a user setting for a cost, the utility model allows a user to assess a parameter v, indicating a dollar value of receiving a correct answer to a question, and where a parameter c represents a cost of each query rewrite submitted to a search engine.

A computer implemented normalization system, comprising the following computer implemented components:

• an interface component that receives data corresponding to a heterogeneous knowledge base;
• a normalization component that applies a model that predicts accuracy or quality of results in conjunction with a utility model to provide a regularized understanding of the value of performing different information extraction actions from the knowledge base;
• and a preference component that enables users to assess or select various parameters that influence the utility model, the preference component processes at least one of a user setting for a cost, a value, and a language preference, the preference component includes a model where a user assesses a parameter v, indicating a dollar value of receiving a correct answer to a question, and where a parameter c represents a cost of each cjuery rewrite submitted to a search engine.