U.S. patent application number 17/141092 was filed with the patent office on 2021-04-29 for computer-implemented system and method for providing contextually relevant task recommendations to qualified users.
The applicant listed for this patent is Palo Alto Research Center Incorporated. Invention is credited to Shane Ahern, Michael Roberts, Simon Tucker.
Application Number | 20210125134 17/141092 |
Document ID | / |
Family ID | 1000005324625 |
Filed Date | 2021-04-29 |
![](/patent/app/20210125134/US20210125134A1-20210429\US20210125134A1-2021042)
United States Patent
Application |
20210125134 |
Kind Code |
A1 |
Roberts; Michael ; et
al. |
April 29, 2021 |
COMPUTER-IMPLEMENTED SYSTEM AND METHOD FOR PROVIDING CONTEXTUALLY
RELEVANT TASK RECOMMENDATIONS TO QUALIFIED USERS
Abstract
A computer-implemented system and method for providing
contextually relevant task recommendations to qualified users is
provided. Contextual data for a user is collected via one more
sensors associated with the user. An activity being performed by
the user is determined based on the contextual data. The activity
is compared with a profile of the user and a determination that the
user is qualified to perform the activity is verified. The activity
is monitored. At least one recommendation for performance of a step
is identified during the activity and transmitted to at least one
recipient including the user or another user.
Inventors: |
Roberts; Michael; (Los
Gatos, CA) ; Tucker; Simon; (Oakland, CA) ;
Ahern; Shane; (Foster City, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Palo Alto Research Center Incorporated |
Palo Alto |
CA |
US |
|
|
Family ID: |
1000005324625 |
Appl. No.: |
17/141092 |
Filed: |
January 4, 2021 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
15203727 |
Jul 6, 2016 |
10885478 |
|
|
17141092 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06Q 10/063112 20130101;
G06F 16/24578 20190101 |
International
Class: |
G06Q 10/06 20060101
G06Q010/06; G06F 16/2457 20060101 G06F016/2457 |
Claims
1. A computer-implemented system for providing contextually
relevant task recommendations to qualified users, comprising:
contextual data collected for a user via one more sensors
associated with the user; and a server comprising a central
processing unit, an input port to receive the contextual data, and
an output port, wherein the central processing unit is configured
to: acknowledge an activity being performed by the user based on
the contextual data; compare the activity with a profile of the
user; verify that the user is qualified to perform the activity
based on the user profile; monitor the activity; identify at least
one recommendation for performance of a step during the activity;
and transmit the recommendation to at least one recipient
comprising at least one of the user and another user.
2. A system according to claim 1, wherein the central processing
unit is further configured to: compare the recommendation with the
user profile; and determine from the user profile whether the user
is qualified to perform the recommendation.
3. A system according to claim 2, wherein the central processing
unit is further configured to: select the other user as the
recipient of the recommendation when the user is not qualified to
perform the recommendation.
4. A system according to claim 1, wherein the central processing
unit is further configured to: place the user and the other user
into one or more groups as members based on one or more topics
associated with each of the groups.
5. A system according to claim 4, wherein the central processing
unit is further configured to: select one or more of the groups
based on the recommendation; and transmit the recommendation to the
members in the selected group.
6. A system according to claim 1, wherein the central processing
unit is further configured to: access guidelines for the activity;
and determine whether the guidelines are performed correctly by the
user during the monitoring of the activity.
7. A system according to claim 1, further comprising: a mobile
computing device to determine the activity being performed by the
user and to transmit the activity to the server.
8. A system according to claim 1, wherein the central processing
unit is further configured to: score candidate recommendations
based one at least one of each recipient's preference, interest,
and location.
9. A system according to claim 8, wherein the central processing
unit is further configured to: select at least one of the candidate
recommendations with the highest score as the recommendation.
10. A system according to claim 1, wherein the central processing
unit is further configured to: provide to the recipient
instructions to perform the recommendation.
11. A computer-implemented method for providing contextually
relevant task recommendations to qualified users, comprising:
collecting contextual data for a user via one more sensors
associated with the user; determining an activity being performed
by the user based on the contextual data; comparing the activity
with a profile of the user; verifying that the user is qualified to
perform the activity based on the user profile; monitoring the
activity; identifying at least one recommendation for performance
of a step during the activity; and transmitting the recommendation
to at least one recipient comprising at least one of the user and
another user.
12. A method according to claim 11, further comprising: comparing
the recommendation with the user profile; and determining from the
user profile whether the user is qualified to perform the
recommendation.
13. A method according to claim 12, further comprising: selecting
the other user as the recipient of the recommendation when the user
is not qualified to perform the recommendation.
14. A method according to claim 11, further comprising: placing the
user and the other user into one or more groups as members, wherein
each group is associated with at least one topic.
15. A method according to claim 14, further comprising: selecting
one or more of the groups based on the recommendation; and
transmitting the recommendation to the members in the selected
group.
16. A method according to claim 11, further comprising: accessing
guidelines for the activity; and determining whether the guidelines
are performed correctly by the user during the monitoring of the
activity.
17. A method according to claim 11, wherein the activity being
performed by the user is determined via a mobile computing device
and is transmitted from the mobile computing device to a
server.
18. A method according to claim 11, further comprising: scoring
candidate recommendations based one at least one of each
recipient's preference, interest, and location.
19. A method according to claim 18, further comprising: selecting
at least one of the candidate recommendations with the highest
score as the recommendation.
20. A method according to claim 11, further comprising: providing
to the recipient instructions to perform the recommendation.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] This application is a continuation of U.S. patent
application Ser. No. 15/203,727, filed Jul. 6, 2016, pending, the
priority date of which is claimed and the disclosure of which is
incorporated by reference.
FIELD
[0002] This application relates in general to determining context,
and in particular to a computer-implemented system and method for
providing contextually relevant task recommendations to users
engaged in tasks.
BACKGROUND
[0003] As younger generations join the workforce, companies are
searching for effective ways to train and interact with these new
employees. Currently, health and safety is taught and regulated
using a combination of printed training material, posted signs, and
supervision. However, younger individuals are used to receiving
information and notices on their mobile computing devices,
including smart phones and tablets. Based on the different
experiences and expectations of the younger workforce, existing
methods for training and communication are not as effective as in
the past.
[0004] Currently, some companies are allowing online access to
their information, including policies, employee handbooks, employee
email, work product, and other documents, such as through a virtual
private network. However, despite the ability to access such
information, employees may have questions regarding the information
or are unable to properly perform particular tasks listed in the
available materials, which can result in accidents or lack of
productivity, as well as a possible violation of compliance
guidelines. Further, the information may not be immediately
available when needed by the user. For instance, if the user is
working in a laboratory and a fire starts, the user is likely
unable to login to the virtual private network to access
information for putting out the fire and proper protocol.
Additionally, even if information about preventing and putting out
fires is on a wall in the laboratory, the user may be unable to
access the information or may be unqualified to perform the
necessary procedures to put out the fire or conduct a proper clean
up once the fire is out.
[0005] Therefore, there is a need for an approach to providing
relevant information to an appropriate individual during a time at
which the information is necessary and relevant. Preferably, a
current scenario is identified for a user and one or more tasks to
be performed are sent to a mobile computing device associated with
the user based on the scenario. If the user is unable to perform
one or more of the recommended tasks, a further individual can be
notified to complete the recommended step.
SUMMARY
[0006] To ensure that employees of a company are provided with
information relevant to a current situation, contextual information
of each employee is first collected and used to determine a
projected activity of the employee. Subsequently, based on the
determined activity, one or more tasks can be recommended to the
employee via a mobile computing device associated with the
employee. If the employee is not qualified to perform one or more
of the recommended tasks, one or more other employees are
identified and notified of the recommended task to be performed.
Such employees can provide just-in-time training to the original
employee, or perform the task themselves.
[0007] A computer-implemented system and method for providing
contextually relevant task recommendations to qualified users is
provided. Contextual data for a user is collected via one more
sensors associated with the user. An activity being performed by
the user is determined based on the contextual data. The activity
is compared with a profile of the user and a determination that the
user is qualified to perform the activity is verified. The activity
is monitored. At least one recommendation for performance of a step
is identified during the activity and transmitted to at least one
recipient including the user or another user.
[0008] Still other embodiments of the present invention will become
readily apparent to those skilled in the art from the following
detailed description, wherein is described embodiments of the
invention by way of illustrating the best mode contemplated for
carrying out the invention. As will be realized, the invention is
capable of other and different embodiments and its several details
are capable of modifications in various obvious respects, all
without departing from the spirit and the scope of the present
invention. Accordingly, the drawings and detailed description are
to be regarded as illustrative in nature and not as
restrictive.
BRIEF DESCRIPTION OF THE DRAWINGS
[0009] FIG. 1 is a block diagram showing a computer-implemented
system for providing contextually relevant task recommendations to
qualified users, in accordance with one embodiment.
[0010] FIG. 2 is a flow diagram showing a computer-implemented
method for providing contextually relevant task recommendations to
qualified users, in accordance with one embodiment.
[0011] FIG. 3 is a flow diagram showing, by way of example, a
process for determining an activity associated with a user.
[0012] FIG. 4 is a flow diagram showing, by way of example, a
process for identifying recommendations.
[0013] FIG. 5 is a flow diagram showing, by way of example, a
process for identifying recipients of recommendations.
DETAILED DESCRIPTION
[0014] Currently, most companies train new employees using printed
material, posted signs, and supervision. However, the younger
workforce is accustomed to receiving, maintaining, and accessing
information on their mobile computing devices, such as smart phones
and tablets, which can make using the printed materials and signs
difficult. Further, printed materials are often misplaced or thrown
away, which makes accessing such documents, difficult. Accordingly,
to ensure that all employees can readily access relevant
information when needed or that assistance is provided in a timely
fashion, the activities of each employee can be monitored.
Subsequently, task recommendations can be provided to the employee
based on the current activity being performed. Also, if the
employee is not qualified to perform one of the recommended tasks,
another employee can be contacted.
[0015] Providing relevant information to an appropriate user in a
timely manner can increase employee productivity, reaffirm and
evaluate training, ensure guideline compliance, and prevent
accidents. FIG. 1 is a block diagram showing a computer-implemented
system 10 for providing contextually relevant task recommendations
to qualified users, in accordance with one embodiment. An employee
(not shown) of a company or other individual can access an
application for providing relevant tasks to qualified employees
based on an identified activity. The employee can be associated
with one or more computing devices, including a smartphone 11, a
smart watch 12, and a head-mounted computing device 13, such as
Google Glass manufactured by Google Inc. Hereinafter, the terms
"employee" and "user" are used interchangeably with the same
intended meaning, unless otherwise indicated.
[0016] Each of the computing devices 11-13 can collect contextual
data 23 of the employee, including one or more of location,
acceleration, movement tracking, inventory, use of machinery,
chemicals, and other types of data related to the user's current
surroundings. The contextual data 23 can be collected via sensors
within the computing devices 11-13 or via sound or video recording
via the computing devices. Further, Bluetooth beacons 29 can be
affixed to one or more pieces of equipment 28 located within a
workspace (not shown) owned by the employee's company or at various
locations within the workspace itself. Each of the mobile computing
devices 11-13 can also include Bluetooth to enable communication
with the Bluetooth beacons 29 and identify a location of a user
with respect to one or more of the Bluetooth beacons.
[0017] Once collected, the contextual data 23 can be separately
transmitted from each computing device 11-13 to a server 15 via an
internetwork 14, such as the Internet. The server 15 includes a
context tracker 16, an activity identifier 17, a recommender 18, a
qualifier 19, a recipient identifier 20, and a notification module
21. Further, the server is interconnected to a database 22, which
can be used to store the contextual data 23, activity models 24,
user profiles 25, recommendations 26, and groups 27.
[0018] The contextual data 23 from each of the computing devices
11-13 is received by the contextual tracker 16 for analysis and
storage in the database 22. Specifically, the contextual tracker 16
can identify one or more low level activities being performed by
the employee based only on the contextual data. In a further
embodiment, the low level activities can be identified by the
specific computing device that collected the contextual data.
Subsequently, the activity identifier 17 identifies high level
activities of the employee using activity models 24 based on one or
more of the low level activities. Identifying activities is further
described in detail with reference to FIG. 3. Once identified, the
activities of the user can be stored in the database 22 in an
activity log (not shown).
[0019] Based on the identified activity, helpful advice or tasks to
be performed can be provided to the employee, regardless of whether
a need for assistance by the employee is identified. The helpful
advice can include messages, information, instructions, schematics,
or plans, while the tasks can include service procedures, requests
for action, and agent-actions to make on behalf of the employee.
The recommender 18 can provide the helpful advice and tasks as
recommendations 26 for display on at least one of the computing
devices 11-13 associated with the employee. Identifying appropriate
recommendations can be based on the identified activity, as
described in further detail in commonly-owned U.S. Patent
Application Publication, Serial No. 2018/0012229, published on Jan.
11, 2018, pending, the disclosure of which is incorporated by
reference. Additionally, once the activity is identified, a
predicted outcome of the activity can also be determined, as
described in detail in commonly-owned U.S. Patent Application
Publication, Serial No. 2018/0012126, published on Jan. 11, 2018,
pending, the disclosure of which is incorporated by reference.
Subsequently, based on the outcome prediction, further assistance
or recommendations can be provided to the employee.
[0020] Prior to or subsequent to providing the recommendations 26
to the employee, the qualifier 19 can compare each recommendation
with a profile 25 of the employee to determine whether the employee
is qualified to perform that recommendation. The user profile 25
can include a name of the employee, birthdate, sex, age, address,
phone number, date of employment, and experience, such as work
history, training courses completed, skills, access permission to
specified areas, current activity, and predicted future activity.
Other types of profile information are possible. If the employee is
qualified, the recommendations can be transmitted to the employee
only. However, if the employee is not qualified to perform one or
more of the recommended tasks, the recipient identifier 20
identifies one or more other employees for providing the
recommended tasks. In one embodiment, the employees can each be
associated with one or more groups 27 based on their skills and
experience as provided by their user profile. Each group is then
associated with one or more topics or key words describing the
skills and abilities of the member employees in that group. The
recommended tasks can then be recommended to one or more of the
groups based on the designated skills and abilities of the group
members.
[0021] The mobile computing devices 11-13 and server 15 can each
include one or more modules for carrying out the embodiments
disclosed herein. The modules can be implemented as a computer
program or procedure written as source code in a conventional
programming language and is presented for execution by the central
processing unit as object or byte code. Alternatively, the modules
could also be implemented in hardware, either as integrated
circuitry or burned into read-only memory components, and each of
the client and server can act as a specialized computer. For
instance, when the modules are implemented as hardware, that
particular hardware is specialized to perform the data quality
assessment and other computers cannot be used. Additionally, when
the modules are burned into read-only memory components, the
computer storing the read-only memory becomes specialized to
perform the data quality assessment that other computers cannot.
The various implementations of the source code and object and byte
codes can be held on a computer-readable storage medium, such as a
floppy disk, hard drive, digital video disk (DVD), random access
memory (RAM), read-only memory (ROM) and similar storage mediums.
Other types of modules and module functions are possible, as well
as other physical hardware components.
[0022] Monitoring and tracking employees can increase the
productivity of a company as well as ensure that work is being
performed according to company guidelines, rules, and regulations
by identifying a contextual situation of an employee and providing
tasks to be performed based on the employee's situation. FIG. 2 is
a flow diagram showing a computer-implemented method 30 for
providing contextually relevant task recommendations to qualified
users, in accordance with one embodiment. Prior to or during a work
shift, an employee is associated with one or more sensors via at
least one mobile computing device. For instance, a researcher
working in a laboratory of a drug company can carry a smart phone
in his pocket, as well as wear a mobile computing headset or a
smart watch.
[0023] During the employee's shift, movement and actions of the
employee are monitored (block 31). Contextual data, including
location, acceleration, movement tracking, inventory, and other
types of data related to the user's current surroundings, are
collected by the sensors during the monitoring. The contextual data
is then used to help identify (block 32) an activity associated
with the employee, such as an action being performed or experienced
by the employee or an activity describing an existing situation
surrounding the employee. Identifying an activity is further
described below in detail with reference to FIG. 3.
[0024] Returning to the above laboratory example, contextual data
collected for the researcher includes a location of the employee,
which is obtained via GPS on the smart phone, a location of the
employee with respect to a particular piece of equipment, as
identified via Bluetooth, and a video of the employee obtained via
a camera, such as on the mobile headset. Based on the collected
contextual data, the researcher is determined to be performing the
activity of protein purification.
[0025] Once an activity is determined, one or more recommendations
can be identified (block 33) for transmitting to the employee via
one or more of the associated computing devices. The
recommendations can be identified using numeric calculations to
combine elements from the employee's profile with data from each of
the recommendations stored in a database. Selecting one or more
recommendations for providing to the employee is further discussed
below in detail with reference to FIG. 4. Once the appropriate
recommendations are identified, recipients are identified (block
34) for sending the identified recommendations. In one embodiment,
recipients of the recommendations can be identified based on the
each of the recipients' ability to complete the recommended tasks,
as further described below in detail with reference to FIG. 5. Upon
identifying the recipients, the recommended tasks are provided
(block 35) to each recipient via mobile computing devices
associated with that recipient. Finally, one or more tasks can be
selected (block 36) for performance by at least one of the
recipients. When multiple recipients are identified, the task
selection can be displayed to the other recipients to prevent
duplicate task selection. Alternatively, a determination can be
made that one of the recipients is performing one of the
recommended tasks based on contextual data collected for that
recipient and a notification can be sent to the other recipients.
In one embodiment, each task can be stored with a set of
instructions. Upon assignment of a task to a recipient, the
associated instructions can be displayed on a wireless computing
device of that recipient to assist in performing the task.
[0026] Returning to the above laboratory example, the researcher is
determined to be performing protein purification. Based on the
purification activity, selected recommendations provided to the
employee can include running the proteins through a further
purification process, using for example, high-performance liquid
chromatography; performing an SDS-PAGE gel to visualize the
isolated protein; and cutting extra tags off of the isolated
protein after nickel purification using an enzyme. Upon review of
the researcher's profile, a determination is made that the
researcher is qualified to perform each of the recommended tasks
and thus, is provided with all three recommendations.
[0027] Determining and tracking activities associated with an
employee helps anticipate a need of the user in real-time and
provide immediate assistance and resolution. Identifying such
activities can be performed based on sensor data collected from the
employee's surroundings. FIG. 3 is a flow diagram showing, by way
of example, a process 40 for determining an activity associated
with an employee. Contextual data is collected (block 41) from
sensors encompassed by mobile computing devices associated with the
employee. The mobile devices can include one or more of a smart
phone, smart watch, and head-mounted computing device, as well as
other types of mobile computing devices. Each of the mobile devices
can include multiple sensors to measure contextual data, including
speed, location, acceleration, physical movement, eye gaze, object
presence, inventory, scenery, and traffic, as well as other types
of data. Further, video and sound data can be recorded, and a
location of the employee with respect to a particular room or piece
of equipment within a work facility can be identified using
Bluetooth.
[0028] One or more low-level activities being performed by the
employee can be identified (block 42) directly from the contextual
data. Each low-level activity describes a raw action being
performed by the employee. For instance, if an accelerometer
provides a reading of zero, then the employee is determined to be
still and not accelerating or moving to a different location.
However, a different sensor may identify movement of the employee
by the pressing of a button on one of the mobile devices. The
low-level activities are then compared (block 43) with a set of
activity models to determine (block 44) a high-level activity
expressed by the model. A high-level activity describes a specific
action being performed by the employee based on the raw actions
detected. For instance, returning to the example above, the
employee is determined to be still, but moving with respect to the
mobile device button selection, which may indicate some sort of
work being conducted on the mobile device. Combined with data for
tracking computer use, the employee activity is determined to be
pressing send on an email.
[0029] In one embodiment, each high-level activity can be stored as
a model that includes one or more raw actions to identify that
specific high-level activity. Those models that most closely
resemble the detected raw actions of the employee are identified
and selected as the high-level activity identified as being
performed by the employee. Each activity model can be focused on
the specific employee based on actions performed by that employee
over time, as well as on background information regarding the
employee's job title and skill set. Alternatively, the activity
models can be based on a population of employees with the same or
similar job titles and skills as the employee. In one embodiment,
the low-level activities can each be detected by the mobile devices
associated with the employee, while the high-level activities can
be determined by a remote server using the activity models.
[0030] Returning to the above laboratory example, contextual data
collected for the employee identifies low level activities, such as
a presence of the employee in Lab Room E203, as identified via GPS,
standing in front of a fume hood, as identified via Bluetooth, and
holding a bottle of mercaptoethanol, as identified via a camera,
such as on the mobile headset. Based on the low level activities,
the employee is determined to be performing the high level activity
of preparing a lysing buffer, for use in protein purification, to
keep the isolated protein stable by preventing formation of
disulfide bonds and protein precipitation. Mercaptoethanol can also
be used for RNA isolation. However, referencing the employee's
profile, in addition to analysis of the contextual data, helps
identify that the employee is performing protein extraction as
opposed to RNA isolation since the profile indicates that the
employee is a research scientist and currently studying
protein-protein interactions to determine specific contact between
two or more proteins based on particular biochemical events.
[0031] In one embodiment, the employee's profile can include
hierarchical categorizations of skills and experience. For
instance, for profession, the hierarchical categorization can
include health industry, which covers medical providers,
researchers, insurers, and pharmaceuticals. Further, researchers
cover laboratory researchers and computational researchers, while
laboratory researchers covers chemistry, biochemistry, and biology.
Research of protein-protein interactions falls under biochemistry.
The hierarchical categories associated with an employee can be used
to identify activities, as described above, or to determine the
employee's ability to perform recommended tasks, as described below
with reference to FIG. 5.
[0032] Once identified, the activity can be used to track an
employee's progress with respect to a particular procedure being
performed. Specifically, a list of guidelines or steps for
performing activities is stored in a database. For instance,
returning to the above laboratory example, the guidelines for
protein purification can be accessed from a database and the
employee can be monitored to ensure that each of the predetermined
steps are correctly performed. The guidelines for protein
purification include: 1) prepare lysing buffer with
mercaptoethanol; 2) suspend cells containing protein of interest in
prepared buffer; 3) lyse cells; and 4) perform Nickel purification.
Based on the contextual data and identified activity, the employee
appears to be performing the first step of preparing a buffer for
protein purification. Thus, the next activity to be performed by
the employee should be suspending the cells in the buffer and
recommendations for performing the activity can be provided, as
described in detail below.
[0033] In a further embodiment, distributed activity detection can
be used to identify activities performed by the user. Distributed
activity detection helps offset some of the processing typically
required by a server and can result in faster and more accurate
identification of an activity using high frequency data. First,
contextual data is collected for a user via one or more mobile
computing devices. Features are extracted from the data to generate
a feature vector. The feature vector is then compared with one or
more activity models stored on at least one of the mobile computing
devices and a similarity measure is determined for each model. If
one of the models satisfies a predefined amount of similarity to
the feature vector, an identification label for the activity
associated with that model is assigned to the feature vector.
However, if none of the models satisfy the similarity, the user is
requested to assign an activity label to the activity represented
by the feature vector and the activity label is transmitted to a
server with the feature vector for training a new model. Once
trained, the server transmits the new model to the mobile computing
device for running. Distributed activity detection is described in
further detail in commonly-owned U.S. patent application Ser. No.
15/203,764, entitled "Computer-Implemented System and Method for
Distributed Activity Detection," Docket No. 20150448US01, filed on
Jul. 6, 2016, pending, the disclosure of which is incorporated by
reference.
[0034] Once an activity performed by the service provider is
identified, one or more recommendations can be identified for
providing to qualified users to assist with the activity. The
recommendations can be identified by looking at all possible tasks
that can be performed to make completing the activity easier,
faster, or in compliance with a set of guidelines. In one
embodiment, the recommendations can be identified using a
mixed-model recommender, as described in commonly-owned U.S. Patent
Application Publication No. 2009/0077057, pending, the disclosure
of which is incorporated by reference. FIG. 4 is a flow diagram
showing, by way of example, a process 50 for identifying
recommendations for providing to one or more recipients using a
mixed-model recommender. A database of tasks is accessed and each
task is processed (block 51) as a candidate task for
recommendation.
[0035] The tasks can be created and identified using, for example,
a web application based simulator. A user can establish a
contextual situation represented by a set of activity signals, by
selecting actions for input into the simulator to define the
situation, such as a chemical spill. The actions can be input via
drop down boxes or by dragging and dropping each action. For
instance, for a chemical spill, the actions can include presence in
a room with chemicals and the sound of breaking glass.
Subsequently, the user can select one or more tasks for performance
when the particular situation or activity is identified by
selecting one or more actions from a list. Each task can be
represented via a vector of relevant factors and annotated with
metadata, such as links to training materials, instructions, and
helpful hints, as well as other types of metadata. An exemplary
task vector might contain numeric representations of various
properties of the task, for example, requires experience with
chemical splits (1.0), requires use of fume hood (1.0), involves
chemical mercaptoethanol (1.0), with all other task vector values
being 0. The task vector can include a potentially large number of
possible values, most of which will be 0 for any given task, but
which together represent the whole of the task space in which the
user is working. An exemplary activity description vector will
contain a number of values in the same sense, but the labels for
the slots in the vectors do not need to be the same, and the slots
can represent low-level activities. Machine learning, such as SVM,
can be used to relate lower-level contextual situations represented
by the activity vector to a set of tasks which could be performed
given the contextual situation and an available training set of
activity-task training pairs, generated using the web-based
simulator.
[0036] Given the set of likely task recommendations, one or more
task recommendations can be produced. Recommendable tasks are
scored as candidates to determine which recommendations are most
relevant to the identified activity. In one embodiment, the
candidate recommendations to be scored can be identified via the
simulator upon determination of the employee's activity. The
recommendation scoring can occur via four scoring models, including
collaborative filter scoring, soft query model scoring, content
preference model scoring, and distance model scoring are used to
generate a list of recommendations. Scoring by the individual
models can occur in the order shown in FIG. 4, in a different
order, or simultaneously. The recommendations can include pushing
textual or other information to the user, contacting another user
for help in performing a task, or a suggestion for a supplemental
task, as well as other types of recommendations.
[0037] For each candidate task recommendation, a collaborative
filtering score can be determined (block 52) by matching the
employee to a large pool of users based on information such as age,
gender, income level, education, location, and occupation, as well
as other personal factors. Subsequently, recommendations for the
employee can be identified based on preference information of the
other users in the pool. The soft query model score is determined
(block 53) for the candidate task by considering the employee's own
preferences, which can be stored in the employee profile. Then, the
soft query model score is assigned based on that employee's
preferences. The context preference model score is determined
(block 54) by boosting recommendations that are semantically
related to the employee's interests, as determined, for example, by
the employee's online search history. The distance model score is
determined (block 55) by determining the employee's location using
GPS coordinates and estimating the employee's range of motion in
relation to a location or range of locations at which a task
recommendation could be performed. Subsequently, the distance model
score is assigned to each recommendation based on the employee's
proximity to the place where that recommendation would be
performed.
[0038] Once the collaborative filtering, soft query, content
preference, and distance model scores have been computed for the
candidate recommendation, the scores are summed (block 56) for a
total recommendation score for each task recommendation. A total
recommendation score is determined (block 57) for each candidate
task recommendation until no more candidate task recommendations
remain for scoring. Subsequently, the task recommendations with the
highest scores are selected (block 58), as most relevant to the
identified activity. The selected task recommendations are then
provided to one or more recipients, such as the employee or other
employees, as recommendable items, as discussed above with respect
to FIG. 2. The selected task recommendations can include all tasks
having a recommendation score above a predetermined threshold or a
predetermined number of recommendations with the highest
recommendation scores. Other methods for selecting the task
recommendations are also possible, such as via a rule-based
recommender.
[0039] Upon identifying appropriate tasks for recommendation,
qualified recipients must also be identified for providing with one
or more of the recommended tasks. FIG. 5 is a flow diagram showing,
by way of example, a process 60 for identifying recipients. For
each recommended task (block 61), a determination is made as to
whether the employee is qualified (block 62) to perform such task.
Qualification of the employee can be determined based on a
comparison of the selected task with the employee's profile. For
example, returning to the above laboratory example, the
recommendations include running the proteins through a further
purification process using high-performance liquid chromatography,
performing an SDS-PAGE gel to visualize the isolated protein, and
cutting extra tags off of the isolated protein after nickel
purification using an enzyme were identified. Based on the work
experience and education listed in the profile, the employee is
determined to be qualified to perform each of the recommended
tasks.
[0040] If the employee is qualified (block 62), the employee is
added to a list of recipients for that task, but if not, the
employee is not added to the recipient list. After consideration of
the employee, one or more predetermined groups of employees can be
considered and selected (block 64). Each group can be selected
based on attributes, such as work experience and skills, associated
with that group. For example, groups can be formed for employees
sharing common training, skills, or experience, as well as many
other attributes.
[0041] Subsequently, the common attributes shared by the members of
each group are compared with the recommendation to determine
whether the members of one or more groups would be qualified to
perform the task associated with the recommendation. If qualified
(block 64), the employees in the group are added (block 65) to the
list as recipients. However, if none of the groups are qualified,
no further users are added to the list. Once the recipients have
been determined for one of the selected recommendations, the next
recommendation is analyzed (block 66). Once all of the selected
task recommendations have been considered, the recommendations are
grouped (block 67) for sending to the intended recipients. The
recommendations can be sent via text message, email, voice message,
Instant Message, or a social networking messaging service. Other
methods for transmitting recommendations are possible.
[0042] If no qualified employees can be identified for one or more
task recommendations, step-by-step instructions can be provided to
the employee, or just-in-time training and learning can be enabled
to ensure that the employee properly and safely performs the
recommended task with guidance. In addition to or alternatively, an
unqualified determination can be used to initiate training of the
employee, including providing a notification to enroll the employee
in a training program, provide the employee with printed training
material, or schedule a meeting between a qualified employee and
the unqualified employee.
[0043] Returning to the above identified example, the researcher is
monitored during the different steps of the protein purification
process. While preparing the buffer, the researcher spills some of
the mercaptoethanol, which is considered toxic. Mercaptoethanol can
cause irritation of the nasal passages and respiratory track if
inhaled, as well as skin irritation, vomiting, and potentially
death for cases of extreme exposure. Once identified as a currently
occurring activity, one or more recommendations are identified,
including evacuating the laboratory room, warning other employees
in the laboratory room, such as via an alarm, turning on fans, air
conditioning units, or forced-air heating units, and cleaning up
the spill.
[0044] Upon review of each of the recommendations, the researcher
is determined to be qualified to perform all of the recommended
tasks, except for the toxic spill clean-up. However, a group of
other employees that are qualified to perform mercaptoethanol
clean-ups is identified and the clean-up recommendation is provided
to each of the employees in that group, while the other
recommendations are provided to the employee. In one embodiment, at
least one of the members of the group can select the clean-up
recommendation to notify other members in the group that he will
perform the clean-up. In a further embodiment, the members of the
group can access a chat system associated with the task
recommendation system to communicate with one another to determine
who can and will perform the clean-up. Further, the member
performing the clean-up can be monitored to ensure that the
clean-up occurs according to standard procedure.
[0045] Additionally, a chat system can be used to provide messages
to the employees and groups of employees, as well as allow the
employees to communicate with one another. Further, a watcher or
artificial agent can monitor the communications and inject
contextually relevant information into the channel for the
employees. The tasks can also be provided via the chat system and
employee selections of the tasks can be displayed.
[0046] In a further embodiment, a predicted outcome of an activity
performed by a user can also be determined, as described in detail
in commonly-owned U.S. patent application Ser. No. 15/203,740,
entitled "Computer-Implemented System and Method for Predicting
Activity Outcome Based on User Attention," Docket No. 20141587US01,
filed on Jul. 6, 2016, pending, the disclosure of which is
incorporated by reference, prior to identifying and requesting
qualified individuals to assist the user. For example, upon
identifying an activity performed by a user, a connection between
the user and a remote expert is made such that the remote expert
can provide assistance to the user. During this connection, actions
of the service provider are monitored to determine the user's
subject of focus. Based on the determined focus subject, an outcome
for completion of the service is determined and further assistance,
including identifying and assigning at least one qualified
individual to assist the user.
[0047] While the invention has been particularly shown and
described as referenced to the embodiments thereof, those skilled
in the art will understand that the foregoing and other changes in
form and detail may be made therein without departing from the
spirit and scope of the invention.
* * * * *