Utilizing semantic hierarchies to process free-form text

• US 10,402,435 B2
• Filed: 06/30/2015
• Issued: 09/03/2019
• Est. Priority Date: 06/30/2015
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Jain, Navendu, Dubey, Rajeev, Betz, Adam J., Haase, Bill A., Dao, Huy P. M.
• Primary Examiner(s): Mamillapalli, Pavan
• Application Number:
  US14/788,247
• Publication Number:
  US 20170004205A1
• Time in Patent Office: 1,526 Days
• Field of Search: 707723
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition

Abstract

User feedback may be analyzed with semantic hierarchies. In some instances, the user feedback includes free-form text. The user feedback may be mapped to one or more semantic hierarchies that include multiple levels of nodes, where each node corresponds to a class. Information of the one or more semantic hierarchies may be mapped to an ontology model. The mapped information of the ontology model may be used to identify an actionable item for the user feedback, such as a problem, suggestion, question, or other issue. Information regarding the actionable item may be made available to an individual for evaluation of the actionable item.

Claims

1. A system comprising:

one or more processors;

memory communicatively coupled to the one or more processors and storing data regarding a semantic hierarchy having one or more levels of nodes, with each node representing a class;

an input device communicatively coupled to the one or more processors and configured to receive data including free-form text;

mapping instructions stored in the memory and executable by the one or more processors to automatically map at least one word or phrase of the data to one or more levels of the semantic hierarchy and to map a class of the semantic hierarchy to a class of an ontology model;

feedback analysis instructions stored in the memory and executable by the one or more processors to use a mapping by the mapping instructions to tag words or phrases in the free-form text with one or more levels of the semantic hierarchy or respective classes of the ontology model and to match a grammar pattern of the tagged words or phrases to a grammar pattern associated with a concept of the ontology model comprising an actionable item in the received data, wherein the concept comprises at least one of a problem, suggestion, question, activity, and action that identifies one of an entity and a question and at least one of an associated state, condition, and symptom; and

feedback routing instructions stored in the memory and executable by the one or more processors to utilize the mapped at least one word or phrase of the data to send information regarding the analysis of the data by the feedback analysis instructions to a device associated with an individual for evaluation of the actionable item in the received data.

2. The system of claim 1, wherein the semantic hierarchy includes:

a first level having at least one node representing customer terms; and

a second level having at least one node representing at least one of morphological forms of the customer terms or technical terms for the customer terms.

3. The system of claim 1, wherein the actionable item includes at least one of a problem with a software or hardware entity, a suggestion for improvement of the software or hardware entity, or a question regarding a software or hardware entity.

4. The system of claim 1, wherein the data includes feedback that is provided via an operating system, feedback from an online source, feedback from a conversation with an agent, feedback provided via a customer support portal, feedback from an electronic message, or feedback from an evaluation form.

5. The system of claim 1, wherein the mapping instructions are configured to:

map the at least one word or phrase of the data to the semantic hierarchy by mapping the at least one word or phrase to a node at a first level in the semantic hierarchy and determining a node at a second level of the semantic hierarchy that corresponds to the node at the first level of the semantic hierarchy; and

map the node at the first or second level of the semantic hierarchy to a class of the ontology model.

6. The system of claim 1, wherein:

the feedback analysis instructions are configured to determine a type of the actionable item based at least in part on the mapped at least one word or phrase of the data; and

the feedback routing instructions are configured to send information to the device associated with the individual regarding the type of the actionable item.

7. A method comprising:

generating, by a computing device, an entity hierarchy having nodes that represent classes for an entity and an entity state hierarchy having nodes that represent classes for a state of the entity, each of the entity hierarchy and the entity state hierarchy including different levels of nodes;

receiving, by the computing device, user feedback including free-form text;

automatically mapping, by the computing device, a first word or phrase of the user feedback to the entity hierarchy and mapping a second word or phrase of the user feedback to the entity state hierarchy, and mapping a class of the semantic hierarchy is mapped to a class of an ontology model;

determining, by the computing device, that the user feedback relates to an actionable item for a software or hardware entity by tagging words or phrases in the free-form text with one or more levels of the entity hierarchy or respective classes of the ontology model and matching a grammar pattern of the tagged words or phrases to a grammar pattern of the received user feedback associated with a concept of the ontology model comprising an actionable item in the received user feedback, wherein the concept comprises at least one of a problem, suggestion, question, activity, and action that identifies one of an entity and a question and at least one of an associated state, condition, and symptom;

determining, by the computing device, an individual to inform of the user feedback based on at least one of the mapped first word or phrase to the entity hierarchy or the mapped second word or phrase node to the entity state hierarchy; and

making available information regarding the user feedback to a device associated with the individual to evaluate the actionable item in the received user feedback.

8. The method of claim 7, wherein at least one of the entity hierarchy or the entity state hierarchy includes a first level having at least one node representing customer terms, and a second level having at least one node representing at least one of morphological forms of the customer terms or technical terms for the customer terms.

9. The method of claim 7, wherein the entity hierarchy includes nodes that represent words or phrases that relate to the software or hardware entity and the entity state hierarchy includes nodes that represent words or phrases that relate to at least one of a problem for the software or hardware entity, a suggestion for the software or hardware entity, or a question regarding the software or hardware entity.

10. The method of claim 7, wherein:

the mapping includes mapping the first word or phrase to a node in the entity hierarchy and mapping the second word or phrase of the user feedback to a node in the entity state hierarchy; and

the method further comprises;

determining a type of the actionable item based on at least one of the node to which the first word or phrase is mapped in the entity hierarchy or the node to which the second word or phrase is mapped in the entity state hierarchy; and

sending information to the device associated with the individual regarding the type of the actionable item.

11. The method of claim 7, wherein the actionable item includes a problem with the software or hardware entity, a suggestion for improvement of the software or hardware entity, or a question regarding the software or hardware entity.

12. The method of claim 11, wherein:

the mapping includes;

mapping the first word or phrase of the user feedback to a node at a lowest level in the entity hierarchy and traversing the entity hierarchy to a node at the higher level of the entity hierarchy; and

mapping the second word or phrase to a node at a lowest level in the entity state hierarchy and traversing the entity state hierarchy to a node at the higher level of the entity state hierarchy; and

the method further comprises;

mapping the node at the higher level of the entity hierarchy to a first class of the ontology model, and mapping the node at the higher level of the entity state hierarchy to a second class of the ontology model;

tagging the first word or phrase of the user feedback with the first class of the ontology model, and tagging the second word or phrase of the user feedback with the second class of the ontology model; and

determining that the user feedback relates to at least one of the problem, the suggestion, or the question for the software or hardware entity based at least in part on the grammar pattern of the tagged user feedback.

13. The method of claim 7, wherein the user feedback includes one or more sentences in a natural language format.

14. The method of claim 7, further comprising:

ranking the actionable item associated with the user feedback with respect to actionable items of other user feedback based at least in part on occurrences of the actionable item within a plurality of pieces of user feedback; and

making available information regarding the ranking to the device associated with the individual.

15. One or more computer-readable media storing computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform acts comprising:

receiving a plurality of documents, individual ones of the plurality of documents including free-form user text;

analyzing individual ones of the plurality of documents based at least in part on one or more semantic hierarchies that each have different levels of nodes for different classes of words or phrases;

identifying actionable items for at least some of the plurality of documents based at least in part on the analysis by tagging words or phrases in the free-form user text with one or more levels of a semantic hierarchy or respective classes of an ontology model, wherein a class of the semantic hierarchy is mapped to a class of the ontology model, and matching a grammar pattern of the tagged words or phrases to a grammar pattern of the free-form user text associated with a concept of the ontology model comprising an actionable item in the free-form user text, individual ones of the actionable items including at least one of a problem, suggestion, question, activity, and action that identifies one of an entity and a question and at least one of an associated state, condition, and symptom;

generating information regarding at least some of the actionable items; and

making the information available to evaluate at least some of the actionable items in the free-form user text.

16. The one or more computer-readable media of claim 15, wherein the identifying the actionable items includes identifying, for individual ones of at least some of the plurality of documents, a node in the one or more semantic hierarchies that is identified from the analysis that relates to a problem, suggestion, or question class.

17. The one or more computer-readable media of claim 15, wherein the generating the information includes:

generating information that identifies a particular actionable item that occurs most frequently, within the identified actionable items, for a particular software or hardware entity.

18. The one or more computer-readable media of claim 15, wherein the generating the information includes:

ranking at least some of the actionable items based at least in part on a number of occurrences of respective actionable items within the identified actionable items; and

generating information regarding the ranking of at least some of the actionable items.

19. The one or more computer-readable media of claim 15, wherein the information indicates a class of at least one of the actionable items.