Automated self-service user support based on ontology analysis

• US 10,162,885 B2
• Filed: 05/29/2018
• Issued: 12/25/2018
• Est. Priority Date: 05/27/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Rozsa, Eniko I., Javahery, Homa, Yamamoto, Paul Alan, Boyle, Peter C.
• Primary Examiner(s): Shin, Seong Ah A
• Application Number:
  US15/990,889
• Publication Number:
  US 20180276293A1
• Time in Patent Office: 210 Days
• Field of Search: 704 4, 704 9, 704275, 707603, 707706, 707751
• CPC Class Codes: G06F 16/3338 Query expansion
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

A method for providing information to a user in response to a received user query. A natural language analysis generates substrings relevant to an input user query pertaining to a problem a user experiences with a product or service. An ontology analysis outputs: terms of an ontology matching the relevant generated substrings; and relationships between the terms. Elements of a perfect or complete query are captured and include information pertaining to the user'"'"'s problem. The input user query is refined based on the outputted terms and relationships between the terms. A search query is performed, based on the refined user query. The results of the search are provided to the user.

Claims

1. A method for providing, by a self-service user support software application for assisting customers having a product or service problem, information, received via a search conducted over an Internet computer network using a refined user query, to a user in response to a received input user query, said method comprising:

receiving, by a processor of a computer system, an input user query from a user interface component in a screen, said received input user query expressed in a free-form text format, said computer system comprising the screen, said input user query pertaining to a problem of the user which is a problem that the user experiences with a product or service;

said processor performing a natural language analysis to generate substrings relevant to the received input user query, wherein said performing the natural language analysis comprises extracting details from the user query, and wherein said performing the natural language analysis comprises identifying a language of text in the input user query, recognizing a misspelling of one word in the input user query, determining a canonical form of another word in the input user query, recognizing a term in the input user query pertaining to a technical support domain, and semantically recognizing an incident expressed in the input user query;

after said performing the natural language analysis, said processor performing an ontology analysis to output terms of an ontology of domain-specific information specific to a domain pertaining to products and to further output relationships between pairs of said terms, said outputted terms constrained to match the relevant substrings generated by said performing the natural language analysis;

said processor capturing, via an ontology model included in the ontology, elements of a perfect or complete query, wherein the elements of the perfect or complete query include information pertaining to the problem of the user;

during said performing the ontology analysis, said processor identifying multiple outputted terms that match one of the relevant substrings, requesting from the user a selection of one outputted term of the multiple outputted terms, and receiving from the user the selection of the one outputted term of the multiple outputted terms;

after said performing the ontology analysis, said processor performing a query analysis to analyze the input user query with respect to the outputted terms and relationships between the terms;

said processor refining the input user query based on the outputted terms and relationships between the terms;

said processor generating a search query based on the refined user query;

said processor initiating a search by sending the search query across the Internet computer network to a search engine configured to perform the search, based on the search query, via one or more databases;

said processor receiving from the search engine results of the search via the user interface component in the screen; and

after said performing the ontology analysis, said processor performing a query analysis to analyze the input user query with respect to the outputted terms and relationships between the terms.

2. The method of claim 1, wherein the details extracted from the user query include a type of the problem of the user, what the user was doing when the problem of the user occurred, an environment in which the problem of the user occurred, product components affected by the problem of the user, and conditions that have changed as a result of the problem of the user.

3. The method of claim 1, wherein the information pertaining to the problem of the user include:

what the problem of the user is, where the problem of the user occurs, an environment of the problem of the user, and what activity of the user led to the problem of the user.

4. The method of claim 1, wherein said capturing elements of a perfect or complete query is a use of the domain.

5. The method of claim 1, said method further comprising:

said processor soliciting feedback from the user concerning a relevancy of the results of the search.

6. The method of claim 1, wherein said performing the query analysis comprises:

determining that the outputted terms and relationships between the terms are not sufficiently specific to enable a search for requested information in the input user query after the input user query is refined based on the outputted terms and relationships between the terms;

presenting questions to the user pertinent to the outputted terms and relationships between the terms not being sufficiently specific;

receiving, from the user in response to the presented questions, clarification with respect to a specificity of the outputted terms and relationships between the terms; and

modifying the input user query, based on the received clarification, to enable the refined user query to be sufficiently specific to enable said search for requested information in the refined user query, wherein said refining the input user query comprises refining the modified user query.

7. The method of claim 1, wherein said performing the query analysis comprises:

determining that the input user query is not sufficiently complete due to the input user query not including sufficient information to enable an expert in a field pertinent to the input user query to respond to a request for information in the input user query;

presenting questions to the user pertinent to the input user query not being sufficiently complete;

receiving, from the user in response to the presented questions, information missing from the input user query; and

modifying the input user query, based on the received missing information, to make the input user query sufficiently complete to enable the expert in the field to respond to the request for information in the input user query, wherein said refining the input user query comprises refining the modified user query.

8. The method of claim 1, wherein the ontology is represented as a tree comprising nodes and edges connecting adjacent nodes, wherein each node is associated with a term of the ontology such that the nodes of the tree comprise the outputted terms that match the relevant substrings, and wherein each edge connecting adjacent nodes represents a relationship between the terms to which the adjacent nodes are respectively associated.

9. The method of claim 1, wherein said performing the ontology analysis comprises:

identifying multiple terms of the ontology that match a selected substring of the generated substrings;

in response to said identifying multiple terms, presenting questions to the user for assisting the user to select a term of the multiple terms;

after said presenting questions to the user, receiving from the user a selected term of the multiple terms that best matches the selected substring.

10. The method of claim 1, wherein said performing the ontology analysis comprises annotating a substring of the generated substrings with a term of the outputted terms.

11. The method of claim 1, wherein said performing the natural language analysis comprises assigning a linguistic category to different words of the input user query, wherein each linguistic category is specific to the word to which each linguistic category is assigned, and wherein generation of the substrings by performing the natural language analysis comprises utilizing the linguistic category assigned to the different words of the input user query.

12. The method of claim 9, wherein a first, second, and third linguistic category assigned to a first, second, and third word is a noun, a verb, and an adjective, respectively.

13. The method of claim 1, wherein said performing the natural language analysis comprises performing a morphological analysis to determine a single word consisting of a lemma that is specific to each word of a plurality of words of the user input query, and wherein the generated substrings comprise the lemma specific to each word of the plurality of words of the input user query.

14. The method of claim 1, wherein the generated substrings comprise a first substring that matches a concept of the ontology and attributes and relationships associated with the concept, wherein said performing the ontology analysis comprises annotating the first substring with the concept and with pillars of the concept, and wherein the pillars of the concept include situations, activities, products, and IT components of the concept that mirror requirements of a perfect query.