Ontology-driven conversational interface for data analysis

Patent metadata

Field	Value
Publication number	US-12019995-B2
Application number	US-202016920693-A
Country	US
Kind code	B2
Filing date	Jul 4, 2020
Priority date	Jul 4, 2020
Publication date	Jun 25, 2024
Grant date	Jun 25, 2024

Abstract

A computer-implemented method for generating an ontology-driven conversational interface includes generating an ontology from a description of a domain schema of a Data Analysis (DA) model, in which the DA model is a defined in terms of quantifiable, qualifying or categorical entities and their relationships as described by the domain schema. Conversational artifacts of a conversation space including a conversational pattern framework are generated by extracting DA-related intents, entities, and a dialog from the generated ontology for the conversational interface. A dialog logic table maps DA-related patterns to intents, extracted quantifiable, qualifying or categorical attributes to entities, and the dialog to user-prompts for one or more parameters in an identified DA pattern. The conversation space is integrated with at least one of an external data source or an analytics platform that stores and processes data.

First claim

What is claimed is:

1. A computer-implemented method for generating an ontology-driven conversational interface, the method comprising:
generating an ontology from a description of a domain schema of a Data Analysis (DA) model, in which the DA model is defined in terms of quantifiable, qualifying or categorical entities and their relationships as described by the domain schema;
generating conversational artifacts of a conversation space by extracting DA-related intents, entities, and elements in building a dialog from the generated ontology for the conversational interface;
mapping the extracted DA-related intents to DA patterns, the extracted entities to quantifiable and qualifying or categorical attributes and the dialog to agent-prompts for one or more parameters in an identified DA pattern of the DA patterns; and
integrating the conversation space with at least one external data source or an analytics platform that stores and processes data.

2. The computer-implemented method according to claim 1 , further comprising transmitting, by the conversational interface, a natural language response to a natural language query to access information in the external data source described by the DA model.

3. The computer-implemented method according to claim 1 , further comprising retrieving data and visualizations from the analytics platform by the conversation space.

4. The computer-implemented method according to claim 1 , further comprising providing meta-concepts as groupings of one or more of the quantifiable and qualifying or categorical attributes.

5. The computer-implemented method according to claim 1 , further comprising:
forming a dialog logic table by specifying one or more parameters associated with each intent; and
identifying each of the specified one or more parameters as optional or required.

6. The computer-implemented method according to claim 5 , wherein the dialog logic table is formed by specifying for each intent, natural language utterances of a user corresponding to the intent.

7. The computer-implemented method according to claim 1 , wherein the conversation space includes one or more training samples for each intent, and the method further comprises training the conversational interface by machine learning to learn a model to identify an intent in a user utterance.

8. The computer-implemented method according to claim 7 , wherein the training of the conversational interface is performed by a classification mechanism.

9. A computer-implemented method of generating a conversation space of a conversational interface for a Data Analysis (DA) application, the method comprising:
generating an ontology from a description of a domain schema of a DA model described in terms of DA-related quantifiable and qualifying or categorical attributes as entities and relationships between them;
annotating the ontology with semantic information from the DA model;
mapping of DA-related intents and entities extracted from the ontology to a dialog logic table, the dialog logic table including a quantifiable entity, a categorical attribute, a filter, and a relationship of mapped components;
generating one or more conversational artifacts of the conversation space from the ontology in terms of intents and entities and in building a dialog for the conversational interface; and
integrating the conversation space with at least one of an external data source or analytics platform to store and process data.

10. The computer-implemented method according to claim 9 , further comprising grouping of the entities from the ontology into one or more meta-concepts and a domain-dependent interpretation of each meta-concept.

11. The computer-implemented method according to claim 9 , further comprising:
forming an ontology graph from the generated ontology;
enhancing the ontology graph by adding one or more of new concepts, groupings, hierarchies, relationships discerned from a data-driven machine learning, a deep learning, an embedding based technique for a named entity recognition, or a link prediction; and
periodically enhancing the ontology graph and a subsequent process of generating the conversational artifacts as more data is consumed.

12. The computer-implemented method according to claim 9 , further comprising adding a set of generic terms to the conversational space as synonyms for entities in user utterances.

13. The computer-implemented method according to claim 9 , wherein generating the ontology further comprises obtaining a taxonomy or hierarchies from a domain description of the DA model in terms of parent-child relationships.

14. The computer-implemented method according to claim 9 , further comprising:
configuring a generic dialog structure for making a series of complex open requests for one or more DA-related requests including one or more of analytic queries, trend requests and comparison requests.

15. The computer-implemented method according to claim 9 , wherein a set of generic operations for one or more DA-related requests is configured for DA, and includes operations for drill down, roll up, and a pivot on a previous query.

16. The computer-implemented method according to claim 9 , further comprising integrating one or more DA-related requests and operations with a Natural Conversation Space, which includes a persistent context, natural conversation activities and conversation management actions.

17. An ontology-driven conversational interface of a conversation device, comprising:
an intent module configured to identify goals and actions from utterances received from a user as one or more intents;
an entity module configured to identify information associated with a user intent as one or more entities;
a dialog module configured to provide a response to the user based on the identified one or more intents, the identified one or more entities, and a context of a conversation;
a processor configured to:
generate an ontology from a description of a domain schema of a Data Analysis (DA) model; and
generate one or more conversational artifacts of a conversation space by extracting DA-related intents, entities and elements in building a dialog from the generated ontology for the conversational interface; and

an analytics platform configured to store and process data that the conversation space returns to the user, and to provide one or more responses to user queries generated through structured queries in a form of at least one of charts, visualizations, or audio,
wherein the ontology includes at least one meta-concept as a grouping of one or more of quantifiable or qualifying or categorical attributes associated with the extracted entities of the generated ontology.

18. The conversational interface according to claim 17 , wherein the DA model is defined in terms of quantifiable, qualifying or categorical entities and their relationships as described by the domain schema, and the conversational interface further comprises:
a training module configured to train for identification of different types of user intent from one or more training samples including user utterances.

19. The conversational interface according to claim 17 , wherein the processor is further configured to:
form an ontology graph from the generated ontology;
enhance the ontology graph by adding one or more of new concepts, groupings, hierarchies, relationships discerned from a data-driven machine learning, a deep learning, an embedding based technique for a named entity recognition, or a link prediction; and
periodically enhance the ontolog

Assignees

• IBM

Inventors

• Quamar Abdul
• Moore Robert John
• Miller Dorian Boris
• Ozcan Fatma
• Kreulen Jeffrey Thomas

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06F40/279

  Recognition of textual entities · CPC title

• G16H50/50

  for simulation or modelling of medical disorders · CPC title

• G06Q40/00

  Finance; Insurance; Tax strategies; Processing of corporate or income taxes · CPC title

• G06N20/00

  Machine learning · CPC title