KR101931128B1 - Method and apparatus for evaluating trust of device in internet-on-the things environment, and computer-readable medium - Google Patents

Abstract

A method and apparatus are provided for evaluating the reliability of a device in an object internet environment. A method for evaluating reliability of a device in an object internet environment includes: a context information recognition step of recognizing context information of a situation in which a device is found in an object Internet environment; A situation information classification step of referring to the situation factors constituting the situation information and classifying the situation according to a machine learning algorithm; A basic reliability determination step of determining a basic reliability for the device by referring to a trust disposition of the user with respect to the classified status; And a final reliability calculation step of calculating a final reliability for the device based on the determined basic reliability using at least one of a user's direct experience with the device and a degree of a priori confidence. According to the present invention, according to the present invention, the SDP can be grasped by a knowledge base using adaptive probing, and assistance can be provided between a user and a service and a smart device based on background knowledge and analysis between a person and a device.

Description

[0001] The present invention relates to a method and an apparatus for evaluating the reliability of a device in an object internet environment, and a computer-readable medium,

The present invention relates to a method and an apparatus for evaluating the reliability of a device in an object internet environment and in particular to a method and apparatus for evaluating the reliability of a device based on a determined basic reliability using at least one of direct experience with a device and a degree of a priori confidence And to a method and an apparatus for calculating the same.

In the Internet On The Things (IoT) environment, multiple devices can be placed in familiar environments, such as the home or office, but in some cases they may approach unfamiliar environments. In a new environment, various types of IoT devices coexist. Different types of terminals communicate with different kinds of protocols, and it is important to understand what kinds of protocols each other uses to communicate with each other. In network communication, there is a service discovery protocol (SDP) for finding out what services each terminal provides to each other. However, there is a problem that legacy services and new services are incompatible because the existing SDP is updated and applied as a base technology of new services after SDP-based services are distributed. Although existing studies have introduced methods for creating run-time adapters to solve them, there is a precondition that specifications for new protocols must be provided in advance. Therefore, a technique has been studied to enable communication with the base SDP of the other party service through inference without specification of a new SDP.

In addition to this incompatibility issue, it is also an important issue in the Internet of things that IoT devices are increasingly exposed to personal information as they receive and communicate with other devices in new environments. The fact that the smart device is deeply involved in the user's life means that access to the user's personal information can be facilitated. Therefore, the reliability of Internet devices has become more important than ever, and the possibility of interacting with unfamiliar objects Internet devices anytime and anywhere is also increased. As the accessibility of these unfamiliar devices increases, the likelihood of a malicious user attacking the device and the risk of an end user being attacked by a malicious device increases. Moreover, as the Internet becomes more universal, the damage can continue to increase.

For example, if personal information is provided to provide personalized services, such personal information may be leaked and privacy may be infringed. In particular, as the importance of personal information is heightened, more attention needs to be paid to these risks. In order to prevent or minimize such damage, a method of evaluating the reliability of a user's Internet device has been studied. Furthermore, to reduce this risk, the nature of trust between people and devices needs to be understood and applied to specific assessment models.

In particular, the following issues should be considered in a situation where a combination of IoT and virtual world service is required in accordance with the universalization of smart IoT and the increase of service application services, and a combination in which reality and virtual world are required.

First, the reliability of people and things Internet devices is important.

Nitti, M., Girau, R., and Atzori, L., "Trustworthiness Management in the Social Internet of Things," IEEE Transactions on Knowledge and Data Engineering, vol. 26, no. 5, 2014 (hereinafter referred to as "Nitti") is a device for evaluating the reliability between devices. It is an object of the present invention, . However, since this method does not take into account the situational factors that influence the initial trust formation on the stranger object of the person, it is difficult to apply it to the interaction of the person as the subject.

Second, it is not easy to apply human-to-human research or device-to-device research directly between people and devices. In order to evaluate the reliability of strangers among others, in addition to the experience of interaction with people who have met at places similar to the place where they are currently located, A stereotype that represents the goal, interest, and features of a person and a method to utilize the similarity with the person who is facing the present are proposed. In other words, when the direct experience with the other party is insufficient, the indirect reliability evaluation is performed through the stereotype, and the change of the stereotype according to the change of the place is considered. However, this method is not applicable when evaluating the reliability of a person's unfamiliar device, because most devices have their roles and purposes fixed in advance and rarely changed. Therefore, it is difficult to apply this method to a new device because similar type devices having similar roles and objectives share similar characteristics and consequently can have the same stereotype and reliability regardless of the change of situation.

According to the prior art, a technique has been introduced to perform reliability evaluation around reliability and sociability used to define the relationship between devices. This means that the relationship of devices is defined by applying sociological factors and collects information of devices based on the defined relationship. Reliability is considered to be an important factor in the evaluation of how reliable the opinions of other devices in gathering information of the device should be made. However, the factors of this methodology are insufficient to change the result of reliability evaluation according to the change of situation based on the result of interaction between devices. Therefore, another factor must be considered to accurately assess reliability.

Also, I.-R. Chen, F. Bao, and J. Guo, " Trust-based Service Management for Social Internet of Things Systems, " IEEE Trans. Dependable Secur. Comput., Vol. 5971, no. c, pp. 1-1, 2015. describe how devices, like users belonging to a particular group of devices, can also classify devices based on similar interests or capabilities, Describe technologies that directly or indirectly collect and evaluate honesty, cooperativeness, and community-interest, and aggregate the results to determine trustworthiness. However, since there is no factor that takes into account changes in the situation, there is a need for additional factors as the accuracy of the reliability evaluation is lowered. In addition, as in these two studies, reliability computation methods based on SIoT inherently have the limitation that they can not guarantee full observability of the social network assumed by SIoT in the real world.

Methods for using profile similarity have been proposed to calculate human reliability according to the prior art. However, this method requires the exchange of profiles among users, so there is a high possibility that personal information is leaked, and there is a problem that it is practically difficult to receive a profile from a stranger. In order to solve this problem, stereo type based reliability evaluation methods have been proposed. Stereotype is a kind of type that represents similar objects by vectorizing shared features from objects sharing similar characteristics. It is a kind.

G. Huerta-Canepa, S. Han, D. Lee, and B. Kim, "A place-aware stereotypical trust supporting scheme," Proc. - 12th IEEE Int. Conf. Trust. Secur. Priv. Comput. Commun. Trust. 2013, pp. 821-828, 2013. introduces a technique for evaluating the credibility of an opponent based on their interaction experience with those who have the same stereotype. However, this assumes that people with the same stereotype will always behave the same, without considering the interaction with an unfamiliar object, so that people with the same stereotype always have the same confidence. The proposed method focuses on improving the accuracy of the reliability evaluation using the characteristics of the place and the device among the contextual characteristics to which the person and the device belong.

C. Burnett, T. Norman, and K. Sycara, " Bootstrapping trust evaluations through stereotypes, " Proc. 9th Int. Conf. Auton. Agents Multiagent Syst., Vol. 1, p. 8, 2010. In this paper, we propose a methodology that introduces placeability to existing stereotypes and changes its reliability accordingly. In the case of humans, the stereotype is variable depending on the situation, whereas the device is not so varied as to show a striking difference from stereotypes. This is because stereotypes and reliability can be similarly evaluated because the same type of devices have the same characteristics. Although these techniques have advantages in introducing placeability to existing stereotypes, they have limitations in that they can not yield variable results depending on the situation change in the reliability evaluation between a person and a device.

In this way, in existing reliability studies of people and people, when the diversity of individual is emphasized and studies in which individual diversity is respected are main, in evaluating reliability between human and device, While harmonizing the same device, it is necessary to focus on what elements are to be expressed differently in the reliability evaluation and how much of each element is harmonized with the situation. Therefore, a technique is required to evaluate the reliability according to the situation in the reliability evaluation process.

Therefore, there is a desperate need for a knowledge-based device reliability evaluation technique that enables a user to discover heterogeneous < RTI ID = 0.0 > IoT services < / RTI >

Nitti, M., Girau, R., and Atzori, L., "Trustworthiness Management in the Social Internet of Things," IEEE Transactions on Knowledge and Data Engineering, vol. 26, no. 5, 2014. I.-R. Chen, F. Bao, and J. Guo, " Trust-based Service Management for Social Internet of Things Systems, " IEEE Trans. Dependable Secur. Comput., Vol. 5971, no. c, pp. 1-1, 2015. G. Huerta-Canepa, S. Han, D. Lee, and B. Kim, "A place-aware stereotypical trust supporting scheme," Proc. - 12th IEEE Int. Conf. Trust. Secur. Priv. Comput. Commun. Trust. 2013, pp. 821-828, 2013. C. Burnett, T. Norman, and K. Sycara, " Bootstrapping trust evaluations through stereotypes, " Proc. 9th Int. Conf. Auton. Agents Multiagent Syst., Vol. 1, p. 8, 2010.

It is an object of the present invention to provide a technique for assisting interaction between a user, a service, and a smart device based on knowledge-based SDP identification technology using adaptive probing and background knowledge and analysis of reliability between a person and a device.

It is another object of the present invention to provide a technology for classifying social factors into relations of devices and evaluating reliability through direct and indirect experiences and opinions in formed groups.

에 따라 상기 최종 신뢰도를 계산하되, x는 평가자를 나타내고, y는 평가 대상을 나타내며, P(W ^x _y )는 x의 y가 긍정적으로 행동할 것이라는 x의 확신 정도를 나타내는 확률값이고, r ^x _y , u ^x _y , v ^x _y , a ^x _y 는 각각 x의 y에 대한 긍정적 경험, 불확실성, 개인적 경험, 및 스테레오 타입에 의한 확신 정도를 각각 나타내는 단계를 포함하는 것을 특징으로 한다. 바람직하게는, u ^x _y 및 a ^x _y 는

및

로 각각 표현되고, s ^x _y 는 x의 y에 대한 부정적 상호작용 경험의 횟수를 의미하며, ρ는 다른 사용자를 나타내고, t _y ^ρ 는 y에 대한 신뢰도이며, c _ρ 는 상기 t _y ^ρ 에 대한 상기 사용자의 신뢰도를 나타내는 것을 특징으로 한다.According to an aspect of the present invention, there is provided a method for evaluating reliability of a device in an object internet environment. A method for evaluating reliability of a device in an Internet environment for objects according to the present invention includes: a context information recognition step of recognizing context information about a situation in which a device is found in an object Internet environment; A situation information classification step of referring to the situation factors constituting the situation information and classifying the situation according to a machine learning algorithm; A basic reliability determination step of determining a basic reliability for the device by referring to a trust disposition of the user with respect to the classified status; And a final reliability calculation step of calculating a final reliability for the device based on the determined basic reliability using at least one of a direct experience and a degree of a priori confidence of the user with respect to the device. In particular, the status information recognition step may include recognizing the status information using a sensor included in the object Internet environment. The context information classification step may include generating a decision tree from the interaction experience of the user using an M5 model tree learning algorithm and adding the contextual factors to the decision tree Wherein the decision tree is learned using the interaction experience of the user and the contextual factors. In addition, the step of determining the basic reliability may include: forming a confidence matrix of the user based on the interaction experience of the user with respect to the classified situation using a ST (swift trust) technique; And determining a basic reliability for the device by referring to the trust substrate of the user for the situation classified according to the decision tree. The final reliability calculation step may be performed using SL (subjective logic)

, Where x represents an evaluator, y represents an evaluation object, P ( W ^x _y ) is a probability value representing the degree of assurance of x that y of x will act positively, and r ^x _y , u ^x _y , v ^x _y , and a ^x _y each include a step of representing positive experience, uncertainty, personal experience, and degree of assurance by stereotype for x of y, respectively. Preferably, u ^x _y and a ^x _y are

And

, S ^x _y represents the number of negative interaction experiences of _y for x, ρ represents another user, t _y ^ρ is the confidence for y, and c _ρ is the confidence level for t _y ^ρ And represents the reliability of the user.

According to another aspect of the present invention, there is provided an apparatus for evaluating reliability of a device in an object internet environment. An apparatus for evaluating reliability of a device in an object internet environment includes a situation information recognition unit for recognizing situation information on a situation in which a device is found in an object Internet environment; A situation information classifying unit for classifying the situation according to a machine learning algorithm with reference to situation factors constituting the situation information; A reliability determining unit for determining a reliability of the device by referring to a trust disposition of the user for the classified situation; And a final reliability calculation unit for calculating a final reliability for the device based on the determined basic reliability using at least one of a direct experience and a degree of a priori confidence of the user with respect to the device. In addition,

에 따라 상기 최종 신뢰도를 계산하고, x는 평가자를 나타내고, y는 평가 대상을 나타내며, P(W ^x _y )는 x의 y가 긍정적으로 행동할 것이라는 x의 확신 정도를 나타내는 확률값이고, r ^x _y , u ^x _y , v ^x _y , a ^x _y 는 각각 x의 y에 대한 긍정적 경험, 불확실성, 개인적 경험, 및 스테레오 타입에 의한 확신 정도를 각각 나타내는 것을 특징으로 한다. 더 나아가, u ^x _y 및 a ^x _y 는

및

로 각각 표현되고, s ^x _y 는 x의 y에 대한 부정적 상호작용 경험의 횟수를 의미하며, ρ는 다른 사용자를 나타내고, t _y ^ρ 는 y에 대한 신뢰도이며, c _ρ 는 상기 t _y ^ρ 에 대한 상기 사용자의 신뢰도를 나타내는 것을 특징으로 한다.Wherein the context information classifier is configured to recognize the context information using a sensor included in the object Internet environment, wherein the context information classifier classifies the context information by using an M5 model tree learning algorithm, Generating a decision tree, and applying the situation factors to the decision tree to classify the situation. In particular, the decision tree is learned using the interaction experience of the user and the contextual factors. In addition, the basic reliability determination unit may be configured to form a confidence matrix of the user based on the interaction experience of the user with respect to the classified situation using the ST (swift trust) technique, Wherein the final reliability calculation unit determines the basic reliability for the device by referring to the user's trust substrate with respect to the situation,

( W ^x _y ) is a probability value indicating the degree of assurance of x that y of x will act positively, and r ^x _y ⁽ _y ^x _y) , u ^x _y , v ^x _y , and a ^x _y are each a positive experience, uncertainty, personal experience, and confidence level of stereotypes for y of x, respectively. Further, u ^x _y and a ^x _y are

And

, S ^x _y represents the number of negative interaction experiences of _y for x, ρ represents another user, t _y ^ρ is the confidence for y, and c _ρ is the confidence level for t _y ^ρ And represents the reliability of the user.

A method for evaluating the reliability of a device in an object internet environment according to an aspect of the present invention can be implemented by a computer program comprising instructions executed by a processor stored in a computer readable medium.

According to the present invention, it is possible to grasp the SDP with the knowledge base using adaptive probing, and to assist the interaction between the user, the service and the smart device based on background knowledge and analysis between the person and the device.

Further, according to the present invention, reliability of a device can be classified by applying social factors to the relationship of devices, and can be calculated through direct and indirect experience and opinion within a formed group.

1 is a diagram illustrating an Internet environment of objects in which the present invention is implemented.
2 is a block diagram conceptually illustrating an operation of a method for evaluating reliability of a device in an Internet environment of objects according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating a method of evaluating reliability of a device in an object Internet environment according to an embodiment of the present invention.
4 is a diagram illustrating a history record of a specific device.
FIG. 5 shows an example of classifying learned situations according to the M5 model tree learning algorithm.
6 shows an example of a linear regression analysis model for estimating the reliability value.
FIG. 7 is a block diagram schematically illustrating an apparatus for evaluating reliability of a device in an object Internet environment according to another aspect of the present invention.

In order to fully understand the present invention, operational advantages of the present invention, and objects achieved by the practice of the present invention, reference should be made to the accompanying drawings and the accompanying drawings which illustrate preferred embodiments of the present invention.

Hereinafter, the present invention will be described in detail with reference to the preferred embodiments of the present invention with reference to the accompanying drawings. However, the present invention can be implemented in various different forms, and is not limited to the embodiments described. In order to clearly describe the present invention, parts that are not related to the description are omitted, and the same reference numerals in the drawings denote the same members.

1 is a diagram illustrating an Internet environment of objects in which the present invention is implemented.

Referring to FIG. 1, it can be seen that the medical system, the education system, the transportation system, the logistics system, the leisure system, and various control systems are integrated and interacted with each other in the object Internet environment. The Internet has been used as a place where human beings can share information with producers / consumers of information. In the Internet of things, even things around us, such as household appliances and sensors, are connected to the network, Can be shared. In other words, devices supporting IoT are expected to increase rapidly in the future. When IoT enables communication, interaction, and information sharing between people and people, people and objects, and objects and objects, it becomes possible to provide intelligent services for judging things on their own, and companies can reduce costs effectively. Things are expected to be diverse with the arrival of the Internet environment between objects and objects. IoT-enabled devices, such as sensors and home appliances, can be connected to devices supporting IoT. In order to do this, it is necessary not only to retrieve information about various devices or applications that can be accessed, but also to obtain information such as authentication and access IP (Internet Protocol) for accessing the retrieved objects. Therefore, the progress of IoT has enabled new products and services in various fields such as smart home, smart home appliance, smart car, smart grid, health care, wearable appliance.

2 is a block diagram conceptually illustrating an operation of a method for evaluating reliability of a device in an Internet environment of objects according to an embodiment of the present invention.

Referring to FIG. 2, an interaction database is usefully used to calculate the reliability of a device. The interaction database stores similar prior experiences based on the situation so that current situation information can be easily identified.

Similar prior experience with specific situations can also be used to calculate personal confidence based on this experience. Personal trust is referred to herein as trust disposition and is described in greater detail below in the relevant portions of this specification. Therefore, repetitive descriptions are omitted for the sake of simplification of the specification.

The individual reliability calculated by the trustworthiness is used again in the process of calculating the non-personal reliability and in deriving the final reliability. The process of calculating the final reliability will be described in detail later in this section of the specification. Therefore, repetitive descriptions are omitted for the sake of simplification of the specification.

Based on the calculated final reliability, the user decides whether to perform the interaction with the new device. When deciding to perform an interaction, the user updates the interaction database with an experience similar to the result of performing the interaction. Machine learning algorithms may be used herein to classify and identify contextual information, and such algorithms are described in greater detail in the relevant portions of the specification, and thus repetitive descriptions are omitted for the sake of brevity.

Hereinafter, the operation of the method for evaluating the reliability of a device in the Internet environment of objects according to the present invention will be described in detail.

FIG. 3 is a flowchart illustrating a method of evaluating reliability of a device in an object Internet environment according to an embodiment of the present invention.

First, in order to determine the reliability of the device, it is necessary to recognize the status information of the situation in which the device is found in the object Internet environment (S310). The situation information can be recognized by combining various sensor information measured using the sensors included in the object internet environment. Objects Because the various sensors used in the Internet have a microprocessor, data can be transmitted wirelessly or wired. The microprocessor embedded in the sensor may be various from a simple passive RFID without a power source to a powerful CPU with a large number of cores (for example, a quad core). Therefore, these sensors are also called smart sensors or IoT controllers.

However, the data transmitted by the smart sensor is a plain text that is not encrypted, and the contents of the data can be grasped easily from the outside. Therefore, stepless modulation is possible. Therefore, the malicious code having the sniffing function and the spoofing function can penetrate into the network gateway for collecting and processing the data, so that even if there is data to be encrypted and input in the sensor terminal, Decryption may not be secure because the decryption key may be exposed. Therefore, it is possible to provide a secure communication (content leakage or modulation is impossible) between the smart sensor and the network gateway in the Internet environment of objects, and a technique of safely decrypting data in the network gateway can be applied.

When the situation information is recognized, the situation is classified according to the machine learning algorithm by referring to the situation factors constituting the situation information (S330). The reason for classifying the situation in this way is to maximize the quality of the calculated reliability.

According to the results of various social psychology studies, people form their own 'disposition' of each situation based on past interaction 'experience' and the results, and categorize the given new 'situation' , And determines whether or not to trust the first person to meet based on such classified situation and temperament. The reason for this is that people predict the outcome of interaction with an opponent in a particular situation based on the experience of interaction with a similar person in a similar situation in the past. For example, if a banker who first met at an arbitrary bank requests specific personal information for a particular transaction, and if the other bank has previously provided similar personal information to another banker and successfully completed the transaction, the trader is willing to We will provide personal information. However, if the person who requested personal information is not a banker, or if the requested place is not a bank, it will not provide the personal information in general. In this way, people can lower the uncertainty of the interaction result and determine the trust through experience-based contextualization when there is insufficient information about the partner. In addition, experience - based situational classification has a close relationship with the formation of initial trust by influencing the 'familiarity' and 'stability' of a given situation.

To classify contextual information, the M5 model tree learning algorithm can be used. Using this learning algorithm, a decision tree can be created from the user 's interaction experience and the situation can be classified by applying the contextual factors to the decision tree. This M5 model tree learning algorithm is described in, for example, Quinlan, J., " Learning with continuous classes, " In Proceedings of the 5th Australian Joint Conference on Artificial Intelligence, pp. 343-348, 1992. < Desc / Clms Page number 2 >

At this time, it is necessary to extract experience according to the situation. That is, when the characteristics of the current situation are acquired, the experience of the user's corresponding situation can be tracked to look up the maintained interaction history, and the history record of the device whose type is the same as the type of the target device can be extracted.

4 is a diagram illustrating a history record of a specific device.

Referring to FIG. 4, each history record consists of four columns, each of which is a context feature vector, a number of positive experiences, a number of negative experiences, and a confidence value. The constituent elements and elements of the context characteristic vector included in the history are the same as the current context characteristic vector described in Table 1. Experimental examples based on this situation are stored in a trusted server and loaded from there.

Table 1 is a table illustrating the device type, the device role, and the place type. A device type includes a field of a manufacturer and a model name, and a device role includes a task field. In addition, the Place Type includes a Category, a Main Function, a Sub-Function, and a Privacy Level. For example, the main function may be divided into Service or Leisure, and the classification may include fields such as Office, Park, and Hospital. Can be divided.

Referring again to FIG. 4, if a device type record of the device is examined, it is known that the type of the device is an air conditioner and a Samsung Electronics product. In addition, an example of the interaction history (Interaction History) can be seen as the number of positive experiences (#PosInt), the number of negative experiences (#NegInt), and the reliability value (TrustVal). These records are continuously recorded as a Situational characteristic vector. As described above, the context characteristic vector can be recorded and updated in the interaction database.

According to the M5 model tree learning algorithm, decision trees can be learned using user interaction experience and contextual factors. The M5 model tree learning algorithm uses a given learning data set (dataset) to generate a decision tree, which quickly and intuitively classifies simple data sets. Each intermediate node of the generated decision tree represents a branch through a value of a specific element, and each leaf node represents a specific linear model. In the present invention, the M5 model is learned using the results of the user's interaction experience and the main contextual factors when the experiences occur. When reliability evaluation is required for any device, the situation is classified by inputting the given contextual factors into the learned M5 model, and the reliability of the user's device according to the situation is obtained.

FIG. 5 shows an example of classifying learned situations according to the M5 model tree learning algorithm.

Referring to FIG. 5, a service record and a public record in a record of a status characteristic vector are classified into a privacy level and a main function, respectively, and configured as a tree. That is, the model tree is divided into the main function node with the privacy level node, and the main function node includes the sub-function, the linear model 1, the linear model 4, Model 5 (Linear Model 5). In addition, additional functional nodes are divided into Linear Model 2 and Linear Model 3. A record of the situation characteristic vector is applied to each of these nodes and stored in the interaction database.

For example, if the privacy level is public, it proceeds along the left branch, and if the privacy level is semi-private, it proceeds along the right branch. Also, if the main function is leisure, proceed to the right branch and proceed to the additional functional node.

It is to be understood that the decision tree shown in FIG. 4 is provided only for convenience of explanation, and that a different decision tree can be constructed when the situation characteristic vector is different. Therefore, it is apparent that the decision tree of FIG. 4 does not limit the present invention.

If the context information is classified as described above, the reliability of the device is determined by referring to the user's trustworthiness of the classified situation (S350). To evaluate the reliability of an unfamiliar device for a user, one must first understand what factors influence the initial trust formation between the person and the device and how the cumulative experience changes the reliability. Assuming that factors affecting the formation of trust between people maintain their influence even when they form trust in the device, social psychological studies on trust between people are the starting point in evaluating the reliability of devices . Therefore, it is necessary to take into consideration the factors which can be applied to the reliability evaluation of the device while giving an influence to the initial reliability formation.

For example, the most important and central factor in the classification of situations is the 'place' where the interaction occurred. People who belong to similar "groups" do not only show similar characteristics but also behave according to the given norms to protect the reputation and interests of their group. However, since people often become aware of related groups through their own places, location awareness can be an important factor in predicting and generalizing the characteristics of related persons. This can reduce uncertainty in the situation and give a sense of stability to help form the initial confidence. That is, if the interacting parties belong to the same group or are very closely related, they can form a trust relationship more quickly.

Similarly, since most devices belong to a particular organization, devices in a similar location will exhibit similar characteristics according to the goals and norms of their organization. For example, banks' devices are likely to maintain high reliability due to bank specificity and maintenance regulations, but café devices can have relatively low reliability. Therefore, in order to evaluate the reliability of an unfamiliar device, it is necessary to consider similarities and related experiences between the place where the user is located and the place where he / she has heard in the past.

The perception of a place also helps to know whether it is the other 's role, which is another important factor that influences the reliability of a stranger. The recognition of the place allows us to know what the related organization is and what kind of person it is, and by recognizing these two things, we can deduce the purpose and the mission of the people involved in the business in the place. If you know the role of the other person, you can understand the other person's behavior based on their interaction with the person who has a similar role. In addition to being able to anticipate and cope with the actions to be followed, You can give.

Thus, contextual information makes interacting parties feel friendly to a given situation, thereby providing stability to the interaction and leading to rapid formation of trust relationships. Likewise, most devices are built for specific 'purposes' and 'uses' and are properly installed where they are needed, so location awareness helps the device to be a role. And the recognition of the role of the device can be helpful in evaluating the reliability of the device based on its experience of interaction with devices having similar roles in the past.

In the present invention, reliability can be evaluated using the ST (Swift Trust) concept. For more information on ST concepts, see Meyerson, D., Weick, K., and Kramer, R., "Swift Trust and Temporary Groups," Trust in organizations: Frontiers of theory and research, pp. 166-196, 1996. < / RTI > The ST concept explains what situational factors influence the process of building trust for mutual cooperation when people lacking information are gathered to achieve a certain task. According to the ST concept, initial reliability among people is determined by non-personal factors such as stereotypes and reputations. As the experience of direct interaction accumulates, they form their own 'trustworthiness' for each situation. In order to determine the reliability of a stranger, we classify a given situation based on distinctive situational factors, determine the trustworthiness of the opponent according to his or her trustworthiness of the situation, And determines whether it works.

6 shows an example of a linear regression analysis model for estimating the reliability value.

Referring to FIG. 6, a new result value is obtained by adding or subtracting a different value to an existing result value according to each record value of the linear model. As described above, the regression analysis model shown in Fig. 6 is only provided as an example for estimating the reliability value, and thus should not be construed as limiting the present invention.

Table 2 summarizes the factors that can be applied to the reliability evaluation of human devices, affecting the formation of human trust.

As described above, it can be seen from Table 2 that the purpose / use of the device, the management organization, the maker, the place, the place type, etc. may influence the formation of people's trustworthiness.

As described above, the reliability of the device is evaluated using the contextual factors. That is, according to the present invention, a machine learning algorithm is used to classify a given situation according to major situation factors, and to calculate a physical reliability for a device encountered in a given situation based on a user's past experience.

As such, when the basic reliability is calculated, the final reliability of the device is calculated based on the basic reliability using the direct experience and a priori confidence level of the user's device (S370). SL (Subjective Logic) can be used to calculate the final reliability based on the underlying reliability. The SL technique mathematically models the degree of belief calculus that the subject will have a positive outcome based on experience with a particular subject. SL allows an arbitrary subject to express his opinion by quantifying the degree of belief and uncertainty based on experience for a given proposition. For details of the SL technique, see Josang, A., "Artificial reasoning with subjective logic," In Proceedings of the 29th Australian Computer Society, Inc., pp. 85-94, 2006. < / RTI >

This SL technique can be used to quantify the degree of certainty that a given device will behave positively or negatively based on direct experience with a particular device, and if the experience is lacking, based on priori degrees of belief The degree of conviction can be quantified. A priori confidence level means that subjective judgment is made using indirect information when it is difficult to make an objective judgment because there is no direct experience and uncertainty is high. The characteristics of this SL technique can be used as an intuitive method that can express both objective and subjective judgment in the reliability evaluation of human devices. The formula for SL is shown in Equation 1 below.

In Equation (1), x represents an evaluator and y represents an evaluation object. P ( W ^x _y ) is a probability value indicating the degree of assurance of x that y of x will behave positively, and r ^x _y , u ^x _y , v ^x _y and a ^x _y are positive Experience, uncertainty, personal experience, and confidence level by stereotype, respectively.

u ^x _y and a ^x _y are again expressed by Equations (2) and (3), respectively.

In Equations (2) and (3), s ^x _y denotes the number of negative interaction experiences of x with respect to y, and r denotes another user. Also, t _y ^ρ is the reliability for y, and c _ρ represents the reliability of the user with respect to t _y ^ρ .

Without any direct experience of y in x, confidence and uncertainty are 0 and 1, respectively, so that the confidence of y in x is equal to the expectation for a priori confidence. In the present invention, the above-described calculated reliability is applied as an expectation value for this a priori confidence. When P ( W ^x _y ) for _y is calculated, x determines whether to interact with the object based on it. If it decides to interact, it is judged to be a positive interaction and added to calculate r ^x _y , which is then used to determine whether to interact with other similar objects in the future.

As described above, according to the method of evaluating reliability of devices in the object Internet environment according to an aspect of the present invention, in order to evaluate the reliability of a person and object Internet device, it is not necessary to evaluate the reliability of the device through a stereotype, Consider what type of device has been interacting with the past in the same situation. In other words, reliability can be evaluated by using given 'experience on the situation'. Thus, the degree of false positives is minimized by allowing the user to flexibly assess the reliability according to the current situation.

FIG. 7 is a block diagram schematically illustrating an apparatus for evaluating reliability of a device in an object Internet environment according to another aspect of the present invention.

7, the device reliability evaluation apparatus 700 according to the present invention is connected to various sensors 720, 722, and 724 through the object Internet 710. [ The device reliability evaluation apparatus 700 includes a situation information recognition unit 730, a situation information classification unit 740, a central control unit 750, a basic reliability determination unit 760, a final reliability calculation unit 770, And an interaction database 790.

The situation information recognizing unit 730 recognizes the situation information on a situation in which the device is found in the object Internet environment. The reason for recognizing the situation is that in order to evaluate the reliability of a new device on the part of the user, it is necessary to first understand what factors influence the formation of initial trust between the person and the device and how the accumulation of experience changes the reliability As described above. In other words, since most devices are appropriately installed for a specific purpose and purpose, recognizing the situation information such as a place can grasp the role of the device, and if the role is recognized, Based on interaction experience, it can be helpful to evaluate the reliability of the device.

When the situation is recognized, the situation information classifier 740 refers to the situation factors constituting the situation information and classifies the situation according to a machine learning algorithm. It is the same as described above that the classified situation can constitute a decision tree as shown in Fig. To classify such situations, a machine learning technique such as an M5 model tree learning algorithm can be applied.

If the situation is classified, the quality reliability determining unit 760 determines the device reliability for the device by referring to the user's trust disposition for the classified situation. To this end, the material reliability determination unit 760 forms a trust substrate of the user based on the user's interaction experience with the classified situation using the ST (swift trust) technique. In addition, the user's trustworthiness of the situation classified according to the decision tree can be referred to to determine the underlying reliability for the device.

Once the trust level of the user is determined, the final reliability calculator 770 calculates the final reliability for the device based on the determined basic reliability, using at least one of the user's direct experience with the device and the degree of a priori confidence . A technique such as Subjective Logic (SL) can be used to calculate the final reliability as described above. Using SL, an arbitrary subject can quantify the degree of confidence and uncertainty based on experience for a given proposition, and the regression model illustrated in FIG. 6 can be applied in this process.

Using the device reliability evaluating apparatus 700 shown in FIG. 7, it is possible not to evaluate the reliability of a device through a stereotype, but to consider what type of device has been interacted with in a past situation which was the same as a given situation You can evaluate the reliability of people and things Internet devices. Therefore, it is possible to evaluate the reliability by using a given 'experience on the situation' and to minimize the false positives by flexibly evaluating the reliability according to the current situation of the user.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. For example, the present invention has been described as exhibiting excellent performance when determining the reliability of a new device, since reliability can be calculated through classification of a situation against a new device that has no direct experience with the object. However, the present invention is not necessarily used to determine the reliability of a new device, but can be applied equally to a case in which there is sufficient direct experience with the object. Where direct experience with objects is used, both direct experience and stereotypes are used, whereas without direct experience, only stereotypes are used. Therefore, it should be understood that the present invention can be usefully used to determine the reliability of an existing device as well as a new device.

In addition, the method according to the present invention can be embodied as computer-readable code on a computer-readable recording medium. A computer-readable recording medium may include any type of recording device that stores data that can be read by a computer system. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like. The computer readable recording medium may also store computer readable code that may be executed in a distributed manner by a distributed computer system connected to the network.

As used herein, the singular " include " should be understood to include a plurality of representations unless the context clearly dictates otherwise, and the terms " comprises & , Parts or combinations thereof, and does not preclude the presence or addition of one or more other features, integers, steps, components, components, or combinations thereof. The terms "part", "unit", "module", "block", and the like described in the specification mean units for processing at least one function or operation, And a combination of software.

Therefore, it should be understood that the present invention and the drawings attached hereto are only a part of the technical idea included in the present invention, and that those skilled in the art will readily understand the technical ideas included in the specification and drawings of the present invention It is obvious that all the variations and concrete examples that can be deduced are included in the scope of the present invention.

The present invention can be applied to a social matching and communication service technology for safe and happy life in the IoT-based smart home community.

Claims (15)

A method for evaluating reliability of a device in an object internet environment,
A situation information recognition step of recognizing situation information on a situation in which a device is found in the object internet environment;
A situation information classification step of referring to the situation factors constituting the situation information and classifying the situation according to a machine learning algorithm;
A basic reliability determination step of determining a basic reliability for the device by referring to a trust disposition of the user with respect to the classified status;
A final reliability calculation step of calculating a final reliability for the device based on the determined basic reliability using at least one of a direct experience of the user and a degree of a priori confidence in the device,
The final confidence calculation step
Using SL (subjective logic)

(W ^x _y ), where x is an evaluator, y is an evaluation object, and P (W ^x _y ) is the degree of certainty of x that y of x will act positively the indicating probability values, the r ^x _y is uncertainty about the positive experience, the u ^x _y is y of x to y of x, the v ^x _y is personal experience, the a ^x _y to y of x is the x y is the degree of assurance by the stereotype for y,
The uncertainty (u ^x _y ) for _{y of} ^x is

, And the degree of assurance (a ^x _y ) by the stereotype for y of x is calculated according to

But calculated according to the s ^x _y is the number of negative interaction experience for the y of the x, wherein ρ is the reliability of the other users, the t _y ^ρ is y, the c _ρ is the user with respect to the t _y ^ρ And
Wherein the method further comprises increasing r ^x _y whenever the user decides to interact with the device. The method according to claim 1,
In the situation information recognition step,
And recognizing the status information using a sensor included in the object Internet environment. 3. The method of claim 2,
Wherein the status information classification step comprises:
Generating a decision tree from the user's interaction experience using an M5 model tree learning algorithm and classifying the situation by applying the contextual factors to the decision tree, Wherein the reliability of the device in the Internet environment of objects is evaluated. The method of claim 3,
Wherein the decision tree is learned using the interaction experience of the user and the contextual factors. 5. The method of claim 4,
Wherein the basic reliability determination step comprises:
Using the ST (swift trust) technique to form the user's trust temperament based on the user's interaction experience with the classified situation; And
And determining the underlying reliability for the device with reference to the user's trustworthiness for the situation classified according to the decision tree. ≪ RTI ID = 0.0 > 31. < / RTI > delete delete An apparatus for evaluating the reliability of a device in an Internet environment,
A situation information recognition unit for recognizing the situation information of a situation in which a device is found in the object Internet environment;
A situation information classifying unit for classifying the situation according to a machine learning algorithm with reference to situation factors constituting the situation information;
A trustworthiness determiner for determining trustworthiness of the device with reference to a trust dispositon of the user for the classified situation; And
And a final reliability calculation unit for calculating a final reliability for the device based on the determined basic reliability using at least one of a direct experience of the user with the device and a degree of a priori confidence,
The final reliability calculation unit
Using SL (subjective logic)

(W ^x _y ), where x is an evaluator, y is an evaluation object, and P (W ^x _y ) is the degree of certainty of x that y of x will act positively the indicating probability values, the r ^x _y is uncertainty about the positive experience, the u ^x _y is y of x to y of x, the v ^x _y is personal experience, the a ^x _y to y of x is the x y is the degree of assurance by the stereotype for y,
The uncertainty (u ^x _y ) for _{y of} ^x is

, And the degree of assurance (a ^x _y ) by the stereotype for y of x is calculated according to

But calculated according to the s ^x _y is the number of negative interaction experience for the y of the x, wherein ρ is the reliability of the other users, the t _y ^ρ is y, the c _ρ is the user with respect to the t _y ^ρ And
And increases the r ^x _y whenever the user decides to interact with the device. 9. The method of claim 8,
The situation information recognizing unit,
Wherein the status information is recognized using a sensor included in the object Internet environment. 10. The method of claim 9,
Wherein the status information classifier comprises:
A decision tree is generated from the user's interaction experience using an M5 model tree learning algorithm, and the situation is classified by applying the contextual factors to the decision tree Wherein the device evaluates reliability of the device in the Internet environment. 11. The method of claim 10,
Wherein the decision tree is learned using the interaction experience of the user and the contextual factors. 12. The method of claim 11,
The basic reliability determination unit may determine,
A trust trust (ST) technique is used to form the trust substrate of the user on the basis of the user's interaction experience with the classified situation, and the confidence level of the user for the situation classified according to the decision tree And determines a basic reliability for the device by referring to the device reliability information. delete delete delete

Images