Patent Application
20150242645
More and more people are expressing and demonstrating their interest in having computers understand them and provide personalized assistance tailored to their specific needs and context. Of course, to provide personalized assistance that is tailored to the specific needs and context of a person, the assisting process must be aware of many aspects of the person, i.e., his or her personal information. Indeed, the more aspects of the person a process knows, the better that process is in personalizing information for the person.
Quite often, the computing device that a person most often has in his or her possession is a mobile computing device, e.g., a smartphone, a tablet computing device, or the so-called “Phablet”—the category of computing devices that are a hybrid of a smartphone and a tablet computing device. Generally speaking, mobile computing devices have less processing capabilities and are operating with limited power resources. While a person may often believe that he/she may have many services and applications running on a desktop computer at one time, in the context of mobile computing devices this is not feasible. As such, the services and apps executing on mobile computing devices must be managed, especially in the context of providing personalized assistance.
The following Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. The Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
According to aspects of the disclosed subject matter, systems and methods of a personal daemon, executing as a background process on a mobile computing device, for providing personal assistant to an associated user is presented. Also executing on the mobile computing device is a scheduling manager. The personal daemon executes one or more personal assistance actions on behalf of the associated user. The scheduling manager responds to events in support of the personal daemon. More particularly, in response to receiving an event the scheduling manager determines a set of apps that are responsive to the received event and from that set of apps, identifies at least a first subset of apps for execution on the mobile computing device. The scheduling manager receives feedback information regarding the usefulness of the executed apps of the first subset of apps and updates the associated score of each of the apps of the first subset of apps.
According to additional aspects of the disclosed subject matter, a mobile computing device configured to provide personal assistance to a user associated with the device is presented. The computing device comprises a processor and a memory, where the processor executes instructions to provide personal assistant to the associated user in conjunction with additional components. These additional components include a personal daemon and a scheduling manager. The personal daemon is configured to operate in the background of the mobile computing device and executes one or more personal assistance actions on behalf of the associated user. The scheduling manager is configured to, upon receiving notice of an event, determine a set of apps that are responsive to the event. Additionally, the scheduling manager identifies and executes a first subset of apps of the determined set of apps on the mobile computing device. The first subset of apps are determined/identified according to a score associated with each of the apps of the set of apps, where the first subset of apps includes apps with an associated score that meets or exceeds a predetermined threshold. In addition to executing the apps of the first subset of apps, the scheduling manager obtains feedback information regarding the usefulness of the executed apps on the mobile computing device. Based on this feedback, the score associated with each of the apps of the first subset of apps is updated for further reference.
The foregoing aspects and many of the attendant advantages of the disclosed subject matter will become more readily appreciated as they are better understood by reference to the following description when taken in conjunction with the following drawings, wherein:
For purposes of clarity, the term “exemplary” in this document should be interpreted as serving as an illustration or example of something, and it should not be interpreted as an ideal and/or a leading illustration of that thing. The term “personal information” corresponds to information, data, metadata, preferences, behaviors, of the associated user, as well as rules for interacting with the user. Generally speaking, personal information is information about the associated user that represents some aspect of the user. The personal information may comprise data such as (by way of illustration and not limitation) gender, age, education, demographic data, residency, citizenship, and the like. Personal information may also comprise preferences and interests, expertise, abilities, and the like. Still further, personal information may comprise rules (including rules established by the associated user as well as rules that are learned and/or inferred through analysis as described below) for interacting with the associated user in providing personal assistance.
The term event, as received and acted upon by the On{Event} framework described below, refers to a picked up by an executing service (such as the On{Event} framework) that originates from either hardware sensors, software instructions, remote devices, detected conditions, and the like.
One solution in providing personalized assistance could be to deploy an online service that can provide personalized assistance to a large number of subscribers by deploying a large numbers of computers and/or processors that gather, store, collate, analyze and manipulate large amounts of data gathered from all over the world. In this monolithic model, subscribers wishing to receive personalized assistance and/or recommendations provide various items of personal information to the online service and, typically, further permit the online service to monitor numerous aspects of the subscribers' lives to learn additional personal information about them. Nearly every activity a subscribers might take (especially with regard to their computer) is captured and analyzed to identify addition personal information, these activities including but not limited to online behaviors, purchases, preferences, affiliations, banking information, etc. The online service then deploys various processes to provide personalized assistance, based on the amassed personal information that it gathers and maintains of its subscribers.
Of course, running a massive monolithic online service as described above is expensive. In order to keep such a large online service operational, the online service must have a revenue stream. Generally speaking, however, subscribers/individuals want the personalized assistance for free. Rather than directly charging the subscribers a fee for the personalization service, the online service turns to monetizing the personal information of its subscribers. A common phrase for this monetization is “ad-funded” or “vendor-funded.” The online service monetizes the personal information of its subscribers by identifying individuals among its subscribers having various traits, interests, demographics, and attributes (as determined by the personal information that the online service has received and learned of its subscribers) and monetizing the identified information by placing advertisements to those individuals on behalf of advertisers. Of course, selling advertisements directed to its subscribers is only one way in which the monolithic online service (as described above) can monetize the personal information of its subscribers. Alternatively, the online service may simply sell contact lists and/or information.
Subscribers are often delighted to receive personalized assistance, so much so that they tolerate the advertisements that are frequently presented to them. Moreover, they are largely unaware of and would be extremely uncomfortable with how much of their personal information the online services possesses and monetizes/exposes to third parties (e.g., advertisers, vendors, organizations, etc.) Of course, an online service might placate its subscribers by telling stating that it will do no harm to its subscribers, yet the online service is conflicted: the online service generates revenue by providing personal information of its subscribers to third parties (whether by advertisements, selling contact lists, etc.). Moreover, the more specific the personal information that is provided to third parties, the greater the monetary reward is for the online service. Unfortunately, the more specific personal information that is exposed, the greater the risk and the more potential for abuse to the person or persons of the exposed personal information.
Of course, even without considering the risk of exposing personal information to known third parties, by simply storing substantial personal information for a large number of subscribers an online service creates an inviting, enticing target for identity thieves. So, while the level of personalized assistance can be directly correlated to the amount of personal information that is known of a person, the personal security of that person (as posed by the risk of exposure or misuse the personal information) is also a function of the amount of personal information of the person that the online service possesses. As shown in
In contrast to a monolithic online service and according to aspects of the disclosed subject matter, a personal daemon operating on a person's own computing device is presented. By way of definition, a “daemon” is a process or thread of execution, run on a computing device that is executed in the background of the computing device rather than being executed under the direct control of a computer user. However, while a daemon executes in the background of the computing device, a computer user can interact with a daemon and, through the interaction, direct the activities of the daemon. A “personal daemon” is a daemon that has access to, acquires, infers, maintains, and acts upon personal information of a computer user in providing personalized assistance. A personal daemon monitors numerous aspects of an associated user's activities to identify, infer, and/or learn additional personal information (when and where available) regarding the user as well as inferring and learning rules for acting on the user's behalf, i.e., providing personalized assistance to the user. Additionally, a personal daemon may learn and/or confirm personal information, particularly in regard to inferred information and/or rules for acting on the user's behalf, regarding the user through dialog and other interaction with the user, including confirming previously derived inferences regarding the user, requesting user preferences and other personal information, and the like.
The phrase “personal assistance,” in the context of a personal daemon providing personal assistance to the associated user, should be interpreted as carrying out one or more actions on behalf of the user based. Typically, though not exclusively, the personal assistance is triggered by one or more events related to aspects of the user's current context. By way of example and not limitation, the one or more actions of personal assistance may include: providing a recommendation to the user that the user take a particular action; obtaining data and/or services on the user's behalf; confirming with the user the inference of personal information from analysis of the user's activities; confirming with the user authorization for the personal daemon take an action on behalf of the user; providing a notification to the user regarding one or more events; providing alternatives to current user activities; recommending a venue; executing an action on behalf of the user on the computing device; recommending alternative and/or related activities or items; and the like. As will be discussed in greater detail below, a personal daemon provides personal assistance to the user based on rules, personal information of the user, and/or the current context of the user.
Unlike monolithic online service option that gathers and monetizes personal information of its subscribers, according to aspects of the disclosed subject matter a personal daemon does not share the associated user's personal information with other, third-party entities, except for and according to explicit direction by the user. A third-party entity corresponds to any entity not owned and/or responsive only to the associated user.
According various embodiments of the disclosed subject matter, the personal daemon operates on the user's computing device solely for the benefit of the user. Advantageously, the personal daemon is not conflicted by the need to monetize the user's personal information to support its operation or other purposes of an external, third-party entity. Accordingly, the personal daemon enjoys a position of intimate trust by the user and can be viewed as a computer-based extension of the user. Indeed, in a real sense the associated user may refer to the relationship as a “we” relationship, i.e., me and my own personal daemon. As a consequence of this high level of trust, the user is more inclined to provide the personal daemon with a greater degree of access to all information related to the associated user and his/her use of a mobile device, including personal and/or confidential information. For example (for illustration and not limitation), because the personal daemon does not share personal information of the associated user with others, the user may be willing to permit the personal daemon to read/scan the emails of the user, have access to and monitor the user's interactions on a social network, track the user's online purchase history, maintain the user's passwords, analyze all files and data streams on the mobile device, and the like. By instilling this higher level of trust in the associated user, and obtaining access to a greater degree of personal information, a personal daemon enhances the level of personalized assistance that can be provided to the user. As will be set forth in greater detail below, based on the enhanced level of access to personal information, through an enrichment cycle of inferring the associated user's preferences and choices, and learning rules of behavior in a given circumstance, and also validating those inferences, the personal daemon becomes an extension of the associated user, reflecting the associated user's personality and providing complimentary personal assistance. Indeed, over time the personal daemon “grows,” becomes more familiar, understands and knows more detail regarding the associated user, and is better able to provide personal assistance.
Turning to
The network environment 200 also includes a network 210 by which the user's computing device 202 (by way of components, applications, apps, etc.) can communicate with and access network accessible devices and/or online services connected to the network, including (by way of illustration and not limitation): one or more other user computing devices, such as computing device 212 associated with user 211; social networking sites, such as social networking site 218; online network services, such as a search engine 216; shopping and/or commerce sites, such as shopping site 214, and the like.
According to aspects of the disclosed subject matter, a personal daemon 204 is configured to operate on the “edge of the cloud,” meaning that the personal daemon operates on the user's computing device 202, with or without connectivity to the network 210. When connectivity to the network 210 is available (via the connection of the computing device 202 to the network), the personal daemon 204 executing on the computing device can access data and services for use in providing personal assistance to the user 201.
It is readily appreciated that many users have more than one computing device. Indeed, it is common for a user to have, by way of illustration, a smart phone, a tablet computing device, a laptop computer, and/or a desktop computer. Thus, according to aspects of the disclosed subject matter, a personal daemon operating on a computing device, such as computing device 204, may be configured to share personal information regarding the associated computer user 201 with a “sibling” personal daemon, i.e., a personal daemon associated with the same user that is operating on another computing device. In other words, as a personal daemon is an extension of one's own self, the personal information maintained by one embodiment of a personal daemon on a first computing devices can share the same and all personal information with another embodiment of the personal daemon (a sibling personal daemon) on another device. Moreover, as discussed below, sibling personal daemons may be configured to cooperate in order to provide personal assistance to the associated user.
Turning now to
The processor 402 executes instructions retrieved from the memory 404 in carrying out various functions, particularly in regard to executing a personal daemon 204 that provides personal assistance to the associated user. The processor 402 may be comprised of any of various commercially available processors such as single-processor, multi-processor, single-core units, and multi-core units. Moreover, those skilled in the art will appreciate that the novel aspects of the disclosed subject matter may be practiced with other computer system configurations, including but not limited to: personal digital assistants, wearable computing devices, smart phone devices, tablet computing devices, phablet computing devices, laptop computers, desktop computers, and the like.
The system bus 410 provides an interface for the various components of the mobile device to inter-communicate. The system bus 410 can be of any of several types of bus structures that can interconnect the various components (including both internal and external components). The computing device 400 further includes a network communication component 412 for interconnecting the computing device 400 with other network accessible computers, online services, and/or network entities as well as other devices on the computer network 210. The network communication component 412 may be configured to communicate with the various computers and devices over the network 108 via a wired connection, a wireless connection, or both.
The computing device 400 also includes executable apps/applications 416. As those skilled in the art will appreciate, an application corresponds to a collection of executable instructions that carry out (through its execution on a processor) one or more tasks on a computing device, such as computing device 400. Applications are typically, but not exclusively, executed at the direction of a user of the computing device. Applications combine features available on the computing device in carrying out the various tasks (as designed by the composition of the application.) While the term “apps” is sometimes uses as a shorthand name for applications, in the alternative an app similarly corresponds to a collection of executable instructions for carrying out one or more tasks. However, in contrast to applications, apps typically, though not exclusively, are directed to a limited set of tasks, often focused to a narrow topic/feature. As the scope of an app is typically more limited than that of an application, apps typically require a smaller footprint with regard to system resources and are often more suited for execution by computing devices of limited resources. While apps/applications 418 are typically stored in memory 404, for illustration purposes only they are called out separately from the memory 404.
The exemplary computing device 400 also includes sensors 418. Typically, sensors correspond to various hardware devices that sense particular events relating to the computing device 400. Sensors 418 may include, by way of illustration and not limitation, accelerometers, haptic sensors, capacitive sensors, audio sensors, optic sensors, timers, temperature sensors, power sensors (AC vs. DC sensors, voltage sensors, etc.), wireless signal sensors, geo-location sensors, magnetic sensors, altimeters, barometric sensors, and the like. Sensors may be based on communication information, such as internet routing data, HTTP request/response inspection, MAC addresses, cellular/wireless triangulation, and the like. As those skilled in the art will appreciate, a suitably configured computing device 400 may various combinations of hardware sensors 418. Moreover, these hardware sensors, as well as software sensors (as will be discussed below), are used in monitoring the user context via an On{Event} framework.
The exemplary computing device 400 further includes a personal daemon component 420 and an On{Event} framework 440. The personal daemon 420 is the executable component which, when executed, is the personal daemon 204 that provides the personal assistance to the user. As shown in
Regarding the On{event} framework 440, the On{event} framework (“framework”) is an extensible event/action framework, i.e., the framework detects events that occur with regard to the one or more sensors (including sensors 418 as well as events of other applications, apps, services, remote signals, and the like) and, in response, executes actions associated with the detected events on the computing device 400. It is extensible in that sensors can be added, including software sensors, and subscribers can subscribe to sensed events.
According to aspects of the disclosed subject matter, sensors are registered with the framework 440. By default or as part of an initialization process, all or some of the sensors 418 may be registered with the framework 440. Additionally, apps and/or applications (including the apps/applications 416) can register as software sensors with the framework 440, where a software sensor identifies the event (or events) that it will signal and the data that may be associated with the signaled event. Software sensors register with the framework 440 through a publisher interface 448. Sensors, including sensors 418 and software sensors, signal the sensed event through a sensor input interface 442. As indicated, upon receiving a sensed event, a rules executor 444 executes one or more actions on the computing device 400 associated with the sensed event, as established in the On{Event} data store 450. Apps and applications can register as subscribers to sensed/signaled events in the framework 440 by way of a subscription interface 446. In subscribing to a sensed event, an app or application, as well as the personal daemon 204 executing on the computing device 400, indicates the events for which the subscribing app, application, or daemon, wishes to be notified.
Regarding the On{event} framework 440, while those skilled in the art will appreciate that there may be a variety of alternatives to implement the framework, in one embodiment the framework 440 is implemented as a background service built upon the Node.js technology from Node.js Developers. The Node.js technology is extensible and robust such that it can interface with hardware sensors, such as sensors 418, as well as software sensors. Similarly, the personal daemon component 420 may be implemented upon the Node.js technology. Apps and applications, including apps/applications 416, interface with Node.js processes by way of JavaScript® code. While both the On{event} framework 440 and the personal daemon component 420 may be implemented using other technologies than Node.js, Node.js may be advantageously used as it enjoys a relatively small footprint on the host computing device, such as computing device 400, has configurations for deployment on a number of various operating system platforms, and JavaScript® programming languages enjoys broad support.
The exemplary computing device 400 further, still, includes a scheduling manager 460. As will be described in greater detail below, the scheduling manager 460 coordinates the execution of the apps and applications 416, as well as other services that may available to the computing device 400 but offered by another, external device or service. According to various embodiments, the scheduling manager 460 operates in conjunction with, or is implemented as a part of, the On{Event} framework 420 in responding to detected events from the various sensors 418 (including both hardware and software sensors), as well as in providing the contextual information/factors, services and data that the personal daemon 420 may need in providing personalized assistance to an associated user. In other, alternative embodiments, the scheduling manager 460 may be implemented as a component of, or independently in cooperation with, the personal daemon 420.
Regarding the various components of the exemplary computing device 400, those skilled in the art will appreciate that these components may be implemented as executable software modules stored in the memory of the computing device, as hardware modules (including SoCs—system on a chip), or a combination of the two. Moreover, each of the various components may be implemented as an independent, cooperative process or device, operating in conjunction with one or more computer systems. It should be further appreciated, of course, that the various components described above in regard to the exemplary computing device 400 should be viewed as logical components for carrying out the various described functions. As those skilled in the art will readily appreciate, logical components and/or subsystems may or may not correspond directly, in a one-to-one manner, to actual, discrete components. In an actual embodiment, the various components of each computer system may be combined together or broke up across multiple actual components and/or implemented as cooperative processes on a computer network.
Regarding the exemplary computing device 400, it should be appreciated that while the personal daemon is configured to interact with the associated user via the components of the computing device, generally speaking the personal daemon is independent of any particular configuration of computing device. Indeed, the personal daemon may be implemented on any suitable computing device and may communicate via displayed messages on a display component, text messages, audio and/or voice communications, haptic signals, and combinations thereof.
In addition to being implemented on one computing device, or across multiple computing devices via sibling personal daemons, a personal daemon may be further configured as a public mask to cooperatively operate in a joint computing manner with other services and/or processes in providing personal assistance to the associated user and/or performing analysis of user activity in order to learn and/or infer additional personal information regarding the user. However, the personal daemon operates in such a configuration (joint computing) according to the approval of the associated user and is restricted in sharing personal information with the joint processes/services according to the associated user's rules for doing so. According to aspects of the disclosed subject matter, in addition to sharing personal information with other third-party entities (e.g., processes and/or services) according to the associated user's explicit rules, the personal daemon may be configured to track what personal information is disclosed to these other entities. In tracking the disclosure of personal information to other entities, the personal daemon is able to inform the associated user what has been disclosed such that the user may identify limits to the amount of personal information that may disclosed. Indeed, an associated user may establish a limit of personal information that may be disclosed where after the personal daemon obfuscates any additional personal information that may be requested by any one entity or set of entities.
Turning now to
Upon receiving notice of the subscribed event 501 and according to information associated with the event, the personal daemon determines whether to provide personal assistance to the associated user in regard to the event, as indicated by circle 502. This determination is based on the information regarding the current context of the associated user, including personal information of the user, as well as rules previously established for the particular combination of events and context. For example, assume that the associated user is currently at work and the personal daemon knows this according to events received regarding the geo-location of the user's smart-phone/computing device according to rules and personal information in the personal daemon data store 432. Additionally, as a rule (which rule the personal daemon has either learned through inference, explicit direction from the user, or a combination of the two), the user typically does not take phone calls on his or her smart-phone while at work. However, yet another rule established with the personal daemon (again by inference, explicit instruction, or both) that the associated user will answer his or her smart-phone if it is during lunch or it is from specific individuals (such as a spouse.) Thus, at circle 502, when the subscribed event 501 is in regard to an incoming telephone call, the personal daemon receives the event and provides personal assistance to the user according to its rules regarding the user and the user's current context. Thus, if the information associated with the event indicates that the incoming telephone call is from an acquaintance, the personal daemon 204 according to its internal rules may immediately direct the incoming telephone call to an answering service. Alternative, if the information associated with the event 501 indicates that the incoming telephone call is from a spouse, the then personal daemon 204 can provide personal assistance to the associated user by permitting the incoming call to ring on the user's smart phone.
In addition providing immediate personal assistance, as indicated in circle 504, another part of the personal daemon 204 records information/data in regard the received event 501 in a user information data store 503. According to aspects of the disclosed subject matter, the personal daemon 204 records and logs events, contexts, and data associated with the user and the user's activities. This information is then used later in the analysis of user information, as indicated by circle 506, in learning and making inferences regarding additional personal information regarding the user, and in also learning rules for providing personal assistance to the user in regard to various events and contexts. This learning activity is described below in regard to routine 700 of
In analysis activity, as indicated by circle 506, the personal daemon 204 analyzes the information, as found in the user information data store 503, regarding the associated user, as well as and in light of the personal information know about the associated user in the personal daemon data store 432. The analysis activity uses neural networks, machine learning models, pattern recognition, and the like to infer information regarding the associated user. The analysis activity may further validate its inferences with the associated user by way of a confirmation dialog, though not necessarily in synchronicity upon deriving various inferences. The inferences may include static personal information (e.g., where the associated user works, the username/password of the user on a social networking site, etc.) or dynamic personal information (e.g., rules for responding to particular events, etc.) Based on the results of the analysis, the personal information regarding the associated user is refined and/or augmented in the personal daemon data store 432. Regarding the confirmation dialog, while this “dialog” may involve a user-interface control often referred to as a “dialog” or “dialog control”, in fact the dialog may take on any number of forms of a human-to-personal assistant interaction including, by way of illustration and not limitation, spoken instructions and/or confirmations, haptic feedback, gestures, and the like.
It should be appreciated that the analysis activity, as indicated by circle 506, will often include a confirmation dialog with the associated user. Typically, inferences are associated with some level of confidence. Except for the occasions in which the analysis activity produces an inference with near certainty of confidence, the personal daemon will often need to interact with the user in a confirmation type dialog, where inferences of personal information are presented to the user for confirmation or rejection. With regard to the example of determining the location where the associated user works, upon the first inference the personal daemon may engage the associated user with a dialog such as “Is this your work location?” The associated user may confirm or reject the inference. For example, the associated user may indicate that inferred location it is not a work location, but location of a school that the associated user attends. Through confirmation dialogs, as well as explicit review of inferred personal information and rules, the user exercises complete control over his/her personal information.
As part of or as a result of learning/inferring addition personal information regarding the user, and as part of providing personal assistance to the associated user (circle 502), the personal daemon may take proactive steps such as downloading data that may be relevant to the user. For example, as part of learning the location where the associated user works and based on personal information about the user that he or she likes a particular cuisine, the personal daemon may proactively download restaurant information surrounding the user's work location for future reference. Based on personal information regarding the associated user's work location and commuting habits, the personal daemon may associate a rule with a timer event to check the traffic situation for the commute and provide recommendations to the user when poor commuting conditions are present.
A distinct advantage that a personal daemon 204 has over a monolithic online service is that the personal daemon needs only maintain data relevant to the associated user. Maps, restaurants, calendars of events, etc. that are relevant to the associated user, as well as recording user related information such as search queries, browsing history, social networking profiles, etc., requires substantially less storage capacity than capturing and storing all information to serve a large number of users. Indeed, while the amount of information that may be of relevance to the user is not insignificant, in the context of the capacity of current computing devices, maintaining such information on a computing device is manageable. Additionally, as the personal daemon is situated on the “edge of the cloud,” to the extent that information is not currently available, is temporal, or is beyond the capacity of its host computing device, the personal daemon 204 can access such information online. For example, in the above-mentioned example of obtaining traffic information regarding the associated user's commute, the personal daemon may be configured to access the traffic information from an external source rather than retrieving and storing the information in the user information data store 503.
As indicated above, the personal daemon 204 does not share personal information regarding the associated user with other entities except as explicitly directed by the user. For example, the user may subscribe to a social networking site where access to the site is gained by supplying a password. Further, the personal daemon may establish rules for providing notice to the associated user whenever content is posted on the social networking site by a particular user. While the personal daemon 204 may associate a timer rule to periodically check on the social networking site for such posts, to access the information the personal daemon would need to provide the user's password and account information to the site to gain access. This activity, of course, is divulging the user's personal information. However, based on rules established by the personal daemon and according to explicit or inferred authorization by the associated user, the personal daemon may be authorized to divulge the personal information in providing personal assistance to the user. Of course, as the personal daemon 204 improves its efficiency at providing personal assistance to the user, actions may be taken by the personal daemon on the user's behalf when confidence, based on developed patterns and levels of trust, is high.
Of course, in the preceding example, the networking site may capture certain personal information regarding the user, e.g., user preferences, demographic information, geographic information, etc. Moreover, the networking site may also be vendor-funded such that advertisements are presented to the user when accessing the site. This, then, illustrates that while the personal daemon 204 does not share personal information regarding the associated user, the associated user is not restricted out of accessing and interacting with sites that may be vendor-funded through the disclosure of personal information, including the monolithic online sites discussed above.
At block 608, the personal daemon identifies personal assistance actions to be taken in regard to the received event. At decision block 610, if there are no actions to be taken, the routine 600 terminates. However, if there are actions to be taken, at block 612, the actions are configured according to current constraints. Generally speaking, configured the action according to current constraints comprises adapting the execution of the action according to the current context of the associated user. Personalization rules for adapting an action may be determined for the current context from the personal daemon data store 432. For example, if the received event is in regard to traffic congestion on the associated user's typical route home, the action may be to notify the user of the traffic congestion and suggest an alternative. Further still, the current context of the user may be that he/she is currently in a meeting and he/she should not be notified of non-emergency items during meetings. Hence, configuring the action according to current constraints would mean delaying the delivery of the suggested alternative route until the meeting is over. At block 614, the configured actions are executed in according to the various constraints, if any, from block 612. Thereafter, the routine 600 terminates.
As suggested above, one of the advantages of the trusted nature of a personal daemon is that it can use its access to the associated user's personal information to learn additional personal information through analysis activity (see circle 506 of
At block 704, one or more inferences are generated according to the analysis activity of block 702. These inferences generate additional and/or refined personal information of the associated user, as well as additional and/or refined rules for providing personal assistance to the user. As used herein, generating inferences regarding the associated user corresponds to inferring information about the user, rules for providing personal assistance to the user and the like. As indicated above, the generated inferences are made upon the various events and associated contexts regarding the user, both current and past, the user's interaction and behaviors with regard to the events, personal information of the user, as well as previously inferred rules for providing personal assistance to the user. As those skilled in the art will appreciate, inference can be employed to identify a specific context or action, or can generate a probability distribution over candidate states. An inference can be probabilistic, i.e., the inference may be associated with a probability or likelihood of occurrence with regard to a given state of interest based on a consideration of data and events. Inference techniques can be employed to generate higher-level events, e.g., rules for providing personal assistance from a set of recorded events and/or know or assumed data. Thus, inferences can result in the construction of new information or actions/rules from a set of observed events and/or stored event data. Advantageously, the inferences may be generated from events and data are not necessarily correlated in close temporal proximity, and/or from events and data that come from one or more sources.
Assuming that the generated inferences were determined as a probabilistic inference, at decision block 706 a determination is made as to whether or not any of the generated inferences are sufficiently “strong” that they do not need to be confirmed by the associated user. In one exemplary embodiment (for illustration and not limitation), an inference is sufficiently strong if the likelihood of occurrence is greater than a predetermined threshold value, e.g., a 95% estimated likelihood of occurring given the same (or substantially similar) events, context, and data. In an alternative embodiment, all inferences regarding the user's personal information or rules for providing personal assistance to the user that are generated in the analysis activity are confirmed with the user before implementation. Alternatively still, implementation and use of the inferred personal information and rules may conditionally occur, pending further confirmation, when the probabilistic likelihood exceeds a predetermined threshold, e.g., a 75% estimated likelihood of occurrence.
In the event that one or more generated inferences are not sufficiently strong, or that all inferences should be confirmed, at block 708 the inferences are confirmed with the user. Confirming inferences typically involves user interaction to confirm inferred personal data and/or rules for providing personal assistance. In confirming the generated inferences, the bases for the inference may be presented to the user, i.e., the event, personal information and context upon which the inferences was drawn. As will all of the personal information (including both data and rules for providing personal assistance) maintained by the personal daemon, the associated user has full control over this data such that he/she may delete, modify, confirm any and all parts of such personal information. This is important as an inferred rule may involve disclosing personal information regarding the user to another service or entity, in which case it is important for the associated user to be able to exercise control over such data (including stopping the dissemination of the data, permitting the disclosure in the particular context, and the like.)
Confirming inferences may involve a dialog between the personal daemon and the associated user (on the user's mobile device) in which the personal daemon iterates through the unconfirmed inferences, iteratively presenting each unconfirmed inference (and, potentially, the bases for its generation) and requests feedback from the user, including acceptance, modification, delaying a decision, or rejection. As an alternative to this dialog approach, or in addition to this iterative dialog approach, when conditions in which an unconfirmed inference may be used in providing personal assistance to the user, a dialog (i.e., a presentation to the user on the mobile device which may involve displaying information on a display screen, an audio presentation, signaling the user in some fashion, etc.) specifically directed to the unconfirmed inference at current issue may be presented to the user. For example, assuming that the personal daemon recognizes that the associated user is preparing to leave work for his/her home residence, a notice may be generated to the user from the personal daemon suggesting that the daemon check on the traffic status of the user's typical route home.
At block 710, after having confirmed the generated inferences or, the generated inferences are of sufficient strength that the user does not wish to confirm them, the associated user's personal information, including both data and rules for providing personal assistance, are updated. Thereafter, routine 700 terminates.
As those skilled in the art will readily appreciate, through this process of analysis of user activity and data, generation of inferences regarding the user, and confirmation of inferences, the personal daemon continually adapts itself to provide ever improving personal assistance. Continued application of these steps (analysis, inference, confirmation, and—of course—application of the information via personal assistance) refines the personal daemon to the point that it literally becomes an extension of one's self, reflecting the preferences and habits of the associated user.
Regarding routines 600 and 700, as well as routine 1000 described below and other processes, while these routines/processes are expressed in regard to discrete steps, these steps should be viewed as being logical in nature and may or may not correspond to any actual and/or discrete steps of a particular implementation. Nor should the order in which these steps are presented in the various routines be construed as the only order in which the steps may be carried out. Moreover, while these routines include various novel features of the disclosed subject matter, other steps (not listed) may also be carried out in the execution of the routines. Further, those skilled in the art will appreciate that logical steps of these routines may be combined together or be comprised of multiple steps. Steps of routines 600, 700, and 1000 may be carried out in parallel or in series. Often, but not exclusively, the functionality of the various routines is embodied in software (e.g., applications, system services, libraries, and the like) that is executed on computing devices as described in regard to
These routines/processes are typically implemented in executable code comprising routines, functions, looping structures, selectors such as if-then and if-then-else statements, assignments, arithmetic computations, and the like. The exact implementation of each of the routines is based on various implementation configurations and decisions, including programming languages, compilers, target processors, operating environments, and the link. Those skilled in the art will readily appreciate that the logical steps identified in these routines may be implemented in any number of manners and, thus, the logical descriptions set forth above are sufficiently enabling to achieve similar results.
While many novel aspects of the disclosed subject matter are expressed in routines embodied in applications (also referred to as computer programs), apps (small, generally single or narrow purposed, applications), and/or methods, these aspects may also be embodied as computer-executable instructions stored by computer-readable media, also referred to as computer-readable storage media. As those skilled in the art will recognize, computer-readable media can host computer-executable instructions for later retrieval and execution. When the computer-executable instructions store stored on the computer-readable storage devices are executed, they carry out various steps, methods and/or functionality, including those steps, methods, and routines described above in regard the various routines. Examples of computer-readable media include, but are not limited to: optical storage media such as Blu-ray discs, digital video discs (DVDs), compact discs (CDs), optical disc cartridges, and the like; magnetic storage media including hard disk drives, floppy disks, magnetic tape, and the like; memory storage devices such as random access memory (RAM), read-only memory (ROM), memory cards, thumb drives, and the like; cloud storage (i.e., an online storage service); and the like. For purposes of this disclosure, however, computer-readable media expressly excludes carrier waves and propagated signals.
According to aspects of the disclosed subject matter, numerous technical benefits are realized through the use of a personal daemon over alternative solutions. These technical benefits include, by way of illustration, improved latency in providing personal assistance as the personal daemon resides and executes on the “edge of the cloud” thereby eliminating the communication time with a remote service; correspondingly, local execution minimizes the bandwidth usage over the network; elements of the personal daemon (such as determining personal information and inferences by analysis) may be executed during non-peak processing times, i.e., when the user's demands on the computing device are low; enables personal assistance even when the computing device does not have network connectivity; makes use of preemptive caching of information based on predicted needs, which caching may be completed at times of network connectivity; low cost of implementation as the personal daemon operates within bounds of the associated user's own computing device; provides substantially improved data security as personal information is not shared with others without explicit rules to do so; and provides proactive augmentation of personal data and personal assistance without third party surveillance.
With all of the potential apps, applications, and services that could be used in providing personalized assistance to an associated user, when a personal daemon 204 is implemented on a mobile computing device, such as computing device 202, that is typically limited in computing resources (such as processing bandwidth, available power, current processing load, availability of resources, and the like), it is advantageous that only those services (or apps or applications) that are most likely to provide relevant data or features for the associated should be executed. Of course, the services or apps or applications on a mobile computing device that are most likely to provide relevant data or functionality is not a fixed set—the importance of rendered services, data, and functionality may change depending on the current context. For example, services that determine the highway congestion on the associated user's typical route to or from work are important during the user's commuting times and on those days that the user commutes to/from work. They are not particularly useful during the times that the associated user is not commuting to/from work, or on days that the associated user does not go to work. However, even on days or at times that the associated user does not typically commute, when a personal daemon 204 can determine that the associated user is actually commuting to work, executing the services that determine highway congestion would be useful for the personal daemon in providing personalized assistance to the associated user.
According to aspects of the disclosed subject matter, a scheduling manager 460 is used to coordinate the execution of apps, applications, and services (or, more aptly stated, in obtaining services), both on the local device (e.g., mobile computing device 202) and on remote devices including sibling computing devices. The scheduling manager 460 rewards those apps (used generally for apps, applications, and/or services) that are determined to provide useful data given the current context with improved scores, and penalizes the apps that are determined to not provide useful data given the current context with depreciated scores. Generally speaking, using the scheduling manager 406, apps associated with higher (improved) scores are executed more frequently given the same or similar set of contextual conditions as those apps associated with lower (depreciated) scores.
The determination as to whether an app provides useful data (or service or functionality) is made according to various criteria which can be generally classified as feedback and context. Elements of feedback include, but are not limited to: explicit user execution of an app and frequency of user execution; the type and amount of user interaction with the app; execution of the application and frequency of execution by the personal daemon 204; explicit evaluation feedback by the personal daemon, and the like. The context includes items such as the time of day, day of week, day of year, holiday and holiday season, geographic location, current weather conditions, whether the associated user is currently participating in a meeting, whether the associated is in a telephone conversation, whether the associated user is determined to be driving, wearable devices configured to sense user emotions, and the like. Further still, context may be based on more than a list of contextual criteria. Contextual items may be based on inferences: e.g., the concurrent execution of one or more apps with other events may be a strong indication of a context criteria that is not currently “listed.” Of course, as the personal daemon 204 identifies these inferences, the addition to the known set of contextual criteria may be augmented. These criteria are evaluated by a machine learning engine within (or in operation with) the scheduling manager 460. Information identifying the executed app (or service), feedback signals (more than one source of feedback may be obtained), and contextual information are provided to the machine learning engine and a score associated with the app is updated as appropriate. The information, including feedback and contextual information, is stored with information regarding the executed app for further reference. As those skilled in the art will appreciate, machine learning techniques, algorithms, and systems, also referred to as neural networks, are known in the art.
Turning to
Optionally and in addition to identifying a first subset of apps to be executed on the computing device 202, the apps/services executor 804 may also identify a second subset of apps of the initial set that may be executed by other devices, including sibling computing devices such as sibling computing device 206, for the benefit of the associated user 201 of the local computing device 202. These, too, are executed on behalf of the associated user 201.
It should be appreciated, of course, that the apps/services executor 804 of the scheduling manager 460 may further allocate processing resources to various other tasks (including tasks offered by apps, services, and/or various daemons executing on the local computing device—referred to as service processing) in addition to the first and second subsets of apps. Indeed, the apps/services executor 804 may budget or allocate time for service processing especially where such processing demands fall outside of the typical demands of executing an app. For example, the analysis or mining of a large data set may require substantial processing but can be performed as a background service and completed over time. Of course, the apps/services executor 804 may consider current processing loads and demands in budgeting for service processing.
In addition to executing the apps, the apps/services executor 804 monitors for and/or receives the feedback described above. The feedback includes, but is not limited to: explicit user execution of an app and frequency of user execution; the type and amount of user interaction with the app; execution of the application and frequency of execution by the personal daemon 204; explicit evaluation feedback by the personal daemon, and the like. The context includes items such as the time of day, day of week, day of year, geographic location, whether the associated user is currently participating in a meeting, whether the associated is in a telephone conversation, whether the associated user is determined to be driving, and the like. Of course, the feedback and context of execution of the apps (including whether the application was executed locally or remotely), is provided to the learning engine 806 where the scores associated with the executed apps give the current context is updated, if appropriate.
Through updating the scores of apps in view of a given context, the scheduling manager rewards apps for providing useful data, features, and/or services, penalizes apps when the data, features, and/or services provided by an app is less useful, and leverages this information to efficiently manage the execution of apps on the local computing device 202. Typically, a pattern will develop with regard to the execution of apps.
As discussed above and as can be seen in
Turning now to
At block 1010, a first subset of the apps is determined. As suggested above, this first set of apps is determined according to those apps having the highest scores, and meeting or exceeding a predetermined threshold in light of the current capabilities of the local computing device 202. At block 1012, the apps of the first subset are executed.
As indicated above, not all apps must be executed on the local computing device 202 but, instead, may be executed on remote computing devices (such as sibling computing devices). Accordingly, at block 1014, a second subset of apps is optionally identified for execution. As with the first subset of apps, these are apps that meet or exceed a predetermined threshold—though potentially a different threshold, specific to apps executed on remote devices, than those that the apps in the first subset of apps must meet. At block 1016, the apps of the second subset of apps are executed on remote computing devices. More specifically, the apps of the second subset of apps are caused to be executed on remote computing devices, as the scheduling manager may or may not be able to directly execute the apps on a remote device.
At block 1018, the scheduling manager 460 obtains feedback regarding the success or usefulness of the executed apps. In other words, the scheduling manager 460 feedback of the executed apps as to whether they provided useful data given the current context (success) or whether they did not provide useful data (no success) given the current context with depreciated scores. This feedback information is provided to the learning engine 806 and at block 1020 the scores associated with the apps (where appropriate) given the current context of execution (including where the app was executed) is updated. Thereafter, the routine 1000 proceeds to block 1022 where the routine awaits the next event and repeats the steps described above.
As will be appreciated, the iterative process of rewarding those apps, applications and/or services that provide useful information, data, or functionality for a given set of contextual factors will begin to favor a subset of all available apps, applications and/or services over all others. However, those apps that have an associated lower score or, perhaps, no score at all, may actually be of great value to the associated user at certain times. To ensure that scheduling manager 460 continually tries to improve on the overall performance of scheduling apps for execution, various measures may be implemented. Of course, a first measure is that the associated user may explicitly cause the execution of a particular app. The feedback for explicit user execution may be sufficient to cause that the app's associated score be raised to the level that it can meet the predetermined thresholds for a given set of contextual factors.
A second measure to ensuring that that scheduling manager 460 continually tries to improve on the overall performance of scheduling apps is to include a responsive app (one that is identified as being responsive to the triggering event) that does not meet the predetermined threshold for execution on the local device when the processing load of the computing device is sufficient to allow the responsive app to execute with the other apps that meet the predetermined threshold.
Yet another measure is to associate a length of time to be delayed as a function of an app's associated score. In this, the app (who's associated score falls below the predetermined threshold) would be delayed by the associated length of time before it could be considered for execution by the scheduling manager 460. Still further, similar measures may be implemented with the apps for remote execution. Thus, the determinations of blocks 1010 and 1014 may be modified from the description above to include other apps that do not meet the associated predetermined threshold.
While various novel aspects of the disclosed subject matter have been described, it should be appreciated that these aspects are exemplary and should not be construed as limiting. Variations and alterations to the various aspects may be made without departing from the scope of the disclosed subject matter.
This application is a continuation-in-part of U.S. patent application Ser. No. 14/187,567, entitled “Local Personal Daemon,” filed Feb. 24, 2014, and which is incorporated herein by reference. This application is also related to U.S. patent application Ser. No. ______, entitled “Accelerated Training of Personal Daemons,” attorney docket no. 340916.01; U.S. patent application Ser. No. 14/265,656, entitled “Client-side Integration Framework of Services,” filed Apr. 30, 2014; and U.S. patent application Ser. No. 14/219,501, filed Mar. 19, 2014, entitled “Personal Daemon Interaction through Escalating Disclosure.”
| Number | Date | Country | |
|---|---|---|---|
| Parent | 14187567 | Feb 2014 | US |
| Child | 14265882 | US |
