Autonomous learning of actionable models from unstrutured data

Patent metadata

Field	Value
Publication number	US-2018218272-A1
Application number	US-201715840548-A
Country	US
Kind code	A1
Filing date	Dec 13, 2017
Priority date	Jan 31, 2017
Publication date	Aug 2, 2018
Grant date	—

Abstract

Techniques for autonomously generating a domain model and/or an action model based on unstructured data are provided. In one example, a computer implemented method can comprise extracting, by a system operatively coupled to a processor, a plurality of actions from a non-numerical language. The plurality of actions can achieve a goal. The computer-implemented method can also comprise generating, by the system, a domain model based on the plurality of actions. Further, the computer-implemented method can comprise generating, by the system, an action model based on the domain model. In various embodiments, the action model can comprise an action transition for accomplishing the goal.

First claim

What is claimed is:

1 . A computer-implemented method, comprising:
extracting, by a system operatively coupled to a processor, a plurality of actions from a non-numerical language, wherein the plurality of actions achieve a goal; generating, by the system, a domain model based on the plurality of actions; and generating, by the system, an action model based on the domain model, wherein the action model comprises an action transition for accomplishing the goal.

2 . The computer-implemented method of claim 1 , wherein the non-numerical language is a text shared over a global corpus.

3 . The computer-implemented method of claim 2 , wherein the text is in English language and the global corpus is the Internet.

4 . The computer-implemented method of claim 2 , wherein the text is generated by a plurality of entities.

5 . The computer-implemented method of claim 1 , wherein the generating the domain model comprises grouping one or more actions from the plurality of actions into a cluster based on similarity of the one or more actions.

6 . The computer-implemented method of claim 5 , wherein the generating the domain model comprises identifying an action within the cluster as a cluster representative.

7 . The computer-implemented method of claim 6 , wherein the domain model comprises a plurality of cluster representatives.

8 . The computer-implemented method of claim 7 , wherein the generating the action model comprises determining a probability that one or more cluster representatives from the plurality of cluster representatives are effects of another cluster representative from the plurality of cluster representatives.

9 . The computer-implemented method of claim 1 , further comprising determining a quality of the action model.

10 . The computer-implemented method of claim 1 , further comprising determining an efficiency of the action model.

11 . The computer-implemented method of claim 1 , wherein the action model is probabilistic.

12 . The computer-implemented method of claim 1 , wherein the action transition is a sequence of actions from the plurality of actions.

Assignees

• IBM

Inventors

• Manikonda Lydia
• Riabov Anton Viktorovich
• Sohrabi Araghi Shirin
• Srivastava Biplav
• Talamadupula Kartik
• Turaga Deepak Srinivas

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title

• G06F40/289

  Phrasal analysis, e.g. finite state techniques or chunking · CPC title

• G06F30/20

  Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30) · CPC title

• G06N5/04 Primary

  Inference or reasoning models · CPC title