KR20190103499A - Method and apparatus for recommending contents - Google Patents

Abstract

Disclosed are a content recommendation method and an apparatus thereof. The method is operated by at least one processor wherein a content recommendation apparatus including content database communicates with a user terminal to recommend content to the user terminal. The method comprises: a step of receiving an access request including a user identification key from the user terminal; a step of determining a user group to which the user terminal belongs based on the user identification key; and a step of transmitting a recommended content list set in the determined user group to the user terminal. The recommended content list includes content extracted from the content database by using similarity between a plurality of content stored in the content database calculated based on a tag extracted from a plurality of content selected or played from the user terminal. According to the present invention, it is possible to provide environment in which a user can easily select and use content based on the user′s preference by recommending content based on user content usage information.

Description

Content recommendation method and device thereof {METHOD AND APPARATUS FOR RECOMMENDING CONTENTS}

The present invention relates to a content recommendation method and an apparatus thereof.

Currently, a service for providing content to a mobile communication terminal through a wireless Internet has become popular. The types of content provided to the mobile communication terminal include image content, music content, video content, game content, and real-time information content. It is very diverse and it is increasing in its variety.

However, in order to use such content, a user must bear a lot of time and communication cost to access a content providing server and search for a desired content.

In the related art, various methods for selecting and recommending content that a user is satisfied with have been presented. For example, there is a simple recommendation method using content sales volume, buyer recommendation information, and the like. However, in the case of a simple recommendation method based on sales volume, there is a problem in that the recommendation accuracy is lowered because the user's personal preference cannot be reflected.

As another example, there is a personalized recommendation method using preference genre information or preference category information input by each user, or purchase history of each user. In the case of such a customized recommendation method, since only the user's preference information is considered, there is a limit to recommending the optimal content customized to the user's preference in consideration of all the contents that are continuously changing and added. In addition, since the genre, category, etc. of the content to be recommended are limited, there is a problem that various contents cannot be recommended.

The problem to be solved by the present invention is to personalize highly accurate personalization without limitation of genre or category by recommending the tags included in the content and the content that is substantially related to the content used by the user based on the content usage history of the user The present invention provides a content recommendation method and apparatus capable of providing content recommendation information.

According to an aspect of the present invention, a content recommendation method is operated by at least one processor, and a content recommendation device having a content database communicates with a user terminal to recommend content to the user terminal. Receiving an access request including a user identification key from a terminal, determining a user group to which the user terminal belongs based on the user identification key, and recommending content list set in the determined user group to the user terminal And transmitting the content list, wherein the recommended content list includes contents extracted from the content database using a similarity between a plurality of contents selected or reproduced in the user terminal and contents stored in the content database. Similarity is It is calculated on the basis of the tags extracted from the plurality of content groups.

The recommended content list is provided through an application executed in the user terminal, and the user identification key includes an application unique key set in the application and stored in a storage area of the application in a storage space of the user terminal. can do.

After the transmitting, receiving feedback information including a plurality of pieces of content information selected or reproduced by the user terminal from the user terminal, and transmitting the plurality of pieces of content information to a user action log corresponding to the user identification key. Recording and periodically comparing a plurality of contents recorded in the user action log with a list of recommended contents for each user group, and when there is no user group that satisfies a threshold condition, the user group matching the user identification key. And updating the recommended content list.

After the determining, if there is no user group to which the user terminal belongs, the method may further include transmitting a default content list.

The updating may include determining whether a user group including a plurality of contents recorded in the user action log includes a threshold number, and if there is no user group including the threshold number, a user matching the user identification key. Establishing a group, extracting tags included in each content from the plurality of contents, and sorting the extracted tags in order of increasing frequency of duplicate occurrences for each content, and including contents including at least one sorted tag The method may include extracting from a content database, and generating the extracted contents as a list of recommended contents of the newly created user group and transmitting the extracted contents to the user terminal.

The feedback information includes a content ID and a content playing time, and prior to the updating, a weight is set for each content recorded in the user action log based on a ratio between the content playing time and the total playing time of the corresponding content. The method may further include extracting the tags included in each content from the plurality of contents, and applying the weight of the contents including the respective tags to each extracted tag to apply the weights of the extracted tags. The method may include calculating a duplicate occurrence frequency for, and sorting the extracted tags in the order of high calculated duplicate appearance frequencies.

The recommended content list of the newly created user group may be sorted in ascending order of the tags and page view (PV) of the contents.

According to another feature of the invention, the content recommendation device is a communication device connected to a user terminal through a communication network, a content DB consisting of a plurality of content, a user action log recorded a plurality of content information selected or reproduced in the user terminal A storage device including a recommendation content list group DB including a DB and a recommendation content list for each user group, a memory for generating a recommendation content list of a user group to which the user terminal belongs and storing the same as a program for transmitting to the user terminal; And at least one processor for executing the program, wherein the program extracts the plurality of pieces of content information from feedback information received from the user terminal and records the plurality of content information in a user action log corresponding to the user identification key, and the user. Recorded in the action log Sorting the extracted tags based on a duplicate occurrence frequency of each of the contents extracted from a plurality of contents, extracting contents including at least one sorted tag from the contents DB, and matching the extracted contents to the user identification key Instructions for generating, storing, and transmitting the recommended content list of the user group to be included.

The program determines a user group based on a user identification key received from the user terminal, and if there is no user group, transmits a default content list, and displays the content playback time and corresponding content of the content extracted from the feedback information. Instructions for setting a weight of each of the plurality of contents based on a ratio between total playing times, and calculating the duplicate occurrence frequency using the weights, wherein the weights are determined by the content playing time being the total playing time. The larger the occupancy ratio, the closer to one.

The program periodically compares a plurality of content IDs recorded in the user action log with a list of recommended content for each user group, and determines whether there is a recommendation content list including a plurality of content IDs or more. Thus, if present, it may include instructions for maintaining the already created user group and, if not present, creating a new user group.

According to an embodiment of the present invention, by recommending content on the basis of the content usage information of the user, it provides an environment in which the user can easily select and use the content according to his or her taste.

In addition, by identifying a user based on an application unique key or terminal unique information executed in the user terminal and recommending the content, a personalized recommendation is possible even for a user who is not logged in, and has a cumbersome user operation such as membership registration or login procedure. I never do that.

1 is a block diagram of a content recommendation system according to an exemplary embodiment of the present invention.
2 is a detailed block diagram of a content recommendation apparatus according to an embodiment of the present invention.
3 is a flowchart illustrating a process in which a content recommendation apparatus communicates with a user terminal to recommend content according to an embodiment of the present invention.
4 is a flowchart illustrating an operation of a content recommendation apparatus according to an embodiment of the present invention in detail.
5 is a flowchart illustrating in detail the operation of the content recommendation apparatus according to another embodiment of the present invention.
6 is a flowchart illustrating a feedback process of a user terminal according to an embodiment of the present invention.
7 is a flowchart illustrating a process of setting weights based on a playback time of a clip image executed in a user terminal according to an embodiment of the present invention.
FIG. 8 is a flowchart illustrating a process of extracting a clip image for content recommendation based on the weight set in FIG. 7.
9 is a block diagram showing a hardware configuration of a user terminal according to an embodiment of the present invention.
10 is a block diagram illustrating a hardware configuration of a content recommending apparatus according to another embodiment of the present invention.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Throughout the specification, when a part is said to "include" a certain component, it means that it can further include other components, without excluding other components unless specifically stated otherwise.

In addition, the terms “… unit”, “… unit”, “… module” described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software. Can be.

In the present specification, a clip movie is a relatively short movie, and a normal playback time is an image of a minute unit. Such a clip image may be referred to as a partial movie obtained by editing or extracting the entire movie constituting the program for a specific purpose.

The tag is descriptive information about the content and may be defined as an extensible content description unlike metadata for describing the content in a predetermined format. For example, for the content of sound source A, the metadata includes standardized description information such as genre, singer, production year, producer, and the like. On the other hand, the tag may describe not only the information included in the metadata, but also various descriptive information such as an advertisement using the sound source A, a drama, a main listener of the sound source A, and the like.

Page View (hereinafter referred to as 'PV') refers to the display of a page of an Internet website at the request of a user or a unit of counting the number of such requests. As will be described later, the PV of the clip image refers to the number of times a web page including the clip image is displayed or requested according to a user's request.

A content recommendation method and apparatus thereof according to an embodiment of the present invention will now be described with reference to the drawings.

1 is a configuration diagram of a content recommendation system according to an embodiment of the present invention, and FIG. 2 is a detailed configuration diagram of a content recommendation apparatus according to an embodiment of the present invention.

Here, examples of the contents may include music contents, video contents, audio contents, game contents, and the like. However, in the exemplary embodiment of the present invention, the clip image is described as the contents.

Referring to FIG. 1, the content recommendation system includes a content recommendation apparatus 100, a plurality of user terminals 200, and a communication network 300.

The content recommendation device 100 may be a server computer device provided by an online video on demand (VOD) service provider that provides various clip images. The content recommendation apparatus 100 provides a personalized clip image recommendation service to a user who cannot specify personal information without undergoing a login process. In other words, it is recommended that clip images that the user is expected to prefer. In addition, the content recommendation apparatus 100 may provide various VOD contents in addition to a clip image.

Here, the personalized clip video recommendation service may be provided as one service included in the online VOD service, or the personalized clip video recommendation service itself may implement one independent online VOD service.

The user terminal 200 mounts the application 201 and accesses the content recommendation apparatus 100 through the application 201 to receive a personalized clip image recommendation service.

The application 201 is an application that is additionally installed and executed by a user. For example, the application 201 may be downloaded and installed from an app store or the like. The application 201 may be a smart application operating on a smart platform.

The application 201 executes a clip image provided from the content recommending apparatus 100 by interposing a media player or interworking with a media player installed in the user terminal 200.

The communication network 300 connects the plurality of user terminals 200 to the content recommendation apparatus 100. For example, the communication network 300 may include a local area network (LAN), the Internet, a long term evolution (LTE) network, a wideband code division multiple access (WCDMA) network, Wi-Fi, WIBRO, or the like. However, this is only an example and is not limited thereto.

FIG. 2 is a detailed configuration diagram of a content recommendation apparatus according to an exemplary embodiment, and illustrates a detailed configuration of the content recommendation apparatus 100 of FIG. 1.

2, the content recommendation apparatus 100 includes a clip image DB 101, a tag DB 103, a tag connection DB 105, a user recommendation connection DB 107, and a recommendation clip image list group DB 109. , A user action log DB 111, a recommended content generator 113, a feedback collector 115, and a recommended content transmitter 117.

The clip image DB 101 stores clip images that are periodically updated or updated in real time. The clip images are updated by the service provider, and the update target may include, for example, a popular clip image of which PV meets a predetermined criterion, and a clip image of a service target provided by the service provider.

The popular clip picture list and the recommended clip picture list are based on these clip pictures. The clip image is data composed of a plurality of columns as shown in Table 1 below.

Column name type example Explanation clip_seq number 4544 Clip key value clip_media String /clip/media/aaa.mp4 Media file location to play clip_title String Yoo Jae-seok running Clip comment program_title String Infinite Challenge Registered program name status String Y State of use page_view number 90 The number of clicks or requests for pages with clip footage created date 2017-12-29 12:53:02 Registered Date

The tag DB 103 stores tags of each clip image. The tag is data composed of a plurality of columns as shown in Table 2 below.

Column name type example Explanation tag_seq number 215 Tag key values tag_name String Yoo Jae Suk Tag content status String Y State of use

The tag connection DB 105 stores tag connection data for connecting a clip image and a tag of the clip image. Tag connection data is composed as following Table 3.

Column name type example Explanation clip_seq number 4544 Clip key value tag_seq number 215 Tag key values status String Y State of use

According to Table 3, the clip key values are linked to Table 1, and the tag key values are linked to Table 2. For example, the tag of the clip image 4544 is 215, and referring to Table 2, it can be seen that the tag corresponding to 215 is 'Yoo Jae-seok'.

The user recommendation connection DB 107 stores user recommendation connection data which is user group information corresponding to the user identification key. Here, the user identification key is information for identifying a user who has not logged in.

The user identification key may use terminal unique information such as an International Mobile Subscriber Identity (IMSI), terminal hardware information such as a media access control address (MAC), and the like. Use

The application 201 is installed in the user terminal 200 through the app store, and when installed, a unique key is issued to the application 201. The unique key may be a universally unique identifier (UUID).

The application 201 operates in conjunction with the content recommendation apparatus 100. The unique key issued to the application 201 is stored in the storage area of the application 201 among the storage space of the user terminal 200. Even if the user does not register or log in to the application 201, the user is identified through a unique key issued to the application 201.

The user recommendation connection data is composed of a plurality of columns as shown in Table 4 below.

Column name type example Explanation user_seq String 0134fae2-ec9e-11e7-ad07-0a8f510e37e9 User identification value cg_id number 10221 User group key value created date 2017-12-29 12:53:02 Creation date

The recommended clip image list group DB 109 stores a recommended clip image list generated for each user group. Each recommended clip video list is composed of a plurality of columns as shown in Table 5 below.

Column name type example Explanation cg_id number 10221 User group key value clip_seq_list number (4133,5162,4333,1822,1927,1111) Recommended clip video list created date 2017-12-29 12:53:02 Creation date

According to Table 5, the user group key value is linked to Table 4, and the recommended clip image list is composed of clip key values indicating each of the plurality of clip images. These clip key values are linked in Table 1.

The user action log DB 111 stores the user action log generated based on the feedback information received from the user terminal 200. The user action log is data composed of a plurality of columns as shown in Table 6.

Column name type example Explanation user_seq String 0134fae2-ec9e-11e7-ad07-0a8f510e37e9 User identification value action_type String R Action type
(R: City Hall) action_target String {"4544": "2017-12-29 12: 51: 37.429226": "1", "4232": "2017-12-29 14: 31: 17.429226", "0.5", "4471": "2017- 12-29 20: 20: 22.898920 ":" 0 "} Watched clip video ID: Viewing time: Weight updated date 2017-12-29 12:53:02 Update date

According to Table 6, user_seq contains a user identification key value for identifying the user who transmitted the feedback information. That is, the application unique key is stored. In action_type, user's action pattern classification information is recorded. Behavior patterns can be defined as R (view), F (Favorite). The action_target includes a weight assigned according to the watched clip video ID, the viewing time, and the playback time of the clip video. As such, the information contained in the user action log is accumulated. updated contains the most recent date.

The recommended content generating unit 113 generates a list of recommended clip images to be transmitted to the user terminal 200 connected through the communication network 300.

The recommended content generating unit 113 generates a recommended clip image list by using similarity between the plurality of clip images selected or reproduced in the user terminal 200 and the clip images stored in the clip image DB 101. In this case, the similarity is calculated based on a tag extracted from a plurality of clip images selected or reproduced in the user terminal 200. That is, the duplicate appearance frequency in each clip image of the extracted tag or the duplicate appearance frequency applied with a weight according to the playback time of the clip image may be used as the similarity.

The feedback collecting unit 115 receives feedback information including a clip image ID periodically or when executing a clip image from the user terminal 200, generates a user action log, and generates a user action log by the user action log DB 111. Store in

The recommended content transmitter 117 transmits the recommended clip image list generated by the recommended content generator 113 to the user terminal 200.

Operations of the recommended content generator 113, the feedback collector 115, and the recommended content transmitter 117 will be described in detail with reference to FIGS. 3 and 8.

3 is a flowchart illustrating a process in which a content recommendation apparatus communicates with a user terminal to recommend content according to an embodiment of the present invention.

Referring to FIG. 3, the application 201 is executed in the user terminal 200 (S101) and is connected to the content recommendation apparatus 100 (S103).

The recommended content generating unit 113 of the content recommending apparatus 100 may obtain a user identification key through the access request S103 received from the user terminal 200. The user identification key may include hardware information such as a medium address control (MAC) address of the user terminal 200 and a unique key assigned to the application 201.

The recommended content generation unit 113 determines a user group based on the user identification key (S105). The recommended clip video list of the determined user group is extracted (S107). The recommended content generator 113 checks a user group key corresponding to the user identification key from Table 4. The recommended clip video list corresponding to the checked user group key is extracted from Table 5.

The recommended content transmitter 117 of the content recommendation apparatus 100 transmits the recommended clip image list extracted by the recommended content generator 113 to the user terminal 200 (S109).

When the user selects a clip image from the recommended clip image list (S111), the application 201 transmits feedback information including the selected clip image ID to the content recommendation apparatus 100 (S113). In this case, the clip image may be selected only from the recommended clip image list or may be played simultaneously with the selection.

The feedback collecting unit 115 of the content recommending apparatus 100 records a user action log of the user terminal 200 based on the received feedback information (S113) (S115). This user action log is generated as shown in Table 6. Thereafter, the information recorded in the user action log is used to extract the recommended clip image list in step S107.

4 is a flowchart illustrating an operation of a content recommendation apparatus according to an embodiment of the present invention in detail.

Referring to FIG. 4, the recommended content generating unit 113 of the content recommending apparatus 100 checks a user group to which a user belongs based on a user identification key (S201). For example, the unique key of the application executed in the user terminal 200 is checked, and the user group key matching the application unique key is checked from Table 4.

The recommended content generator 113 determines whether a user group corresponding to the user identification key exists (S203). That is, it may be determined whether the user group key matching the application unique key exists in Table 4.

In operation S203, if it is determined that there is no user group, the recommended content transmission unit 117 transmits the recommended clip image list extracted by the recommended content generation unit 113 to the user terminal 200 (S205). Here, the popular clip image list includes a plurality of clip images extracted from the clip images stored in the clip image DB 101 in order of increasing PV values. The list of popular clip images may be sorted in ascending order of recommendation, and each recommendation clip image may include a clip image file link as shown in the 'clip_media' column of Table 1.

If it is determined in step S203 that the user group exists, the recommended content generating unit 113 checks the user group corresponding to the user identification key, and extracts the recommended clip image list of the user group (S207). The recommended content generation unit 113 checks the user group key corresponding to the user identification key from Table 4. The recommended clip video list corresponding to the user group key is extracted from Table 5 below.

The recommended content generating unit 113 compares the extracted clip image list (S207) with the clip image list recorded in the user action log of the user identification key (S209), and the clip image already watched exists in the recommended clip image list. It is determined whether (S211). The recommended content generating unit 113 determines whether there is a clip image ID that matches the clip image ID included in the 'action_target' column of Table 6 among the clip image IDs recorded in the 'clip_seq_list' column of Table 5 (S211).

If there is a matching clip image ID, the recommended content generating unit 113 removes the matching clip image ID, that is, the clip image that the user has already watched from the recommended clip image list (S209) (S213).

The recommended content generating unit 113 extracts the unviewed popular clip images of the removed number from the popular clip image list or the clip image DB 101 and adds them to the recommended clip image list (S215). Here, the recommended content generating unit 113 of the clip images of Table 1 is a clip in the order in which the 'page_view' column has the highest value among the recommended clip image list and clip images not included in the 'action_target' column of Table 6 The video can be extracted as a popular clip video.

The recommended content transmission unit 117 transmits the recommended clip image list extracted by the recommended content generation unit 113 to the user terminal 200 (S217).

The feedback collector 115 receives feedback information including the ID of the clip image selected by the user from the application 201 of the user terminal 200 (S219). The feedback information may be received when the clip image is selected or periodically.

The feedback collector 115 records the clip image ID included in the feedback information in a user action log corresponding to the user identification key (S221). That is, the clip image ID is recorded in the 'action_target' column corresponding to the user identification key of Table 6. If the viewing information includes the viewing time, the viewing time is recorded.

The recommended clip image list is periodically updated, and the period may be in minutes. For example, it may be 5 minutes. The recommended clip image list is updated by receiving feedback of the clip image selection information of the user from the user terminal 200.

5 is a flowchart illustrating a content recommendation method according to another embodiment of the present invention, illustrating a process of periodically updating a recommendation clip image list.

In addition, when the application 201 is executed in the user terminal 200, the content recommendation apparatus 100 determines that the user terminal 200 is in a connected state and accesses a page using the application of FIG. 5. You can perform the operation.

Referring to FIG. 5, when a predetermined period arrives (S301), the recommended content generating unit 113 of the content recommending apparatus 100 searches for the user action log DB 111 (S303).

The recommended content generator 113 determines whether a user action log corresponding to the user identification key of the connected user terminal 200 exists (S305).

If no user action log exists, the step ends.

If there is a user action log, the recommended content generation unit 113 compares the clip image list recorded in the user action log with the previously generated recommended clip image list for each user group (S307). In operation S309, it is determined whether the number of clip image IDs corresponding to the clip image IDs recorded in the user action log among the recommended clip image lists for each user group meets the threshold number. That is, the recommended content generating unit 113 may determine that the number of clip image IDs that match each other among the clip image IDs included in the 'action_target' column of Table 6 and the clip image list included in the 'clip_seq_list' of Table 5 is, for example, 15 You can determine if more than.

In step S309, if there is a user group including a clip image ID of more than a threshold number, the user group is matched and stored with the user terminal 200, that is, the user identification key (S311). That is, in this case, the previously created user group is maintained.

On the other hand, if at least one user group including more than a threshold number of clip image IDs exists, a process of creating a new user group is performed. That is, the recommended content generating unit 113 extracts each tag from the clip images recorded in the user action log (S313). The recommended content generator 113 obtains tag key values corresponding to clip image IDs included in the 'action_target' column of Table 6 from Table 3, and extracts all tags from Table 2 based on the tag key values (see FIG. S313).

The recommended content generating unit 113 sorts the extracted tags in the order of increasing frequency of duplicate appearance in each clip image (S315). At this time, if the overlapping frequency of the tags is the same, the sequence of the tags is sorted in ascending order.

The recommended content generating unit 113 extracts each clip image ID including at least one of the extracted plurality of tags from the clip image DB 101 based on Table 3.

The recommendation content generating unit 113 puts the sorting order of the tags as the highest priority, and if the sorting order of the tags is the same, the clip content IDs are sorted in the order of the page views (PV) (S317). The sorted clip image IDs are generated as a recommended clip image list (S319).

When the recommended content generation unit 113 describes a series of processes (S315 to S319) for generating a list of recommended clip images by way of example, it is as follows. At this time, for convenience of description, unlike the above description, the tag ID and the clip image ID are set to text.

Clip image ID Tag ID PV Watched A a, b, c, d 88 City Hall B a, e 23 City Hall C f, g, h, a, g 31 City Hall D a, g 50 City Hall E d, e 92 City Hall F a, g, f 43 Unviewed G b, f, h 90 Unviewed H c, d, g, e 68 Unviewed I a, h 71 Unviewed

Referring to Table 7, the tag with the highest frequency of duplicate appearance among the tags included in the clip images watched by the user is 'a' included in all of A, B, C, and D. In this way, tags are sorted based on the overlapping frequency (S315). In this case, when the duplicate occurrence frequency is the same, the tag having the sequence key first, that is, 'a' takes precedence over 'b'. This sorting results in a> d> g> e> b> c> f> h.

Clip images that are not viewed by the user are sorted in order of priority of tag sorting (S317). For example, the clip image IDs including a are F and I. In this case, priority is given to the clip video with many PV. In other words, I has more PV than I, so I is sorted first. This sorting results in I> H> F> G. That is, the recommended content generating unit 113 generates a list of recommended clip images arranged in the order of I-H-F-G (S319).

The recommended content generating unit 113 generates a user group key of the recommended clip image list (S321).

The recommended content generating unit 113 matches the recommended clip video list with the user group key and stores the recommended clip video list in the recommended clip video list group DB 109 (S323).

The recommended content generating unit 113 matches the user identification key with the user group key and stores the same in the user recommendation connection DB 107 (S325).

6 to 8, when the recommended clip image list is updated, the user terminal 200 receives feedback whether the recommended clip image has actually been viewed, and applies weights according to the viewing time.

First, FIG. 6 is a flowchart illustrating a feedback process of a user terminal according to an exemplary embodiment of the present invention, and illustrates a process of collecting a viewing time of a clip image.

Referring to FIG. 6, when the user selects a clip image from the recommended clip image list (S401), the application 201 of the user terminal 200 may provide feedback information including the selected clip image ID of the content recommendation apparatus 100. The feedback collecting unit 115 transmits (S403).

The feedback collecting unit 115 of the content recommending apparatus 100 stores the clip image ID included in the feedback information in the user action log DB 111 corresponding to the user identification key of Table 6 (S405).

After the application 201 of the user terminal 200 plays the selected clip image (S407), the playback time is periodically measured according to the setting (S409). Then, the feedback information including the measured playback time and the clip image ID is transmitted to the feedback collector 115 of the content recommendation apparatus 100 (S411).

When the reproduction of the clip image is terminated (S413), the application 201 of the user terminal 200 may provide feedback information including the information indicating the end of reproduction and the clip image ID of the feedback collecting unit 115 of the content recommendation device 100. Transmit to (S415).

The feedback collector 115 of the content recommendation apparatus 100 calculates a total playback time based on the feedback information received in steps S411 and S415, and stores the clip image ID and the playback time in the user action log DB 111. (S417).

7 is a flowchart illustrating a process of setting weights based on a playback time of a clip image executed in a user terminal according to an embodiment of the present invention.

Referring to FIG. 7, the recommended content generating unit 113 of the content recommending apparatus 100 confirms a playback time of each clip image from the user action log DB 111 (S501).

The recommended content generating unit 113 determines whether the reproduction of the clip image is completed based on the reproduction time, that is, whether the user has finished watching the clip image (S503).

If it is determined that the reproduction of the clip image is completed, the recommended content generating unit 113 considers that the clip image matches the taste of the user and sets the weight to 1 (S505).

If it is determined that the reproduction of the clip image is not the case, the recommended content generating unit 113 determines whether the playback time checked in step S501 exceeds 50% of the total playback time in step S507.

If it is determined that the content exceeds 50%, the recommended content generating unit 113 considers that the clip image is somewhat consistent with the user's taste but contains some unwanted contents, and sets the weight to 0.5 (S509).

If it is determined that it is 50% or less, it is determined whether the checked playback time exceeds 20% of the total playback time (S511).

If it is determined that the content exceeds 20%, the recommended content generating unit 113 sets the weight to 0.1 by considering that the clip image does not match the user's taste, but is interested in the additional information of the clip image (S513).

If it is determined that it is 20% or less, the recommended content generating unit 113 considers that the clip image is inconsistent with the user's taste and sets the weight to 0 (S515).

Referring to FIG. 8, a process of generating a recommended clip video list using the weights set as described above is described.

FIG. 8 is a flowchart illustrating a process of extracting a clip image for content recommendation based on the weight set in FIG. 7.

In this case, FIG. 8 omits the same steps as in FIG. 5, and illustrates only the steps corresponding to steps S313 to S317 of FIG. 5, which is a step in which weights are used.

Referring to FIG. 8, the recommended content generating unit 113 extracts each tag from the clip images recorded in the user action log (S601).

The recommended content generator 113 calculates a duplicate occurrence frequency by applying a weight of a clip image including each tag for each tag (S603). The tags are sorted in the order of the high frequency of duplicate appearance (S605).

The clip image ID and the tag ID are shown in Table 8, and at this time, the tag ID and the clip image ID are set to text as shown in Table 7 for convenience of description. At this time, the difference from Table 7 is added to the weight.

Clip image ID Tag ID PV Watched weight A a, b, c, d 88 City Hall One B a, e 23 City Hall 0.5 C f, g, h, a, g 31 City Hall One D a, g 50 City Hall 0.1 E d, e 92 City Hall 0.5 F a, g, f 43 Unviewed - G b, f, h 90 Unviewed - H c, d, g, e 68 Unviewed - I a, h 71 Unviewed -

According to Table 8, the tags are a, b, c, d, e, f, g, h, as shown in Table 9 when the weight of the duplicate occurrence frequency of such a tag.

Tag ID Duplicate occurrence frequency a (1 × 1) + (1 × 0.5) + (1 × 0.1) = 1.6 b (1 × 1) = 1 c (1 × 1) = 1 d (1 × 1) + (1 × 0.5) = 1.5 e  (1 × 0.5) + (1 × 0.5) = 1 f (1 × 1) = 1 g (1 × 1) + (1 × 0.1) = 1.1 h (1 × 1) = 1

According to Table 9, since a is included in the clip images IDs A, B, and D watched by the user, and the weights of A, B, and D are 1, 0.5, and 0.1, respectively, the frequency of overlapping a is (1 It is calculated as x 1) + (1 x 0.5) + (1 x 0.1) = 1.6. In this manner, duplicate occurrence frequencies of all tags are calculated (S603). When the tags are sorted in the order of the high frequency of duplicate occurrence calculated as described above (S605), a> d> g> b = c = e = f = h.

If it is determined in step S607 that the recommended content generating unit 113 has the same overlapping frequency, the recommended content generating unit 113 sorts the tags having a large number of duplicate appearance frequencies excluding weights in priority order (S609). That is, in the case of b = c = e = f = h, since the frequency of overlapping occurrence of e contained in B and E is high, the order is e> b = c = f = h. For the rest of the tags, the sequence keys are sorted in order of precedence. That is, the order is b> c> f> h (S611). Thus, tags are sorted as a> d> g> e> b> c> f> h.

The recommendation content generating unit 113 puts the sorting order of the tags as the top priority, and if the sorting order of the tags is the same, the clip image IDs are sorted in the order of the page views (PV) (S613).

Referring to Table 8, the clip images F, G, H, and I, which are not viewed by the user, are sorted in order of priority. Among the sorted tags, clip image IDs including a sorted first are F and I. In this case, priority is given to the clip video with many PV. Arranging the clip images in this manner results in I> H> F> G.

9 is a hardware block diagram of a terminal according to an embodiment of the present invention, which shows a hardware configuration of the user terminal 200 described with reference to FIGS. 1 to 8.

Referring to FIG. 9, the terminal 400 is composed of hardware including a communication device 401, a memory 403, a display device 405, an input device 407, at least one processor 409, and the like. A program that is executed in conjunction with hardware is stored in a designated location.

The communication device 401 is connected to the processor 409 to transmit and / or receive data through a communication network 300 (in FIG. 1). The memory 403 is connected to the processor 409 and stores a program including instructions for executing the configuration and / or method according to the embodiments described with reference to FIGS. 1 to 8. The program implements the present invention in combination with hardware such as memory 403 and processor 409.

The display device 405 is connected to the processor 409 and outputs data according to the embodiments described with reference to FIGS. 1 to 8 on the screen. The input device 407 is connected to the processor 409 and is a means for a user input operation according to the embodiments described with reference to FIGS. 1 to 8. In this case, the display device 405 and the input device 407 may be implemented as one device.

10 is a hardware block diagram of a content recommending apparatus according to another embodiment of the present invention, and shows a hardware configuration of the content recommending apparatus 100 described with reference to FIGS. 1 to 8.

Referring to FIG. 10, the content recommendation apparatus 500 includes a communication device 501, a memory 503, a storage device 505, and at least one processor 507.

The communication device 501 is connected with at least one processor 507 to transmit and / or receive data via a communication network 200 (in FIG. 1). The memory 503 is connected to at least one processor 507 and stores a program including instructions for executing the configuration and / or method according to the embodiments described with reference to FIGS. 1 to 8. The program implements the invention in combination with hardware such as memory 503 and at least one processor 507. The processor 507 executes the present invention in combination with hardware such as the memory 503, the storage device 505, and the like.

The storage device 505 includes information necessary for the operation of the content recommending device 500. The storage device 505 includes the clip image DB 101, the tag DB 103, the tag connection DB 105, the user recommendation connection DB 107, the recommended clip image list group DB 109, and the user described in FIG. 2. An action log DB 111 is included.

The embodiments of the present invention described above are not only implemented through the apparatus and the method, but may be implemented through a program for realizing a function corresponding to the configuration of the embodiments of the present invention or a recording medium on which the program is recorded.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements of those skilled in the art using the basic concepts of the present invention defined in the following claims are also provided. It belongs to the scope of rights.

Claims (10)

A method of recommending content to a user terminal by a content recommending device operated by at least one processor and having a content database in communication with the user terminal,
Receiving an access request including a user identification key from the user terminal,
Determining a user group to which the user terminal belongs based on the user identification key; and
Transmitting the recommended content list set to the determined user group to the user terminal;
The recommended content list,
Contents extracted from the content database by using similarity between the plurality of contents selected or reproduced in the user terminal and the contents stored in the content database,
The similarity is calculated based on a tag extracted from the plurality of contents. In claim 1,
The recommended content list,
Is provided through an application running in the user terminal,
The user identification key,
And an application unique key set in the application and stored in a storage area of the application in a storage space of the user terminal. In claim 1,
After the transmitting step,
Receiving feedback information including a plurality of content information selected or reproduced in the user terminal from the user terminal,
Recording the plurality of pieces of content information in a user action log corresponding to the user identification key; and
Periodically, a plurality of contents recorded in the user action log are compared with a list of recommended contents for each user group, and when there is no user group that satisfies a threshold condition, a list of user groups and recommended contents matching the user identification key is searched. Step to update
Further comprising, content recommendation method. In claim 3,
After the determining step,
If there is no user group to which the user terminal belongs, transmitting a default content list
Further comprising, content recommendation method. In claim 3,
The updating step,
Determining whether there is a user group including a plurality of threshold numbers or more of contents recorded in the user action log;
If no user group is included above the threshold number, establishing a user group matching the user identification key;
Extracting tags included in each content from the plurality of contents, and sorting the extracted tags in order of increasing frequency of duplicate occurrences for each content;
Extracting from the content database contents containing at least one sorted tag; and
Generating the extracted contents as a list of recommended contents of the newly created user group and transmitting the extracted contents to the user terminal;
Including a content recommendation method. In claim 5,
The feedback information,
Includes a content ID and a content duration,
Prior to the updating step,
Setting a weight for each content recorded in the user action log based on a ratio between the content play time and the total play time of the content;
The sorting step,
Extracting tags included in each content from the plurality of contents;
Calculating a duplicate occurrence frequency of each of the extracted tags by applying a weight of the content including each tag for each extracted tag, and
Sorting the extracted tags in order of calculated duplicate occurrence frequency
Including a content recommendation method. In claim 5,
The recommended content list of the newly created user group is
And a sorting order of the tags and a page view (PV) of the contents are arranged in ascending order. A communication device connected to a user terminal through a communication network,
A storage device including a content DB consisting of a plurality of contents, a user action log DB in which a plurality of content information selected or reproduced in the user terminal is recorded, and a recommended content list group DB including a recommended content list for each user group;
A memory for generating a list of recommended contents of the user group to which the user terminal belongs and storing the recommended content list as a program for transmitting to the user terminal;
At least one processor for executing the program,
The program,
Extracting the plurality of pieces of content information from the feedback information received from the user terminal, and recording the plurality of pieces of content information in a user action log corresponding to the user identification key;
Sorting the extracted tags based on a duplicate occurrence frequency of each of the contents of the tags extracted from the plurality of contents recorded in the user action log, extracting contents including at least one sorted tag from the contents DB,
And instructions for generating, storing, and transmitting the extracted contents as a list of recommended contents of the user group matching the user identification key. In claim 8,
The program,
Determine a user group based on a user identification key received from the user terminal, and if there is no user group, transmit a default content list,
Instructions for setting a weight of each of the plurality of contents based on a ratio between a content playback time in the user terminal extracted from the feedback information and a total playback time of the corresponding content, and calculating the duplicate occurrence frequency using the weights. Include,
The weight is,
The content recommendation apparatus is set to a value close to 1 as the ratio of the content reproduction time to the total reproduction time increases. In claim 8,
The program,
Periodically comparing the plurality of content IDs recorded in the user action log with recommended content lists for each user group, and determining whether there is a recommendation content list including the plurality of content IDs more than a threshold number, and if present, And instructions for maintaining the already created user group and, if not present, creating a new user group.

Images