KR100837798B1 - Method for providing serivce with situation aware personalization, recording medium thereof and system for providing serivce with situation aware personalization - Google Patents

Abstract

A method and a system for providing a situation aware personalization service, and a recording medium thereof are provided to predict a user request by analyzing a web access pattern even if a request is not received from a user, reduce a terminal operation frequency of the user by presenting predicted links, and offer the service optimized to restricted display and interface of a terminal. An access analyzer(150) reflects situation information for link use of a user on a web access pattern. A storage(190) stores the web access pattern. A controller(160) generates current state information of the user by using neighboring nodes selected by a user terminal(100), generates the next state information by applying the current state information to a Markov chain, and transmits the next state information to the user terminal. The terminal includes an extractor(120) for selecting the neighboring nodes by using the web access pattern and extracting links predicting a user request according to the received next state information, and a display unit(110) outputting the extracted links. The access analyzer generates a web access pattern graph including an edge connecting a plurality of nodes by reflecting situation information.

Description

Context-aware personalization service method, recording medium and context-aware personalization service system {Method for providing serivce with situation aware personalization, Recording medium etc and System for providing serivce with situation aware personalization}

1 is a block diagram of a context aware personalization service system according to the present invention.

FIG. 2 illustrates an example of a state diagram for predicting a next state in FIG. 1.

FIG. 3 illustrates an example of a web access pattern graph for predicting a user request in FIG. 1.

4 illustrates an example of a sequence diagram of the access pattern of FIG. 3.

5A and 5B apply Markov chains to the sequence diagram of FIG. 4.

6 is a flowchart of a context aware personalization service method according to the present invention.

7 is a detailed flowchart of the Markov chain application process of FIG. 6.

The present invention relates to ubiquitous computing, and more particularly, to a context-aware personalization service method, a recording medium thereof, and a context-aware personalization service system.

In general, the tuning of the mobile communication terminal only changes the appearance of the mobile communication terminal, and the user interface of the mobile communication terminal is simple and uniform as provided by the mobile communication terminal manufacturer. In addition, such a user interface is designed to utilize a minimum of graphics and sound resources due to the limitations of the storage capacity, processing speed, and computational capacity of the mobile communication terminal, thus failing to satisfy various needs of the user of the mobile communication terminal.

Typically, mobile services perform services for only one specialized domain. That is, the current situation is to provide limited to movie reservation, traffic reservation, location-based information service, etc. that are carried out on the mobile phone (Mobile Phone).

In addition, since mobile services have been built on the basis of each independent service system, in order to integrate them, a newly developed mobile integration system must be built, which causes economic burden on the construction cost.

In addition, the terminal providing the mobile service has a feature of providing various mobile services on a very small screen due to the characteristics of the environment, and the provision of service contents for providing such services is limited, thus integrating various services to the mobile terminal. There was a technical limitation in providing.

Recently, the multi-stage information retrieval process is defined as procedural information retrieval, and the concept of service-pack is established. Also, accumulation, sharing, personalization, and expansion of information retrieval and information retrieval experiences in mobile environment are established. The technology of combining the concept of and reuse with the concept of service pack and designing and presenting a framework for building a service pack environment has been proposed.

In the existing Internet information retrieval environment, the user has experienced discontinuous and multi-level information retrieval process to achieve a specific retrieval purpose. Information retrieval process.

Service-Pack is defined as a bundle of resources for the purpose of sharing, expanding, and reusing the accumulated experience of users' procedural information retrieval in the Internet information retrieval environment. The basic concept of a service pack is a bundle of closely related information for a particular query. The user can requery based on the service pack and the service pack will be expanded again.

On the other hand, the technology to enhance tourists' communication, communication, and tourism effect has been proposed by mounting tourist attractions information contents on the Internet-enabled PDA terminal as well as providing convenience for foreign tourists.

In particular, it is possible to overcome the problem of billing by distributing mobile terminals by renting them at airports for foreign tourists who come to Korea as one of the profit models, and by processing additional services such as the use of contents by using mobile phone call calculation methods. It is expected to be a model of the new cultural tourism industry in the century, and it is expected to contribute to the development of new tourism industry through tourism promotion and reservation system using tourism internet broadcasting station and PDA.

On the other hand, a system for collecting and packaging structural and unstructured information scattered over a network such as the Internet has been proposed as an information collection and integrated service system. The network access interface is manipulated by the user, and the network interface targets a local resource repository and a uniform resource locator (URL) on the Internet. Data movement between the network interface device and the data store is governed by an intermediate gateway system. The collected data is then integrated, customized, personalized and converted to meet the user's information resource needs through the network interface.

However, the conventional personalized service method can provide the service only when the user requests it, and can not predict the user's demand, thus increasing the number of user's terminal operations unnecessarily and optimizing the limited display and interface of the terminal. There is a problem that cannot provide the service provided.

Therefore, the first technical problem to be achieved by the present invention is to analyze the user's web access pattern even when there is no user request, to predict the user's demand, reduce the number of user's terminal operation by presenting the predicted link, The present invention provides a context-aware personalization service method that can provide an optimized service for a limited display and interface.

A second technical object of the present invention is to provide a computer-readable recording medium having recorded thereon a program for executing the context-aware personalization service method on a computer.

The third technical problem to be achieved by the present invention is to provide a context-aware personalization service system to which the context-aware personalization service method is applied.

In order to achieve the first technical problem, the present invention includes the steps of reflecting context information about a user's link usage in a web access pattern, selecting neighbor nodes using the web access pattern, and using the selected neighbor nodes. Generating current status information of the user, applying the current status information to a Markov chain to generate next status information, and linking predicted the user's request according to the next status information to a screen of the user's terminal; It provides a context-aware personalization service method comprising the step of outputting.

In order to achieve the second technical problem, the present invention provides a computer-readable recording medium that records a program for executing the context-aware personalization service method on a computer.

In order to achieve the third technical problem, the present invention provides a system including a service provider and a plurality of terminals, wherein the service provider includes: an access analyzer for reflecting context information on a user's link usage in a web access pattern; The current state information of the user is generated using the storage storing the pattern and the neighbor nodes selected by the terminal, the next state information is generated by applying the current state information to the Markov chain, and the next state information is generated by the user. And a controller for transmitting to a terminal of the extractor, wherein the terminal selects neighbor nodes using the web access pattern, and extracts links for predicting the user's request according to the transmitted next state information. It includes a display that outputs linked links. Cognitive context and provides a personalized service system.

Personalization of services is about tailoring services to the needs and preferences of users or groups. Personalization is important in today's service-oriented societies and is key to receiving the services provided by the Internet and mobile communications.

Hereinafter, nodes, links, current state, and next state are defined as concepts referring to arbitrary web sites.

Two important categories of personalization are:

First, personalization based on external resources is described as follows. An example of this is the Local Based System (LBS). This is a technology that provides various services related to the measured location by measuring the mobile terminal using a mobile communication facility or satellite signals.

Personalization services based on external resources, such as regional-based systems, provide a user-specific service by acquiring a user's location and using location information for the service using location resources of the user and the terminal.

Next, personalization based on internal resources is described as follows. Internal resources for personalization service mainly consist of static information using user's profile.

The profile may include the user's terminal, the user's preferences, and the user's basic information. The method of collecting the profile is to statistically analyze the degree of use of the user and to analyze the information transmitted and the direct input information of the user. In addition, the collected profiles may be produced in a profile language such as CC / PP, GUP, or the like. As with profiles, internal resource-based personalization provides user-centric services.

This divided approach to personalization can be considered a resource provider in terms of context awareness.

Therefore, two approaches to obtaining contextual information resources of a user are integrated by contextual awareness. In addition, the concept of contextual cognitive personalization is obtained based on a variety of resources. In addition, the collected resources of the contextual information of the user by each approach will determine the context factor p.

Hereinafter, with reference to the drawings will be described a preferred embodiment of the present invention. However, embodiments of the present invention illustrated below may be modified in many different forms, and the scope of the present invention is not limited to the embodiments described below.

1 is a block diagram of a context aware personalization service system according to the present invention.

The proposed model for contextual cognitive personalization is as follows.

One of the features of context-aware personalization is to predict user requirements in advance to provide better user-centric services using a variety of resources.

The concept of context aware personalization is to provide a simple site from the user's point of view of the mobile terminal, and to represent one site by the prediction process.

The overall operating model, including the system structure of context-aware personalization, is described as the connection between mobile and the Internet through middleware interfaces such as gateways and controllers 160. In addition, the interactions between the extractor 120 and some essential parts that play an important role are described below.

The access pattern graph is formed by a structure consisting of a connection of an access analyzer 150 and a controller 160, a wireless / internet gateway, and a storage 190.

It is also described to support personalization using operations of parsed patterns, context sequences, and user context sets. Finally, by predicting the matrix using the pattern log and node state, predicting the place of the next step is the overall operational model of contextual cognitive personalization as shown in FIG.

In context-aware personalization, the defined and predictive models are described based on operational models.

In context-aware personalization, prediction means acquiring a pattern of stored web access data over time to provide the user's next request in advance.

The system of FIG. 1 includes a service provider and a plurality of terminals 100. The terminal 100 includes a display 110 and an extractor 120.

The display 110 outputs the extracted links to the screen. The display 110 may include a liquid crystal display panel, an OLED panel, and the like. In this case, the content division display method suitable for the user's web request step may be applied to the screen.

The extractor 120 selects neighbor nodes using the web access pattern, and extracts links predicting the user's request according to the next state information transmitted from the service provider.

Preferably, extractor 120 may select nodes as neighboring nodes whose edge value is greater than or equal to a predetermined threshold. At this time, the edge value indicates its usage factor. Preferably, the extractor 120 may weight the particular node according to the time the user has spent at that particular node.

The service provider includes an access analyzer 150, a controller 160, and a storage 190.

The access analyzer 150 reflects the context information of the user's link usage in the web access pattern. Preferably, the access analyzer 150 may generate a web access pattern graph including an edge connecting the plurality of nodes by reflecting the context information. In this case, the web access pattern graph may be a graph in which an edge value is allocated according to an access frequency between two nodes connected by one edge.

The controller 160 generates current state information of the user by using the neighbor nodes selected by the terminal 100. In addition, the controller 160 generates the next state information by applying the current state information to the Markov chain, and transmits the next state information to the user terminal 100.

Preferably, the controller 160 may calculate the start vector by analyzing the most recent web access data, and calculate the probability of the next state by multiplying the state transition matrix representing the nodes before moving to the current node by the start vector. have. At this time, the controller 160 generates the next state information according to the probability of the next state.

Storage 190 stores web access patterns. In this case, the web access pattern may include an access pattern graph and an access frequency of neighboring nodes.

FIG. 2 illustrates an example of a state diagram for predicting a next state in FIG. 1.

The web access pattern can be generalized by the state diagram as shown in FIG. In addition, while there are various approaches for prediction, the present invention uses web access data to calculate general patterns suitable for mobile environments. The Markov chain model is used to provide an algorithm for predicting the next state of a pattern.

The target is the URL of the web page you want. The source is a web page that contains a link to the target. When the user enters the target or selects on the first page, NULL is written as the source.

When the mobile terminal is connected to the service provider's wireless network, the first page is displayed. The first page contains a list of links that the user frequently visits.

S1 of FIG. 2 is an initial state, and S6 is a new state that can be predicted based on a pattern. With this approach, the user pattern recorded through the link reset algorithm (personalization algorithm) described below is provided to the prediction service, which is the final stage of context aware personalization.

Context-cognitive personalization based on the predictive model is as follows. The prediction model may be produced using the definition of prediction as described above.

The records that make up the unit represent the stored access records for a preset time period and the physical location where the user makes his request. In this case, the preset time period is based on a time slot of one day.

The web access patterns of a particular user are represented in graphs that are connected and directed to each other. The nodes here represent URLs, and each edge represents the relationship of the source and target of web access between any two nodes.

FIG. 3 illustrates an example of a web access pattern graph for predicting a user request in FIG. 1.

3 shows an example access pattern graph. Here each edge is annotated with an edge value representing its usage factor.

The usage factor represents an access period from the source node of the specific edge to the target node for a preset time period.

Each node represents a URL of a web page, but for simplicity, most nodes do not represent a full URL. It is assumed that context aware personalization systems continue to follow the complete URL.

For example, in FIG. 3, the URL for the 'Context Aware' node is {embedded. korea.ac.kr/SituationAware}. By converting the state transition diagram described by the definition of prediction into a graph, the prediction model is represented. In addition, a link reset algorithm is provided, and context-aware personalization based on prediction is described through this model.

The personalization algorithm is described below.

The extractor 120 of the operational model is used to extract link information when receiving the URL of the requested web page from the controller 160 of FIG. 1.

Depending on the time slot of the day and the physical location of the user upon acknowledgment of the request, different access pattern graphs may be applied to reestablish links for the interaction of context-aware personalization.

Resetting the set of links from the applied access pattern graph is based on link usage information. In addition, we present a method for predicting the next state according to the current access pattern link based on the link frequency and link usage reflection used to reflect the link usage.

The core algorithm that reflects the context information in the user's access pattern and predicts the user's request, that is, the link reset algorithm, can be summarized as follows.

First, determine the set of neighbor nodes. Each of the neighbor nodes has an incoming edge from the requested node. At this time, the edge value is greater than or equal to the threshold.

Second, present the current states using the determined set of neighbor nodes and predict the next state using the Markov chain.

Third, repeat the first step in each subgraph. Each of those routes is a node included in the set determined in the first step.

For example, it is as follows. In the following, it is assumed that the set of nodes constituting the web access graph is N. Further, if v is included in set N and e (u, v)> 0, then node v is assumed to be a peripheral node of node u. At this time, e (u, v) represents the value of the edge. Edges exiting node v are defined as edges between node v and surrounding nodes.

The total outgoing edge value of node v is defined as the sum of the edge values of the outgoing edges. Peripheral nodes whose edge value is greater than or equal to the threshold are selected. Threshold [theta] provides a criterion for selecting frequently accessed nodes among the requested nodes.

The threshold is calculated as in Equation 1 below.

ρ is the selection factor. The selection factor is a preset value.

The threshold θ may change depending on the selection factor and may affect the selection of the node.

This selection factor is affected by contextual information deduced from various resources and is set to a value between 0 and 1.

In the present invention, access frequency is used to indicate link usage. However, it can be combined with other contributions or weights based on system resources and network environment.

For example, a weight may be assigned to each node based on the time spent at that node before the user navigates to another page. In addition, the pattern of the node path is examined in this process, and the Markov chain can be used to track the next node.

In FIG. 3, when the user is connected with the service provider's wireless network, the threshold θ = (2 + 15 + 5 + 18 + 5 + 40 + 10 + 5) * ρ / 8.

If the selection factor (current context information) p = 0.8, then θ = 10. Accordingly, nodes outgoing from the user node whose value is greater than or equal to the threshold {embedded.korea.ac.kr}, {www.korea.ac.kr}, {eku.korea.ac.kr } And {faculty} are selected for the first time.

In this example, only node {embedded.korea.ac.kr} has an outgoing edge. Thus, the link reset process continues for that subgraph.

The extractor 120 selects {people.shtml} and {SituationAware} as nodes that are periodically accessed links. The incoming edge of the selected nodes has a value greater than the threshold (θ = 10.66) recalculated for the node {embedded.korea.ac.kr}.

In a similar way, extractor 120 adds nodes {faculty}, {Hohin}, {SituationAware}, {embedded.korea.ac.kr / admissions}, {course-info}, {CSCE316} and {CSCE418}. Choose.

Similarly, if the user selects any one link on the screen, the same link resetting process is applied.

The controller 160 wraps the link at the selected node and sends the wrapped link to the user's display.

The following describes the application of the Markov chain to the personalization algorithm.

The Markov chain model, which is an important part of the prediction algorithm, is described in line with the link reset algorithm. If the Markov chain model is derived based on an algorithm for selecting nodes based on the threshold θ, then the place of the next step selected by the user can be predicted.

One way to implement this is to assume that the state of the model depends only on the previous state. This is called the Markov hypothesis, which can greatly simplify the given problem.

Considering context-aware personalization, the Markov assumption assumes that the current node can always be predicted only by the knowledge given about some past nodes (websites). The first Markov process to construct context-aware personalization is shown below.

States: Web-Page1 (linked), Web-Page2 (linked),… … , Web-Page n (linked).

4 illustrates an example of a sequence diagram of the access pattern of FIG. 3.

In FIG. 4, the transition for contextual cognitive personalization is translated into the state transition diagram of FIG. 5A.

5A and 5B apply Markov chains to the sequence diagram of FIG. 4.

FIG. 5A is a Markov chain for context aware personalization, and FIG. 5B is a simplified context aware personalization Markov chain grouping the states of interest.

The state transition diagram of FIG. 5A can be considered to form a Markov chain since the transition to the next state mostly depends on the transition from the current state and the state under consideration to other states that only depend on links in the current access pattern. .

On the other hand, the π vector applied to the Markov chain defines the likelihood of a system present in each of the states at time zero (starting vector). The π vector can be considered as the threshold θ.

Through analysis of the most recent data, the starting vector is calculated using Equation 1, which satisfies the condition of Equation 2.

In Equation 2, I () represents a start vector.

The likelihood of current nodes using the state transition matrix is represented by the previous node (website). A system that can be described in this way is the Markov process.

Where S is in the form of a state transition matrix.

In the prediction phase of the next web link, the likelihood and frequency of future clicks are estimated. In this case, the generated state transition matrix and the generated start vector are used.

Equation 3 below shows the process of calculating the likelihood of a click occurring.

In Equation 3, P is a probability of occurrence, whereby the next state (website) can be predicted.

In this way, the user can get links associated with a particular link (web page) until the user has specifically defined the content.

When the user wants web content for the selected link, the controller 160 transmits the URL according to the selected link to the terminal 100. On the other hand, when no link information is available by the access analyzer 150, the controller 160 links to the original server to the terminal 100 via the wireless / Internet gateway to restore the web content. Send (URL along link).

6 is a flowchart of a context aware personalization service method according to the present invention.

First, context information about a user's use of a link is reflected in a web access pattern (step 610). This process 610 may include generating a web access pattern graph including an edge connecting the plurality of nodes by reflecting the context information. In this case, the web access pattern graph may be a graph in which an edge value is allocated according to an access frequency between two nodes connected by one edge.

Next, neighbor nodes are selected using the web access pattern (step 620). This process 620 may include selecting nodes as neighboring nodes whose edge value is greater than or equal to a predetermined threshold. Preferably, this process (step 620) may include a process of weighting a specific node according to the time spent by the user in the specific node.

When the neighbor nodes are selected, the current state information of the user is generated using the selected neighbor nodes, and the next state information is generated by applying the current state information to the Markov chain (S630).

Finally, the link predicted by the user according to the next state information is output to the screen of the user terminal (step 640).

FIG. 7 is a detailed flowchart of the Markov chain application process 630 of FIG. 6.

First, a start vector is calculated by analyzing the most recent web access data (step 731). In this case, the most recent web access data may be stored in storage connected to a gateway with the access analyzer and the controller.

Next, the probability of the next state is calculated by multiplying the state transition matrix representing the nodes before moving to the current node by the start vector (step 732).

Finally, the next state information is generated according to the probability of the next state (step 733).

The present invention is a system that uses a user profile and user location information together, grasps the web requirements that change according to the user's situation, and provides a specialized personal service according to the user's situation.

As a method for grasping the user's web requirements, the Markov chain algorithm, which is a mathematical prediction technique, is introduced to calculate the collected information. This enables us to go beyond just understanding the user's web requirements, predicting the user's web requirements, and presenting the service before the user requests it. In addition, in order to overcome poor conditions or characteristics of a mobile display, a content division display method implemented through a web access pattern graph may be provided to the user terminal.

The present invention goes beyond the level of the existing system that provides services according to the needs of the users, grasps the service usage pattern according to the user's situation, predicts the user's web requirements, and presents the web services before the user requests. . In addition, in order to overcome the small size and proportion of a mobile display, a content division display method suitable for a user's web request step is provided to the user terminal.

Preferably, a program for executing the context-aware personalization service method of the present invention on a computer may be recorded and provided on a computer-readable recording medium.

The invention can be implemented via software. When implemented in software, the constituent means of the present invention are code segments that perform the necessary work. The program or code segments may be stored on a processor readable medium or transmitted by a computer data signal coupled with a carrier on a transmission medium or network.

The computer-readable recording medium includes all kinds of recording devices in which data is stored which can be read by a computer system. Examples of computer-readable recording devices include ROM, RAM, CD-ROM, DVD ± ROM, DVD-RAM, magnetic tape, floppy disks, hard disks, optical data storage devices, and the like. The computer readable recording medium can also be distributed over network coupled computer devices so that the computer readable code is stored and executed in a distributed fashion.

Although the present invention has been described with reference to one embodiment shown in the drawings, this is merely exemplary and will be understood by those of ordinary skill in the art that various modifications and variations can be made therefrom. However, such modifications should be considered to be within the technical protection scope of the present invention. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

As described above, according to the present invention, even when there is no user's request, the user's web access pattern can be analyzed to predict the user's demand, and the predicted link is presented to reduce the number of user's terminal operations, There is an effect that can provide a service optimized for a limited display and interface.

Claims (11)

delete delete delete delete delete delete In the system comprising a service provider and a plurality of terminals, The service provider An access analyzer for reflecting context information of a user's link usage in a web access pattern; Storage for storing the web access pattern; And A controller configured to generate current state information of the user using neighbor nodes selected by the terminal, generate next state information by applying the current state information to a Markov chain, and transmit the next state information to the user's terminal Including, The terminal is An extractor for selecting neighbor nodes using the web access pattern and extracting links predicting the user's request according to the transmitted next state information; And Context-aware personalization service system comprising a display for outputting the extracted links. The method of claim 7, wherein The access analyzer A context-aware personalization service system, comprising: generating a web access pattern graph including edges connecting a plurality of nodes by reflecting the context information. The method of claim 8, The web access pattern graph Edge values are assigned based on the frequency of access between two nodes connected by one edge, The extractor Context selection personalization service system, characterized in that for selecting the nodes that the edge value is greater than or equal to a predetermined threshold as neighboring nodes. The method of claim 9, The extractor Context-sensitive personalization service system, characterized in that the user weights the specific node according to the time spent by the specific node. The method of claim 7, wherein The controller Compute the start vector by analyzing the most recent web access data, and calculate the probability of the next state by multiplying the start vector by the state transition matrix representing the nodes before moving to the current node, and according to the probability of the next state. Context-aware personalization service system, characterized by generating state information.

Images