Preventing biased queries by using a dictionary of cause and effect terms

Abstract

A method, computer system, and computer program product for eliminating confirmation bias in a user search query is provided. The present invention may include receiving a user-entered search query from an open ended-search tool. The invention may also include identifying a field associated with the received user-entered search query. The invention may further include creating a dictionary based on the identified field. The invention may also include determining the received user-entered search query relates to an effect by parsing the created dictionary. The invention may further include identifying one or more causes associated with the effect by parsing the created dictionary. The invention may also include generating a plurality of suggested search queries based on the one or more identified causes.

1 . A processor-implemented method for eliminating confirmation bias in a user search query, the method comprising:
 receiving, by a processor, a user-entered search query from an Internet search engine, wherein the user-entered search query includes a string of natural language text, one or more integers, and one or more keywords;   identifying a field associated with the received user-entered search query, wherein the field is an area most closely related to the user-entered search query;   creating a dictionary based on the identified field, wherein creating the dictionary further comprises:
 analyzing one or more documents with a plurality of factual information in the identified field for one or more if/then statements using one or more natural language processing techniques, wherein the one or more documents are a public domain article, a legislative bill, and a medical journal; 
 identifying one or more cause and effect terms within the one or more if/then statements; and 
 labeling the identified one or more cause and effect terms as either a cause or an effect; 
   determining the received user-entered search query relates to an effect by parsing the created dictionary using one or more natural language processing techniques;   identifying all of a plurality of causes associated with the effect by parsing the created dictionary;   generating one or more suggested search queries that each incorporate at least one cause within the plurality of causes based on the one or more identified causes;   determining a search query within the one or more suggested search queries with a lowest confirmation bias based on one or more search terms within the determined search query; and   executing the search query in the internet search engine.