Workflow generation from natural language statements

Patent metadata

Field	Value
Publication number	US-10360066-B2
Application number	US-201715420272-A
Country	US
Kind code	B2
Filing date	Jan 31, 2017
Priority date	Oct 25, 2016
Publication date	Jul 23, 2019
Grant date	Jul 23, 2019

Abstract

In one example in accordance with the present disclosure, a method may include classifying each word in a natural language statement and determining an implementation, from a set of possible implementations, for a workflow platform based on the classified words. The method may also include mapping a first of the classified words to a task selected from a set of possible tasks associated with the implementation and mapping a second of the classified words to an input parameter associated with the task. The method may also include generating a workflow for the workflow platform using the task and the input.

First claim

We claim:

1. A method comprising:
classifying, by a computer, each word in a natural language statement;
determining, by the computer, an implementation, from a set of possible implementations, for a processor-based workflow platform based on the classified words;
determining, by the computer, available tasks for the determined implementation;
mapping, by the computer, a first word of the classified words to a task selected from the determined available tasks, wherein the first word comprises a verb;
mapping, by the computer, a second word of the classified words to an input parameter associated with the task;
determining, by the computer, that, aside from the input parameter mapped to the second word of the classified words, a different input parameter corresponding to the task is missing from the natural language statement;
determining, by the computer, that an output parameter or an input parameter of a previous action meets requirements of the different input parameter; and
generating, by the computer, a workflow for the processor-based workflow platform for performing at least one of deploying or managing a component of a cloud-based computer system using the selected task and the input parameter, wherein generating the workflow for the processor-based workflow platform comprises generating a tool for virtualization.

2. The method of claim 1 , further comprising:
identifying, by the computer, a noun in the natural language statement; and
mapping, by the computer, the noun to the determined implementation from the set of possible implementations.

3. The method of claim 1 , further comprising:
identifying, by the computer, a noun in the natural language statement; and
mapping, by the computer, the noun to an applicable object for the determined implementation.

4. The method of claim 1 , further comprising:
identifying, by the computer, a group of unmapped words in the natural language statement;
determining, by the computer, that a first word in the group is a unit of measurement;
determining, by the computer that a second word in the group is a number;
marking, by the computer, the second word as a parameter corresponding to the unit of measurement;
determining, by the computer, that a third word in the group is a pronoun; and
marking, by the computer, the third word as a value corresponding to the unit of measurement.

5. The method of claim 4 , further comprising:
determining, by the computer, that a word belonging to the group is a preposition; and
removing, by the computer, the preposition from the group.

6. The method of claim 1 , further comprising:
identifying, by the computer, a second natural language statement associated with the natural language statement;
generating, by the computer, a second workflow for the workflow platform based on the second natural language statement; and
marking, by the computer, the second workflow as an error condition for the workflow.

7. The method of claim 1 , wherein classifying, by the computer, each word in the natural language statement comprises assigning each word to at least one of the following classifications: a noun, a verb, an adjective or a preposition.

8. The method of claim 1 , further comprising:
generating, by the computer, data serialization instructions from the task and input; and
generating, by the computer, the workflow from the data serialization instructions.

9. The method of claim 1 , further comprising:
determining, by the computer, that no words from the natural language statement match any implementation from the set of possible implementations;
identifying, by the computer, an unmapped word from the classified words;
determining, by the computer, a context keyword, from a synonyms database, that is associated with the unmapped word;
determining, by the computer, a synonym that is associated with the context keyword;
determining, by the computer, an implementation corresponding to the synonym; and
mapping, by the computer, the implementation to the unmapped word.

10. The method of claim 1 , wherein determining that an output parameter or an input parameter of a previous action meets requirements of the different input parameter comprises:
applying, by the computer, fuzzy logic to determine a semantic similarity between the different input parameter and the output parameter or input parameter of the previous action, and
applying, by the computer, fuzzy logic to determine a semantic similarity between a description of the different input parameter and a description of the output parameter or input parameter of the previous action.

11. A system comprising:
at least one processor; and
a memory to store instructions that, when executed by the at least one processor, cause the at least one processor to:
classify each word in a natural language statement based on a part of speech associated with the word;
determine an implementation, from a first dictionary, for a processor-based workflow platform based on the classified words, wherein the first dictionary includes functionalities supported by the processor-based workflow platform and implementations of the functionalities;
map a first word of the classified words to a task selected from a second dictionary, wherein the second dictionary includes tasks associated with the implementation;
map a second word of the classified words to an input parameter associated with the selected task;
determine that, aside from the input parameter mapped to the second word of the classified words, a different input parameter corresponding to the task is missing from the natural language statement;
determine that an output parameter or an input parameter of a previous action meets requirements of the different input parameter; and
generate a workflow for the processor-based workflow platform to perform at least one of deploying or managing a cloud-based computer system using the selected task and the input parameter, wherein generating the workflow for the processor-based workflow platform comprises generating a workflow to deploy or create a virtual machine.

12. The system of claim 11 , wherein the instructions, when executed by the at least one processor, cause the at least one processor to:
identify a first noun in the natural language statement;
map the noun to the implementation from the first dictionary;
identify a second noun in the natural language statement; and
map the noun to an applicable object for the implementation.

13. The system of claim 11 , wherein the instructions, when executed by the at least one processor, cause the at least one processor to:
identify a verb in the natural language statement;
determine available tasks, from the second dictionary, for the implementation; and
map the task from the available tasks to the verb.

14. The system of claim 11 , wherein the instructions, when executed by the at least one processor, cause the at least one processor to:
identify a group of unmapped words in the natural language statement;
determine that a first word in the group is a unit of measurement;
determine that a second word in the group is a number;
mark the second word as a parameter corresponding to the unit of measurement;
determine that a third word in the group is a pronoun; and
mark the third word as a value corresponding to the unit of measurement.

15. The system of claim 11 , wherein generating the workflow to deploy or create the virtual machine comprises selecting a snapshot or a template.

16. A non-transitory machine-readable storage medium encoded with instructions, the instructi

Assignees

• Hewlett Packard Entpr Dev Lp
• Entit Software Llc

Inventors

• Suparna Adarsh
• Vitala Pramod Annachira

Classifications

• G06F40/242

  Dictionaries · CPC title

• G06F9/5038

  considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration (scheduling strategies G06F9/4881 and subgroups) · CPC title

• G06F40/284

  Lexical analysis, e.g. tokenisation or collocates · CPC title

• G06F40/247 Primary

  Thesauruses; Synonyms · CPC title

• G06F9/4881 Primary

  Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues · CPC title