Browse Prior Art Database

A Method and System for Dynamically Synthesizing Suggestions for Search Queries by Manipulating Query Tokens

IP.com Disclosure Number: IPCOM000242950D
Publication Date: 2015-Sep-02
Document File: 3 page(s) / 35K

Publishing Venue

The IP.com Prior Art Database

Related People

Zhongqiang Chen: INVENTOR [+5]

Abstract

A method and system is disclosed for dynamically synthesizing suggestions for search queries by manipulating query tokens. The method and system receives a query from a user and treats it as a sequence of tokens to select a subset of query tokens from an original query and construct a new query called “sub query”. Further, the method and system generates a set of sub queries by choosing different subset of tokens from the original query using language models and user behavior. Subsequently, the set of sub queries are processed as normal queries to fetch suggestions from database.

This text was extracted from a Microsoft Word document.
This is the abbreviated version, containing approximately 41% of the total text.

A Method and System for Dynamically Synthesizing Suggestions for Search Queries by Manipulating Query Tokens

Abstract

A method and system is disclosed for dynamically synthesizing suggestions for search queries by manipulating query tokens.  The method and system receives a query from a user and treats it as a sequence of tokens to select a subset of query tokens from an original query and construct a new query called “sub query”.  Further, the method and system generates a set of sub queries by choosing different subset of tokens from the original query using language models and user behavior.  Subsequently, the set of sub queries are processed as normal queries to fetch suggestions from database.

Description

Disclosed is a method and system for dynamically synthesizing suggestions for search queries by manipulating query tokens.  The method and system receives a query from a user and treat it as a sequence of tokens to select a subset of query tokens from an original query and construct a new query called “sub query”.  Further, the method and system generates a set of sub queries by choosing different subset of tokens from the original query using language models and user behavior.  Subsequently, the set of sub queries are processed as normal queries to fetch suggestions from database.  The fetched suggestions are ranked according to the characteristics of the suggestions and correlation between the sub queries and the original query to provide suggestion with highest rank as a final result.

The method of generating set of sub queries from the original query by using language models and user behavior is processed by considering the query entered by the user as groups of concepts.  The query entered by the user usually consists of one or more concepts or ideas which are interconnected together to represent a search query.  The concepts or ideas in each query are described with one or more words which are called as tokens.  Here, the words used to present the same idea or concept tends to be contiguous and the similarity between two words is inversely proportional to their distance.  In some cases, the query entered by the user can be incomplete and vague by missing or dropping some words from the query.  Further, by considering the above mentioned observation from the language models and users’ behavior, the method and system dynamically provides suggestion to topics that are not related to each other using traditional information retrieval system.  Thus, the method used to dynamically synthesize suggestions for these queries can be processed using two subsystems; they are offline subsystem and online subsystem.

The offline subsystem is used to build list of tokens that can appear to be as the beginning or ending word of the concept or idea by mining the database of the potential suggestions.  While the online subsystem is used to generate suggestions for the user queries that are unable to get search assistance from...