Function-based action sequence derivation for personal assistant system

Patent metadata

Field	Value
Publication number	US-9619283-B2
Application number	US-201514810778-A
Country	US
Kind code	B2
Filing date	Jul 28, 2015
Priority date	Jul 28, 2015
Publication date	Apr 11, 2017
Grant date	Apr 11, 2017

Abstract

A method is provided for recommending a desired func sequence to a user. The method includes obtaining a user intention list including at least one user intention; separating the user intention into a plurality of tasks; and creating a task flow graph for the plurality of tasks based on user usage data. Each vertex in the task flow graph represents a task and indicating an importance of the task. The method also includes creating a func flow graph based on the user usage data and temporal sequences of the tasks and funcs, and each vertex in the func flow graph represents a func and indicating an importance of the func. Further, the method includes determining a desired func sequence to fulfill the user intention based on the user usage data, the task flow graph, and the func flow graph; and recommending the desired func sequence to the user.

First claim

What is claimed is:

1. A method for recommending a desired func sequence to a user, comprising:
obtaining, by a hardware processor, a user intention list including at least one user intention;
separating, by the hardware processor, the user intention into a plurality of tasks, and determining a plurality of funcs, wherein each task is implemented by a subset of funcs of the plurality of funcs, each func is used to represent an access point of an app:
creating, by the hardware processor, a task flow graph for the plurality of tasks based on user usage data, with each vertex in the task flow graph representing a task and indicating an importance of the task;
creating, by the hardware processor, a func flow graph based on the user usage data and temporal sequences of the tasks and funcs, with each vertex in the func flow graph representing a func and indicating an importance of the func;
determining, by the hardware processor, a desired func sequence to fulfil the user intention based on the user usage data, the task flow graph, and the func flow graph; and
recommending, by the hardware processor, the desired func sequence to the user, wherein: two vertexes in the func flow graph are linked to each other if a traversal from one to another has been found consecutively in the user usage data; and
any two vertexes in the func flow graph are consecutively linked if two funcs of the two vertexes happened in sequential order in the user usage data within a certain time duration; and
an edge weight between any two vertexes in the func flow graph is a total count of recorded traversal between the two vertexes.

2. The method according to claim 1 , further including:
obtaining, by the hardware processor, a user text request as an input from the user to indicate the user intention; and
deducting, by the hardware processor, the intention list from the user text request.

3. The method according to claim 1 , wherein:
the importance of the func is calculated based on general traversal frequencies from the func to every other funcs, and personal traversal frequencies from the func to every other funcs.

4. The method according to claim 1 , wherein:
two vertexes in the task flow graph are linked to each other if a traversal from one to another has been found consecutively in the user usage data;
any two vertexes in the task flow graph are consecutively linked if those two tasks of the two vertexes happened in sequential order in the user usage data within a certain time duration; and
an edge weight between any two vertexes in the task flow graph is a total count of recorded traversal between the two vertexes.

5. The method according to claim 1 , wherein:
the importance of the task is calculated based on general traversal frequencies from the task to every other tasks, and personal traversal frequencies from the task to every other tasks.

6. The method according to claim 1 , wherein determining the desired func sequence further includes:
determining a starting point and an ending point of tasks associated with each user intention of the intention list;
generating a sub-graph containing the plurality tasks including the starting point and ending point of tasks; and
determining a most probable task sequence for each pair of the starting point and ending point of tasks.

7. The method according to claim 6 , wherein determining the most probable task sequence further includes:
determining a most probable task sequence for each pair of the starting point and ending point of tasks based on an Extended Floyd Algorithm.

8. The method according to claim 6 , further including:
determining a desired task sequence of the intention list based on the most probable task sequence for each pair of the starting point and ending point of tasks.

9. The method according to claim 8 , further including:
determining a corresponding desired func sequence based on the desired task sequence of the intention list.

10. The method according to claim 9 , wherein determining the corresponding desired func sequence further includes:
determining a corresponding desired func sequence based on the desired task sequence of the intention list using a Viterbi algorithm, with the tasks as the observation states and the funcs as the hidden states.

11. A system for recommending a desired func sequence to a user, comprising:
at least a memory for storing user usage data; and
a hardware processor coupled to the memory and configured to:
obtain a user intention list including at least one user intention;
separate the user intention into a plurality of tasks, and determine a plurality of funcs, wherein each task is implemented by a subset of funcs of the plurality of funcs, each func is used to represent an access point of an app;
create a task flow graph for the plurality of tasks based on the user usage data, with each vertex in the task flow graph representing a task and indicating an importance of the task;
create a func flow graph based on the user usage data and temporal sequences of the tasks and funcs, with each vertex in the func flow graph representing a func and indicating an importance of the func;
determine a desired func sequence to fulfil the user intention based on the user usage data, the task flow graph, and the func flow graph; and
recommend the desired func sequence to the user, wherein: two vertexes in the func flow graph are linked to each other if a traversal from one to another has been found consecutively in the user usage data;
any two vertexes in the func flow graph are consecutively linked if two funcs of the two vertexes happened in sequential order in the user usage data within a certain time duration; and

an edge weight between any two vertexes is a total count of recorded traversal between the two vertexes.

12. The system according to claim 11 , wherein:
the importance of the func is calculated based on general traversal frequencies from the func to every other funcs, and personal traversal frequencies from the func to every other funcs.

13. The system according to claim 11 , wherein:
two vertexes in the task flow graph are linked to each other if a traversal from one to another has been found consecutively in the user usage data;
any two vertexes in the task flow graph are consecutively linked if those two tasks of the two vertexes happened in sequential order in the user usage data within a certain time duration; and
the importance of the task is calculated based on general traversal frequencies from the task to every other tasks, and personal traversal frequencies from the task to every other tasks.

14. The system according to claim 11 , wherein, to determine the desired func sequence, the hardware processor is further configured to:
determine a starting point and an ending point of tasks associated with each user intention of the intention list;
generate a sub-graph containing the plurality tasks including the starting point and ending point of tasks; and
determine a most probable task sequence for each pair of the starting point and ending point of tasks.

15. The system according to claim 14 , wherein, to determine the most probable task sequence, the hardware processor is further configured to:
determine a most probable task sequence for each pair of the starting point and ending point of tasks based on an Extended Floyd Algorithm.

16. The system according to claim 14 , wherein the hardware processor is further configured to:
determine a desired task sequence of the intention list based on the most probable task sequence for each pair of the starting point and e

Assignees

• Tcl Res America Inc

Inventors

• Wu Hsin-Tai
• Song Xiaoli
• Cheng Weilun
• Shang Yue
• Wang Haohong

Classifications

• G06N7/01

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

• G06F16/9535

  Search customisation based on user profiles and personalisation · CPC title

• G06F9/453

  Help systems · CPC title

• G06F9/4881 Primary

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

• G06N5/02

  Knowledge representation; Symbolic representation · CPC title