Browse Prior Art Database

Method of Predictive Voice Driven Diagramming using Personal Dictionary and Bayesian Probabilistic Ranking

IP.com Disclosure Number: IPCOM000240933D
Publication Date: 2015-Mar-12
Document File: 6 page(s) / 55K

Publishing Venue

The IP.com Prior Art Database

Abstract

The method given can do prediction of User's probable next move in Voice Controlled Diagramming. The method is based on Bayesian Probability. Voice driven diagramming steps made by the User are analyzed and ranked by the algorithm. Before the user makes the next move, the method notifies the Diagramming Application of the most probable next move suggestion. Based on the suggested move, the application can display the appropriate message requesting for user’s confirmation. This description specifies the method to form Voice Vectors based on User Inputs to the Speech Engine. Speech Engine is outside scope. The method shows how to formulate Voice Vectors in Voice Control for Diagramming Applications. Based on Bayesian Classification and Probability calculation, the solution shows how to classify the Standard Voice Vocabulary Meta-Data and calculate the probability for each type of meta-data for different Users.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 37% of the total text.

Page 01 of 6

Method of Predictive Voice Driven Diagramming using Personal Dictionary and Bayesian Probabilistic Ranking

CONCEPT:

Voice driven diagramming steps made by the User are analyzed and ranked by the algorithm. Before the user makes the next move, the method notifies the Diagramming Application of the most probable next move suggestion. Based on the suggested move, the application can display the appropriate message requesting for user's confirmation. This method to form Voice Vectors based on User Inputs to the Speech Engine. Speech Engine is outside scope. The method shows how to formulate Voice Vectors in Voice Control for Diagramming Applications. Based on Bayesian Classification and Probability calculation, the solution shows how to classify the Standard Voice Vocabulary Meta-Data and calculate the probability for each type of meta-data for different Users.

Based on the maximum probability the algorithm calculates the User's Next Step as input to the Diagramming Application to show if the User intends to carry out that step. If the User confirms the suggested action the diagramming application carries out the step.

Additionally a Standard Ranked Library of Users most used sequences is also maintained. Based on the user's final action the Standard Ranked Library or the Learnt Sequences Library is updated to improve most likely user input sequences.

Based on prediction of User's next step, the methodology eases the use of the diagramming application and makes it more user friendly. Since it is based on standardized voice vocabulary meta-data the method can be applied across any diagramming application.

The method can also be applied to non-voice diagramming.

Drawbacks of the current solution:

Current Solutions do not state predictive feedback for voice based diagramming.

Current Solutions do not state intelligent human machine interface for diagramming applications.

Novelty of the solution:

Method to store User actions for Voice Control in Diagramming Applications based on standardized meta-data vocabulary.

Store Subject-Verb-Object-Attribute as a bundled and ranked sequence step.

Bayesian Score (Ranking) based on repetitive verb-subject-object joins formed by the individual and pre-programmed known action sequences with ranking
Verb score decides the next predictive step User would take.

Method to form bundled Vectors based on Nouns/Verbs/Subject/Objects in Voice Controlled Diagramming.

Method to correlate Multiple Vectors for Predictive Output based on Bayesian Probability. Personal Sequence Ranking Dictionary of User's likely steps can be derived from the Learning

1


Page 02 of 6

Database Matrix.

The advantages of this idea are:


 Having a pre-defined ranking database of most-probable steps by a User.


 Simple Bayesian classification applied to User Vectors with simplified data-base objects.


 Having Bayesian score on sequenced vectors which are linked tags to the subject / object / verbs that helps in predicting the most li...