Process of dialogue and discussion

• US 8,473,449 B2
• Filed: 12/22/2009
• Issued: 06/25/2013
• Est. Priority Date: 01/06/2005
• Status: Active Grant

• Current Assignee: Neuric Technologies, LLC
• Original Assignee: Neuric Technologies, LLC
• Inventors: Visel, Thomas A.
• Primary Examiner(s): Fernandez Rivas, Omar F
• Assistant Examiner(s): Seck, Ababacar
• Application Number:
  US12/645,317
• Publication Number:
  US 20100185437A1
• Time in Patent Office: 1,281 Days
• Field of Search: 706/55
• US Class Current: 706/55
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 3/10 Interfaces, programming languages or software development kits, e.g. for simulating neural networks
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

A method for effecting a dialogue with an emulated brain. The method includes the step of receiving a query in the form of a semantic string. The semantic string is then parsed into basic concepts of the query. The basic concepts are then clumped into a clump concept. If the clump concept constitutes part of a dialogue, then the dialogue thread is activated by determining the context of the clump concept and assessing a potential reply from a group of weighted replies, which expected replies are weighted based on the parsed concepts produced in the step of parsing. The heaviest weighted one of the expected replies is selected and the weight of the selected reply after it is selected is downgraded. The selected reply is then generated for output in a sentence structure.

Claims

1. A method for effecting a dialogue with an emulated brain that can receive a query on an input and provide a reply on an output, comprising the steps of:

receiving a query in the form of a semantic string;

providing a plurality of neurons, wherein one neuron constitutes a predetermined activity neuron that represents one concept and at least one concept is that of an activity associated with a type of verb;

defining relational connections between the activity neuron and a second group of the plurality of neurons, each of the second group of the plurality of neurons representing a fixed concept that is unique to the activity associated with the activity neuron;

wherein the meaning of the activity neuron is determined by a relational connection between the activity neuron and the second group of the plurality of neurons established through the relational connection and wherein the relationship between the activity neuron associated with the concept of activity and the second group of neurons comprises a framework of neurons uniquely related to the at least one concept of the predetermined activity associated with the type of verb associated with that predetermined activity;

providing a plurality of verb neurons, each representing the concept of an associated verb and each associated neuron defined by a plurality of relationships to other neurons and each having a relationship defined to the activity neuron;

providing a plurality of neurons associated with other grammatical elements that can be found in a bounded grammatical structure;

storing the created neurons and associated relationships in a database;

parsing the semantic string into basic concepts of the query, each concept associated with one of the plurality of neurons, and clumping the basic concepts into a clump concept by the steps of;

parsing an input bounded grammatical structure to derive the verb associated with the bounded structure and the other associated grammatical elements;

determining if the verb in the bounded structure is a verb of the type associated with the activity neuron;

determining which of the relationships from the activity neuron to the associated second group of neurons exists which will define the relationship of the parsed grammatical elements to the parsed verb; and

creating a clump neuron defining all of the associated neurons for the parsed grammatical structure and the associated relationships as the clump concept;

storing the clump neuron in the database of neurons;

determining if the clump concept constitutes part of a dialogue and, if so;

activating a dialogue thread loop by;

determining a context of the clump concept;

assessing a potential reply from a group of expected replies associated with the clump concept, which expected replies are weighted based on the parsed concepts from the step of parsing;

selecting the heaviest weighted one of the expected replies and changing the weight thereof after selection by downgrading the weight of the selected one of expected replies;

generating the selected reply for output in a sentence structure; and

outputting the reply.

2. The method of claim 1, wherein the grammatical structure is a sentence.

3. The method of claim 1, wherein the predetermined activity is a transitive activity and the type of verb is a transitive verb.

4. The method of claim 3, wherein the transitive verb constitutes the grammatical role of the verb in the grammatical structure and the second group of neurons defines the restrictions on the transitive verb relative to the other grammatical elements, such that an actor and a recipient if the activity can be defined and attributed their associated roles in the transitory activity.

5. The method of claim 1, wherein the step of creating a clump neuron defining all of the associated neurons for the parsed grammatical structure and the associated relationships as the clump concept defines the grammatical relationships between the neurons in the clump.