OBJECT SELECTION BASED ON NATURAL LANGUAGE QUERIES

• US 20120239682A1
• Filed: 02/22/2012
• Published: 09/20/2012
• Est. Priority Date: 03/15/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Wedeniwski, Sebastian
• Granted Patent: US 9,104,754 B2
• Time in Patent Office: Days
• US Class Current: 707/769
• CPC Class Codes: G06F 16/3334 Selection or weighting of terms from queries, including natural language queries
• CPC Class Codes: G06F 16/337 Profile generation, learning or modification
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 16/367 Ontology

Abstract

A method of selecting an object out of a group of objects. The object may belong to a category. The method may include receiving, object-related information indicative of properties of objects of the category by using network connected databases. The method may further include generating, an extended profile for at least one of the objects based on the received object-related information and object-specific information. The method may additionally include determining, a data structure of weighted keywords based on the natural language input using at least one out of the group consisting of a predefined dictionary, a predefined ontology, predefined system rules, and a predefined regular expression, and selecting, the object out of the group of objects based on a mapping of the extended profile and the data structure of weighted keywords resulting in a preference fit score for the selected object.

Claims

1. A method of selecting an object out of a group of objects belonging to a category, the method comprising:

receiving object-related information indicative of properties of objects of the category by using network connected databases;

generating an extended profile for at least one of the objects based on the received object-related information and object-specific information indicative of properties of the at least one of the objects, such that the extended profile comprises the object-specific information as well as object-related information that is related to the object-specific information, and wherein the generating the extended profile comprises using at least one out of the group consisting of a predefined dictionary, a predefined ontology, predefined system rules, and a predefined regular expression;

providing a user interface adapted for receiving natural language input;

determining a data structure of weighted keywords based on the natural language input using at least one out of the group consisting of the predefined dictionary, the predefined ontology, predefined system rules, and a predefined regular expression; and

selecting the object out of the group of objects based on a mapping of the extended profile and the data structure of weighted keywords resulting in a preference fit score for the selected object, wherein the mapping comprises using at least one out of the group consisting of an N-Gram distance function between data elements of the extended profile and data elements of the data structure of weighted keywords, thresholds specified in the extended profile, and weights of rules.

2. The method according to claim 1, wherein the generating the extended profile comprises relating elements of the object-related information to features of the object-specific information.

3. The method according to claim 2, wherein the generating the extended profile comprises building clusters of features of the object-specific information and elements of the object-related information.

4. The method according to claim 1, wherein the preference fit score is determined based on one out of the group consisting of a keyword factor related to the extended profile, a preference factor determined by a quadratic mean of weighted keywords, and a numeric preference factor defined through the user interface or a system of rules.

5. The method according to claim 1, wherein a weighting of at least one of the weighted keywords is modifiable using the user interface.

6. The method according to claim 1, wherein at least one of the values of the data structure is modifiable using the user interface.

7. The method according to claim 6, wherein the modification of the at least one value of the data structure is performed by using a slider with at least three marks, the slider being part of the user interface.

8. The method according to claim 1, wherein the determining the data structure of weighted keywords comprises analyzing a user profile.

9. The method according to claim 1, comprising displaying an identifier of the selected object together with the related preference fit score.

10. A selection engine for selecting an object out of a group of objects belonging to a category, the selection engine comprising:

a receiving unit for receiving object-related information indicative of properties of objects of the category by using network connected databases;

a generating unit for generating an extended profile for at least one of the objects based on the received object-related information and object-specific information indicative of properties of the at least one of the objects, such that the extended profile comprises the object-specific information as well as object-related information that is related to the object-specific information, and wherein the generating unit uses at least one out of the group consisting of a predefined dictionary, a predefined ontology, predefined system rules, and a predefined regular expression, when generating the extended profile;

a user interface for receiving natural language input;

a determining unit for determining a data structure of weighted keywords based on the natural language input using at least one out of the group consisting of a predefined dictionary, a predefined ontology, predefined system rules; and

a selecting unit for selecting the object out of the group of objects based on a mapping of the extended profile and the data structure of weighted keywords resulting in a preference fit score for the selected object, wherein the mapping comprises using at least one out of the group consisting of an N-Gram distance function between data elements of the extended profile and data elements of the data structure of weighted keywords, thresholds specified in the extended profile, and weights of rules.

11. The selection engine of claim 10, wherein the selection engine is implemented using a computer system.

12. A computer program product, provided on a computer usable medium, comprising a computer readable program for performing a method for selecting an object out of a group of objects belonging to a category when executed on a computer device, the method comprising:

receiving object-related information indicative of properties of objects of the category by using network connected databases;

generating an extended profile for at least one of the objects based on the received object-related information and object-specific information indicative of properties of the at least one of the objects, such that the extended profile comprises the object-specific information as well as object-related information that is related to the object-specific information, and wherein the generating the extended profile comprises using at least one out of the group consisting of a predefined dictionary, a predefined ontology, predefined system rules, and a predefined regular expression;

providing a user interface adapted for receiving natural language input;

determining a data structure of weighted keywords based on the natural language input using at least one out of the group consisting of the predefined dictionary, the predefined ontology, predefined system rules, and a predefined regular expression; and

selecting the object out of the group of objects based on a mapping of the extended profile and the data structure of weighted keywords resulting in a preference fit score for the selected object, wherein the mapping comprises using at least one out of the group consisting of an N-Gram distance function between data elements of the extended profile and data elements of the data structure of weighted keywords, thresholds specified in the extended profile, and weights of rules.