KR20110104165A - Incremental cognition and reasoning apparatus and methods - Google Patents

Abstract

Incremental cognitive and reasoning devices and methods are provided.
Incremental cognition and reasoning devices and methods can minimize the amount of sensing data needed to embody inferred intentions, and infer the user's intentions in detail when inferring situational awareness and intentions, leading to practical services. You can.

Description

Incremental Cognition and Reasoning Apparatus and Methods

TECHNICAL FIELD The present invention relates to an incremental cognitive and reasoning apparatus and method for inferring intentions based on a user's situation, further recognizing the situation, and embodying the inferred intention.

Intelligent service is a technology that intelligently provides a service required by a user after recognizing the user's situation and intention. To this end, an intelligent service capable device continuously acquires user's behavior information through a sensor and infers the user's situation and intention. Existing intelligent services can learn the behavior information of the user, which changes dynamically with time, and infer the user's intention based on the learning result. When the intention of the user is inferred, the intelligent service predicts the user's actions in advance and provides the services necessary for each action.

However, the existing intelligent service capable device has a limitation in simultaneously acquiring or analyzing all the information necessary for inferring the user's intention as the use of the sensor spreads and the amount of sensing data increases. In addition, existing intelligent service capable devices have limitations in inferring intentions more accurately, and as a result, they cannot provide services suitable for inferred intentions.

In one aspect, a situation recognition module for recognizing the initial situation of the user by analyzing the initial sensing information collected from the user environment; And an intention reasoning module that infers a user's intention based on the recognized initial situation and requests the context awareness module for context information necessary to embody the inferred intention, wherein the context awareness module is the intention reasoning module. According to a request of the user, additional sensing of the user's situation by collecting and analyzing other sensing information other than the initial sensing information, and the intention inference module embodies the incremental recognition that embodies the inferred intention based on the additional recognized situation. And an inference apparatus.

The context awareness module generates a list of contextual information required to embody the inferred intention, collects and analyzes sensing information corresponding to the created contextual information, at the request of the intention inference module. You can be aware of the situation further.

The intention reasoning module provides the context awareness module with a list of contextual information necessary to embody the inferred intention, and the context awareness module collects and analyzes sensing information corresponding to the provided contextual information to the user. You may be aware of the situation.

The intention reasoning module may plan the service when all the situations corresponding to the list of the situation information are recognized.

The intention inference module infers the intention of the user based on an intention model upon receiving the initial situation recognized by the situation recognition module, and receives the additionally recognized situation based on the intention model. An inference engine that embodies inferred intent; An intention manager that manages recording, updating, and deleting of the intention inferred by the inference engine and the specified intention; And an inference device for providing the intention inferred by the inference engine to the intention management unit and requesting the context recognition module to provide context information necessary to embody the inferred intention.

The inference reasoning module may manage an initial occurrence time, an update time of the inferred intention, a priority given to the inferred intention, and resolution of execution of the inferred intention for resolving a conflict between a plurality of intentions. .

If the intention reasoning module fails to execute a service for the specified intention, the intention model may propose the inferred intention or another intention associated with the specified intention as an alternative intention with reference to an intention model.

An intention model database storing an intention model including a service that a device can provide, an intention to use the service, and a situation in which the intention occurs; And an intention memory for recording the inferred intent or the specified intent.

In another aspect, the step of analyzing the initial sensing information collected from the user environment, recognizing the initial situation of the user; Inferring a user's intention based on the recognized initial situation; In response to a request for context information required to embody the inferred intention, collecting and analyzing other sensing information other than the initial sensing information to further recognize the situation of the user; And embodying the inferred intention based on the additionally recognized situation.

In the recognizing step, situation information necessary to embody the inferred intention may be created, and sensing information corresponding to the list of the created situation information may be collected and analyzed to further recognize the user's situation.

The additionally recognized step may be provided with a list of contextual information necessary to embody the inferred intention, and may additionally recognize the user's context by collecting and analyzing sensing information corresponding to the provided contextual information. .

In the planning of the service, when all of the situations corresponding to the list of the situation information are recognized, the service may be planned.

The method may further include managing an initial occurrence time of the inferred intention, an update time, a priority given to the inferred intention, and information on whether to execute the inferred intention to resolve a conflict between a plurality of intentions.

If the execution of the service for the specified intention fails, the method may further include referring to the intention model and suggesting the inferred intention or another intention related to the specified intention as an alternative intention.

The method may further include recording, updating, or deleting the inferred intention or the specified intention.

Incremental cognitive and reasoning devices and methods are provided.

If initially inferred intentions are not specific enough to plan a service, there is provided an incremental cognitive and inference apparatus and method that can further recognize a user's situation and progressively specify previously inferred intentions.

In addition, there is provided an incremental cognitive and reasoning apparatus and method capable of efficiently analyzing sensing information by specifying intention step by step, materializing intent to a level at which practical service execution is possible, and linking to a substantial service.

In addition, by continually observing and incorporating previously inferred intentions, it is possible to provide a more suitable service to users in a timely manner.

In addition, when the service execution fails to meet the specified intention, it is possible to improve the user convenience by providing an alternative intention.

1 is a block diagram of an example of an incremental cognitive and inference device.
2 is a block diagram illustrating a context awareness module and an intention reasoning module.
3 is a flowchart illustrating a method of incremental cognition and reasoning of an apparatus for incremental cognition and reasoning.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of an example of an incremental cognitive and inference device.

Incremental cognition and reasoning apparatus 100 may materialize the inference result by gradually collecting and analyzing the data in detail according to the inference result. To this end, the incremental cognitive and reasoning device 100 includes a context aware module 110 and an intention reasoning module 120.

First, sensing information input to the situation awareness module 110 may be divided into user input data and sensing data. User input data is data that a user directly inputs to a device. The context awareness module 110 receives user input data through various input interfaces such as a text input unit, a GUI, a touch screen, and a microphone.

The sensing data is data sensed by hardware or software and may be collected through a sensor attached to the device 100 or may be collected through communication with a sensor of an external system. Data sensed by hardware means data collected by sensing physical environment such as location, speed, temperature, and light. Software sensed data is collected by using software such as electronic scheduler, email, message, life log, etc. It means data that becomes.

The situation is recognized by analyzing the sensing information and the user input information and corresponding to the situation defined in the Intent Model. The situation recognition methods include pattern recognition and language processing, and description logic inference may be applied to the corresponding context to abstract or materialize the situation. The intention model is stored in the intention model database 135.

The context awareness module 110 may analyze the initial sensing information collected from the user environment and recognize the initial context of the user. The situation recognition module 110 may recognize the situation by selecting the sensing information based on the intention model of the intention model database 135, convert the recognized situation according to the inference input format, and deliver the detected situation to the inference unit 131.

Situational awareness of contextual awareness module 110 may be divided into general awareness and focused awareness. General cognition is the process of recognizing a situation to discover new or initial intentions. In general cognitive process, sensing information is collected and analyzed at a general level. For example, the situation recognition module 110 may recognize the current activity of the user and the initial situation by identifying only the location of the user, the presence of a schedule, and the application being used. On the other hand, intensive recognition is a process of intensively recognizing a concept or elements related to the intention by paying attention to a specific intention. The intensive cognitive process is performed at the request of the inference unit 131. For example, when the initial inferred intention requires the detailed location of the user and the purpose of the current schedule, the context awareness module 110 requests and analyzes detailed sensing information necessary to recognize the location and the purpose. The context awareness module 110 may efficiently analyze the sensing information through a mechanism capable of intensive analysis according to necessity without analyzing all of a large amount of sensing information.

The intention reasoning module 120 infers the intent of the user based on the recognized initial situation, and may request the context awareness module for the context information necessary to embody the inferred intention.

The context awareness module 110 may additionally recognize the user's situation by collecting and analyzing other sensing information other than the initial sensing information according to the request of the intention reasoning module 120. The intention inference module 120 further embodies previously inferred intentions based on the perceived situation, and can plan a service corresponding to the specified intentions.

For example, the context aware module 110 may generate a list of contextual information required to embody an intention initially or previously inferred at the request of the intention reasoning module. The context awareness module 110 may identify elements by intention stored in the intention model database 135 and set the identified elements as the list of the necessary context information. The context recognition module 110 may additionally recognize the user's situation by selectively collecting and analyzing sensing information corresponding to the created context information.

As another example, the intention reasoning module 120 may provide the context awareness module 110 with a list of contextual information necessary to embody an intention initially or previously inferred, and request the provision of the necessary contextual information. The context aware module 110 may check the list of contextual information provided from the intention reasoning module 120 and collect and analyze sensing information corresponding to the identified contextual information to further recognize the user's context. The intention reasoning module 120, when the situation recognition module 110 recognizes all the situations corresponding to the list of the situation information, infers the intentions in detail based on all the situations, and plans a service corresponding to the specifically inferred intentions. can do.

2 is a block diagram illustrating a context awareness module and an intention reasoning module.

Referring to FIG. 2, the context awareness module 110 includes a context recognizer 111, a pattern recognizer 112, and a language processor 113.

The situation recognition unit 111 receives initial sensing information and recognizes a situation. The situation recognition unit 111 may recognize the user's situation by analyzing the sensing information collected from the user environment with reference to the intention model stored in the intention model database 135 to be described later. The context recognition unit 111 may combine the sensed information collected with the intention model to recognize the user's situation through at least one of pattern recognition, language processing, or a combination thereof. Through this, the perceived situation can be expressed abstractly or concretely. For example, after the situation recognition unit 111 recognizes the user's mobile phone and recognizes the mobile phone, the situation recognition unit 111 sees the situation and the conceptual structure defined in the intention model, and the upper concept of the mobile phone, that is, the abstracted concept, is the 'communication means'. It can be recognized.

In addition, the situation recognition unit 111 outputs the pattern data to the pattern recognizer 112, outputs the language data to the language processor 113, and outputs the pattern recognizer or the language processor 113 according to the type of sensing information input. Receive and collect the analyzed result as status information.

The pattern recognizer 112 inputs pattern data to analyze the meaning of a previously defined pattern, and outputs the analyzed result to the situation recognition unit 111.

The language processor 113 inputs language data to extract key keywords or situation information, and outputs the extracted key keywords or situation information to the situation recognition unit 111.

The intention reasoning module 130 may include a reasoner 131, an inference engine 132, an intent manager 133, a planner 134, and an intention model database ( 135, intention memory 136, rule / plan / library database 137, user model database 138, intent model manager 139, knowledge manager 140, and user model And a management unit (User Model Manager) 141.

The reasoning unit 131 may derive a service plan by receiving an initial situation recognized or additionally recognized situation from the situation recognizer 111. In detail, the inference apparatus 131 may transmit the input initial situation to the inference engine 132, receive the intention inferred by the inference apparatus 131, and deliver the intention manager 133.

When one intent is at a serviceable level, that is, when all the elements required for the service are satisfied, it is delivered to the planner 134 to request a service plan for realizing the intention.

When the inference engine 132 receives the initial situation recognized by the context awareness module 110, it infers a new intention of the user based on the intention model. The inference engine 132 may use a logical reasoning such as Description Logic reasoning, rule-based reasoning, or a probability reasoning method such as a Bayesian network, or a mixture of both methods. The inference unit 131 transmits the intention inferred by the inference engine 132 to the intention manager 133.

The intention manager 133 registers the intention initially inferred by the general cognitive process into the intention memory 136 and updates the intention embodied by the intensive recognition process in the intention memory 136.

Specifically, the intention management unit 133 notifies the reasoning unit 131 when the situation information is further needed to plan the service of the intention specified for the first inferred intention. The reasoning unit 131 may request the situation recognition unit 111 to provide the situation information necessary to specify the intention initially deduced by the reasoning engine 132. In this case, the reasoning unit 131 may receive a list of context information from the intention manager 133 and provide the context information to the context recognizer 111, or provide only the recognized intention to the context recognizer 111. The list of context information provided by the intention manager 133 may be stored in the intention memory 136 or a separate memory.

At the request of the inference unit 131, the situation recognition unit 111 collects and analyzes additional sensing information in order to recognize situation information necessary to specify the intention, and additionally recognizes the situation using the analysis result and the intention model. Can be. When the reasoning unit 131 receives an additionally recognized situation from the situation recognition unit 111, the inference unit 131 transfers the additionally recognized situation to the inference engine 132 as in the first inferred intention. When the inference engine 132 receives a situation additionally perceived by the situation recognition module 110 from the inference device 131, the inference engine 132 embodies previously inferred intentions using the additionally recognized situation and intention model. The intention manager 133 may update the specified intention in the intention memory 136.

The operation of the intention management unit 133 will be described as follows.

The intention manager 133 may manage to record, update, and delete new intentions and specific intentions deduced by the inference engine 132 at the request of the inference device 131. That is, the intention management unit 133 records the new intention inferred by the inference engine 132 in the intention memory 136, and the recorded intention is the situation recognition unit 111, the inference device 131, and the inference engine 132. Once specified, the written intent can be retrieved from the intent memory 136 and updated to the specified intent. In addition, the intention manager 133 manages the initial occurrence time, update time, priority of the inferred intention, and execution information of the inferred intention, and manages the lifecycle of the intention based on this.

Priority is a priority given to inferred intentions for resolving conflicts between a plurality of intentions. If a conflict between intentions occurs when the priority is not assigned, the intention manager 133 may request additional information, that is, additional context information, from the inference device 131 to resolve the conflict. As a result, the reasoning unit 131 may request the situation recognizing unit 111 to provide the situation information necessary to specify the intention currently being processed.

In addition, the intention management unit 133 determines that the initial intention is embodied to the level at which the service is possible, or when all the situation information necessary to embody the inferred initial intention is recognized and the intention is embodied, the service plan required for the specified intention is specified. In order to establish the inference to the speculative 131 conveys the intention specified. The intention manager 133 may determine that all necessary situation information is recognized, for example, when all service elements necessary to provide a service for intention are identified.

In addition, if the intention manager 133 fails to execute the service for the specified intention, the intention manager 133 may suggest the intention associated with the failed intention to the inference device 131 with reference to the intention model. In addition, the intention manager 133 may delete the intent recorded in the intention memory 136 when the service of the specified intention is completed.

On the other hand, if the intention is embodied to the level at which the service is available, the reasoning unit 131 provides the planner 134 with the specified intention and requests a service plan for realizing the intention.

The planner 134 receives the specified intention to derive the service plan.

The intention model database 135 stores an intention model including a relationship between a service that a device can provide, an intention to use the service, and a situation in which the intention occurs. Thus, the main components of the intention model include situations, intentions and services.

Situation defines the context or hierarchy between situations or situations in which user intention occurs. The situation is classified into a physical situation perceived by sensing time, space, speed, temperature, light, and the like, and a logical situation perceived through scheduling, email, a message, a life log, and the like.

Intent defines the user's purpose, linkages and hierarchies to achieve through the use of the service. Intent can have elements as association concepts to enable service execution. For example, the intention to 'travel' may be set in the intention model database 135 in association with an element (purpose, place, period, companion, transportation, expense).

A service defines a service that a device can provide, an association between the services, and a hierarchy. The device may be an incremental cognitive and reasoning device 100 or an electronic device in which the device 100 is installed in hardware and software.

On the other hand, the intention model may be built using Ontology, Semantic Network or rules. For example, if a general intention model is initially constructed and provided, a personalized intention model may be formed by performing machine learning for the user as the user uses the device. The general intention model is a model that derives and defines common intentions of users generally occurring at the point of service use. Personalization intention model is an evolutionary intention model that adapts to the environment of users and users through machine learning.

Intent memory 136 records the initial intent or materialized intent inferred by inference engine 132. Intent memory 136 stores intents that are currently in progress or being managed, and the stored intents may be gradually materialized over time. This is because, as described above, the reasoning unit 131 requests the situation recognition unit 111 for context information, that is, specific information required to embody the intention, and infers the situation recognition and intention more in detail.

The rule / plan library database 137 stores, in live form, the rules needed to infer the intent and the unit plan elements required for the service plan of the intent. The rule / plan library database 137 is used by the inference engine 132 and planner 134.

The user model database 138 stores and maintains a user information model that includes user profiles, preferences, or patterns that are used for personalization during intent refinement. Here, the pattern may be information extracted and processed from data accumulated while the user uses the device. The user model can be updated based on user feedback.

The intention model manager 139 manages generation or change of the intention model recorded in the intention model database 135. In addition, the intention model manager 139 provides an interface for managing an intention model to a user or an administrator, and a learning function is added to automatically update the intention model.

The knowledge manager 140 manages a library required for rules and plans

The user model manager 141 manages generation or change of a user model. In addition, the user model manager 141 may provide an interface for managing the user model to the user or the administrator, or may add a learning function to automatically update the user model based on the feedback of the user.

On the other hand, the incremental cognitive and inference device 100 may further include a service provision unit (150). The planner 134 searches for elements for providing a service planned for a user and provides a service.

In case of inferring situation awareness and intention using the above-described apparatus 100, the amount of analysis data sensed to be handled can be minimized, and the user's intention can be specified to a level at which the service can be executed to connect to a practical service. .

In addition, by using the apparatus 100 described above, the inferred result may be embodied by gradually collecting and analyzing the sensing information in detail according to the initial or previously inferred result.

For example, when the user receives a text of a suggestion to 'travel' from a friend, and the situation where the user searches for a travel place through the Internet is recognized by the situation recognition unit 111, the reasoning unit 131 may be used by the user. It can be inferred to have the intention of 'traveling'. 'Travel' is an abstract level of intent, so it offers only abstract services, such as recommending a travel destination or connecting a travel site. However, since the device 100 knows that the intention to 'travel' is possible by filling in the elements (purpose, place, period, companion, transportation, expenses), the service 100 collects additional information on those elements. can do. The collection of additional information may be obtained by observing the user's activity or by directly asking the user through an on-screen display (OSD). If all necessary elements necessary for realizing the intention are secured, the device 100 may plan a specific service.

3 is a flowchart illustrating a method of incremental cognition and reasoning of an apparatus for incremental cognition and reasoning. The incremental cognitive and reasoning device may be the device 100 described with reference to FIGS. 1 and 2.

In operation 300, the context aware module of the device may recognize initial sensing information of the user by analyzing the initial sensing information collected from the user environment. Step 300 may be a general cognitive process.

In operation 310, the intention inference module of the device may infer a user's intent based on the initial situation recognized in operation 300. The inferred intent can be registered, updated and deleted in the memory of the intention inference module.

In operation 320, the intention reasoning module may request context information module, that is, context information, that is, specific information necessary to embody intention (ie, initial intention) inferred in operation 310.

In operation 330, the context awareness module may additionally recognize the user's situation by collecting and analyzing other sensing information other than the initial sensing information according to a request of the intention reasoning module. Step 330 may be an intensive cognitive process.

In operation 340, the inference reasoning module may specify and update an initially inferred intention using the additional situation recognized in operation 340.

If it is determined in step 350 that the specified intention is embodied to the level of service availability, in 360, the intention reasoning module may plan a service for realization for the specified intention.

In operation 370, the intention reasoning module may search for and provide elements necessary to execute the planned service.

On the other hand, if it is determined in step 350 that the intention is not embodied to the level at which the service is possible, the intention reasoning module executes step 330.

If the execution of the service for the specified intention fails in step 380, in operation 390, the intention reasoning module may plan another service by suggesting another intention related to the specified intention as an alternative intention with reference to the intention model. The proposed alternative intention is passed on to 350 to determine if intent specification is necessary.

Methods according to an embodiment of the present invention can be implemented in the form of program instructions that can be executed by various computer means and recorded in a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

100: incremental cognitive and reasoning device 110: situational awareness module
120: intention reasoning module

Claims (16)

A situation recognition module for analyzing the initial sensing information collected from the user environment and recognizing an initial situation of the user; And
An inference reasoning module that infers a user's intention based on the recognized initial context and requests the context awareness module for context information necessary to embody the inferred intention
Including;
The contextual awareness module collects and analyzes other sensing information other than the initial sensing information at the request of the intention inference module to further recognize the user's situation, and the intention inference module based on the additionally recognized situation. Incremental cognitive and reasoning devices that embody inferred intentions. The method of claim 1,
The context awareness module generates a list of contextual information required to embody the inferred intention, and collects and analyzes sensing information corresponding to the created contextual information according to a request of the intention inference module. Incremental cognition and reasoning device that additionally recognizes. The method of claim 1,
The intention reasoning module provides the context awareness module with a list of contextual information needed to materialize the inferred intent,
The contextual awareness module collects and analyzes sensing information corresponding to the list of contextual information provided to further recognize the context of the user. The method according to claim 2 or 3,
The intention reasoning module, the incremental cognitive and reasoning apparatus for planning the service when all the situations corresponding to the list of the situation information is recognized. The method of claim 1,
The intention reasoning module,
Receiving the initial situation recognized by the context awareness module, infers the intention of the user based on an intention model, and upon receiving the additionally recognized context, embodies the inferred intention based on the intention model. Inference engine;
An intention manager that manages recording, updating, and deleting of the intention inferred by the inference engine and the specified intention; And
An inference device for providing the intention inferred by the inference engine to the intention management unit and requesting the situation recognition module to provide context information necessary to embody the inferred intention.
Incremental cognitive and reasoning device comprising a. The method according to claim 1 or 5,
The intention reasoning module,
Incremental recognition and reasoning apparatus for managing the initial occurrence time of the inferred intention, the update time, the priority given to the inferred intention and the execution information of the inferred intention to resolve conflicts between a plurality of intentions. The method according to claim 1 or 5,
The intention reasoning module,
Incremental cognition and reasoning apparatus, if an execution of a service for the specified intent fails, referring to an intention model and suggesting the inferred intent or other intent associated with the specified intent as an alternative intent. The method according to claim 1 or 5,
An intention model database storing an intention model including a service that a device can provide, an intention to use the service, and a situation in which the intention occurs; And
Intent memory for recording the inferred intent or the specified intent
Incremental cognitive and inference device further comprising. Analyzing the initial sensing information collected from the user environment and recognizing an initial situation of the user;
Inferring a user's intention based on the recognized initial situation;
In response to a request for context information required to embody the inferred intention, collecting and analyzing other sensing information other than the initial sensing information to further recognize the situation of the user; And
Embodying the inferred intent based on the additionally recognized situation
Incremental cognition and reasoning method comprising a. 10. The method of claim 9,
Recognizing the additional step,
Incremental recognition and reasoning method for creating the situation information necessary to embody the inferred intention, and additionally recognize the user's situation by collecting and analyzing the sensing information corresponding to the list of the created situation information. 10. The method of claim 9,
The further recognized step,
Incremental recognition and reasoning method that is provided with a list of the situation information necessary to materialize the inferred intention, and additionally recognizes the user's situation by collecting and analyzing the sensing information corresponding to the provided list of the situation information. The method according to claim 10 or 11, wherein
Planning the service,
Incremental cognition and reasoning method for planning the service when all the situations corresponding to the list of the situation information are recognized. 10. The method of claim 9,
Managing an initial occurrence time of the inferred intention, an update time, a priority given to the inferred intention, and resolving information on whether the inferred intention is executed in order to resolve a conflict between a plurality of intentions;
Incremental cognition and reasoning method further comprising. 10. The method of claim 9,
If the execution of the service for the specified intent fails, suggesting an alternative intent as the inferred intent or other intent associated with the specified intent by referring to the intention model
Incremental cognition and reasoning method further comprising. 10. The method of claim 9,
Recording, updating, or deleting the inferred intent or the specified intent
Incremental cognition and reasoning method further comprising. A computer-readable recording medium for recording a program for executing the method of any one of claims 9 to 15 on a computer.

Images