KR101925950B1 - Method and device for recommending contents based on inflow keyword and relevant keyword for contents - Google Patents

Abstract

According to the present invention, there is provided a content recommendation method based on an influential search term for content, comprising the steps of: (a) selecting, based on a search record of users who accessed the content recommendation apparatus, Extracting an inflow search term for the keyword; (b) generating reverse index information by sorting the candidate contents based on the identified input query word after identifying influent search words of candidate contents; And (c) calculating scores of interest between the input search word extracted from the search record of the target user's terminal and the selection input record of the search result content and the influential search word of the candidate content included in the inverse index information, Wherein the input search word for the content includes at least one of a search window input keyword received from the content recommendation device and a search keyword input keyword received from the content recommendation device before receiving the user input for the link information of the content, to be.

Description

[0001] METHOD AND APPARATUS FOR RECOMMENDING CONTENTS [0002] BASED ON INFLOW KEYWORD AND RELEVANT KEYWORD FOR CONTENTS [

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a content recommendation method and recommendation apparatus based on an input search word and an associated search term for contents.

Recently, according to the development of Internet technology, a recommendation technique of contents corresponding to user's interest is applied to a search engine such as a portal site.

One of the conventional content recommendation techniques is collaborative filtering in which a target user connected to the server is recommended based on contents viewed by many other users having similar symbols. Because it reflects the interests of many users with similar preferences, it is likely to match the interests of the target user, but it does not directly reflect the interests of the target user. It is difficult to satisfy a desired user.

Meanwhile, according to another conventional content recommendation technique, there is a technique of analyzing the text of the content searched by the user and the texts of the contents belonging to the recommendation candidate group, and recommending matched contents among the recommended candidate groups to the user. However, in the case of this technique, since the relative accuracy of the text analysis is lowered, it is not always possible to provide contents suitable for the user's interest, and a considerable load is required in the analysis process.

Korean Patent Laid-Open Publication No. 10-2016-01216878 (published Oct. 20, 2016)

It is an object of the present invention to provide a technique for recommending contents to a user on the basis of an input search word for contents.

It is another object of the present invention to provide a technique for recommending contents to a user in consideration of not only an incoming search word but also an associated search term for the content.

According to an embodiment of the present invention, a content recommendation method based on an influential search term for content, which is performed by a content recommendation apparatus, includes the steps of: (a) Extracting an influential search word for each content based on a selection input record for the content; (b) generating reverse index information by sorting the candidate contents based on the identified input query word after identifying influent search words of candidate contents; And (c) calculating scores of interest between the input search word extracted from the search record of the target user's terminal and the selection input record of the search result content and the influential search word of the candidate content included in the inverse index information, Wherein the input search word for the content includes at least one of a search window input keyword received from the content recommendation device and a search keyword input keyword received from the content recommendation device before receiving the user input for the link information of the content, to be.

According to another embodiment of the present invention, a content recommendation method based on an influential search term for content, which is performed by a content recommendation apparatus, includes: (a) a search history of users who access the content recommendation apparatus, Extracting an influential search word and an associated search term for each content based on a selection input record for the content; (b) generating inverse index information by identifying influent search terms and associated search terms of candidate content, and then sorting the candidate content based on the identified influent search terms and associated search terms; And (c) calculating a score related to the interest between the influent search word extracted from the search history of the terminal of the target user and the selection input record of the search result content, the related search word extracted from the input search term, and the influent search word included in the inverse index information, Selecting candidate content among the candidate contents and providing the selected candidate content to the terminal of the target user, wherein the influent search word for the content includes a recommendation of the content recommendation right before receiving the user input for the link information of the content, And the related search word for the content is a search word excluding the input search word among the search window input keywords received by the content recommendation apparatus before receiving the user input for the link information of the content .

According to another embodiment of the present invention, a content recommendation apparatus based on an influential search term for content may further include a content recommendation apparatus for recommending content based on a search record of users who accessed the content recommendation apparatus, A content query extracting unit for extracting an influential search word for each content; An inverse index generating unit for generating inverse index information by sorting the candidate contents based on the identified influent search word after identifying influent search words of candidate contents; And calculating a score of interest between the input search word extracted from the search record of the terminal of the target user and the selection input record of the search result content and the influential search word of the candidate content included in the inverse index information, And a recommendation content providing unit for providing the selected content to the terminal of the target user, wherein the influent search word for the content is a search window input keyword received by the content recommendation apparatus before receiving the user input for the link information of the content .

According to another embodiment of the present invention, a content recommendation apparatus based on an influential search term for content may further include a content recommendation apparatus for recommending content based on a search record of users who accessed the content recommendation apparatus, A content search term extraction unit for extracting an influent search term and an associated search term for each content; An inverse index generating unit for generating inverse index information by sorting the candidate contents based on the identified influent search terms and related search terms after identifying influent search terms and related search terms of candidate contents; And calculating a score of an influential search word and an associated search term extracted from a selective input record of a search result of the terminal of the target user and a search result of the search result, and a score of an interest between an influential search word and an associated search word of candidate content included in the inverse index information, And a recommended content providing unit for providing recommended content among the candidate contents to the terminal of the target user, wherein the incoming search word for the content includes a content recommendation message And the related search term for the content is a search term other than the influent term among the search window input keywords received by the content recommendation apparatus before receiving the user input for the link information of the content.

According to the present invention, contents are recommended based on an influential search word and an associated search word for a content and a profile vector of a target user, so that it is possible to accurately grasp a user's interest and provide accurate contents according to the user's interest have.

In addition, since the content is recommended based on the input search word and the related search term simply without analyzing the texts constituting the content, the load on the server can be further reduced compared to the conventional technique.

1 is a structural diagram of a system according to an embodiment of the present invention.
2 is a block diagram of a configuration of a content recommendation apparatus according to an embodiment of the present invention.
FIG. 3 is a conceptual diagram for explaining the meaning of an input search word and an associated search word for content according to an embodiment of the present invention.
4 is a conceptual diagram illustrating an example of profile profile vectors and association profile vectors that can be generated for each content according to an exemplary embodiment of the present invention.
FIG. 5 is a conceptual diagram illustrating an example of inverse index information that sorts and represents contents based on an input search word or an associated search word, according to an embodiment of the present invention.
FIG. 6 is a conceptual view conceptually illustrating a process of extracting keyword-based recommendation content and association-based recommendation content using a profile vector of a target user according to an embodiment of the present invention.
FIG. 7 is a flowchart illustrating a content recommendation method based on an influential search term for content according to an exemplary embodiment of the present invention. Referring to FIG.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, which will be readily apparent to those skilled in the art. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when a part is referred to as being "connected" to another part, it includes not only "directly connected" but also "electrically connected" with another part in between . Also, when an element is referred to as " comprising ", it means that it can include other elements as well, without departing from the other elements unless specifically stated otherwise.

In this specification, the term " part " includes a unit realized by hardware, a unit realized by software, and a unit realized by using both. Further, one unit may be implemented using two or more hardware, or two or more units may be implemented by one hardware. On the other hand, 'to' is not limited to software or hardware, 'to' may be configured to be an addressable storage medium, and may be configured to play one or more processors. Thus, by way of example, 'parts' may refer to components such as software components, object-oriented software components, class components and task components, and processes, functions, , Subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. The functions provided in the components and components may be further combined with a smaller number of components and components or further components and components. In addition, the components and components may be implemented to play back one or more CPUs in a device or a secure multimedia card.

The " user terminal " mentioned below may be implemented as a computer or a portable terminal capable of accessing a server or other terminal through a network. Here, the computer includes, for example, a notebook computer, a desktop computer, a laptop computer, and the like, each of which is equipped with a web browser (WEB Browser), and the portable terminal may be a wireless communication device (BLE, Bluetooth Low Energy), NFC, RFID, Ultrasonic, Infrared, WiFi, and LiFi as well as smart phones, tablet PCs and wearable devices. And may include various devices mounted thereon.

The term " network " means a connection structure in which information can be exchanged between each node such as terminals and servers, and includes a local area network (LAN), a wide area network (WAN) : World Wide Web), wired / wireless data communication network, telephone network, wired / wireless television communication network, and the like. Examples of wireless data communication networks include 3G, 4G, 5G, 3GPP, LTE, WIMAX, Wi-Fi, Bluetooth, infrared, Communications, Visible Light Communication (VLC), LiFi, and the like.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

Referring to FIG. 1, a system according to an embodiment of the present invention includes a user terminal 100 and a server 200.

The server 200 may be a server 200 operating a search engine provided in a portal site or the like (i.e., a service for providing corresponding contents among all the contents interposed on the online for an arbitrary search word) . The server 200 may also refer to a device that recommends content based on an entry query for content.

That is, when the user terminal 100 inputs an arbitrary search term on the screen, the query related to the search term is transmitted from the user terminal 100 to the server 200, and the server 200 transmits the corresponding contents to the user terminal 100).

In this process, the search term entered by the user terminal 100 into the search window of the search engine may be recorded in the cookie of the user terminal 100 or may be recorded in the server 200. The server 200 can extract the search terms of the content input from the user terminal 100 to accurately derive the user's interest information and select contents matching the user's interest information among the existing contents, And can be provided as recommended content to the terminal 100.

In one embodiment of the present invention, the recommendation contents are selected based on an input search word and an associated search term for the contents input from the user terminal 100 in providing the recommendation contents, .

The server 200 according to an exemplary embodiment of the present invention may be configured to include a processor that executes a program and a memory in which a program (or an application) for performing a content recommendation method based on an input query based on content is stored. Here, the processor may perform various functions according to the execution of the program stored in the memory. As shown in FIG. 2, the detailed modules included in the processor are classified into a content query extracting unit 210, a target user information extracting unit 220 ), An inverse index generator 230, and a recommended content provider 240.

The content query term extracting unit 210 extracts a search record of users who have accessed the server 200 and have performed content search, and a record of selecting and inputting (e.g., clicking) the content provided as a search result along with the search record And extracts an influential search term and an associated search term for each content.

Herein, an influent search term for content means a keyword entered directly into the content, which means a search window input keyword received by the server 200 right before receiving a user input for link information of the content. In other words, after the user inputs a keyword "patent" into the search window of the search engine and a user's selection input (eg, click and touch) on the link of the specific content among the displayed search results occurs, &Quot; may be an influent search word of a specific content.

The related search term for the content is a keyword indirectly involved in the flow of the content, and is a keyword that is input before the user input for the link information of the content, . This remaining search term usually refers to search terms input in one session (for example, an activity interval before the user leaves the search engine service by server 200). In other words, if a user clicks on a link of a specific content after inputting a keyword "patent" while performing a web surfing while inputting keywords such as "patent attorney" and "trademark" in a search window, "patent attorney" A " trademark " may be an associated search term for a particular content. Alternatively, if a user searches for a " trademark ", clicks a link related to an arbitrary content, and then clicks a link of a specific content placed in the page, the " trademark "

For example, referring to FIG. 3, if the keywords input by the users immediately before entering the bottom content (the article titled "The sentence is correct") were "LG Twins", " These keywords become influent search words for the content. On the other hand, if keywords such as " Kia Tigers ", " wild card ", and " post season " are input into the search window before inputting the influent search word,

The content search term extraction unit 210 can grasp the search history of the users based on the log information of the users recorded in the user terminal 100 or the server 200. [ Based on the search history, it is possible to grasp what contents the user has viewed and which keyword is entered in the search window.

Next, the content query term extraction unit 210 generates a main keyword profile vector for the incoming query word and generates a related word profile vector for the associated keyword word. Since the influent query is a keyword that directly contributes to the influx of the content, it can be said as a main word of the content if interpreted differently.

For example, as shown in FIG. 4, a plurality of influent search words and a plurality of related search words can be derived for any content. At this time, the value of the associative profile vector may be determined differently depending on whether each user has an influent search word and an associated search word for the content, and how many users have utilized a specific influent search word or a specific related search word have. In FIG. 4, the "LG Twins" has the highest value of the subject keyword profile vector and the "wildcard" has the highest value of the association profile vector for the content shown in the figure (the article titled " Is shown.

The content search term extraction unit 210 selects the content in which a user's click has occurred for a predetermined period of time in advance and searches for the content having the highest input count among the plurality of influent search terms related to the selected content and the related search terms, It is possible to generate a keyword and associate profile vector for the search term and related search term. The content search term extracting unit 210 sets the same value for all of the content search terms or generates a vector based on the click inflow number of the content search term or using Equation 1 as below, ) To normalize (normalize) the final vector value.

[Equation 1]

In the equation (1), clicks _{kwd ^ contents} means the number of times a click is generated on a target content after inputting a specific input query or an associated search word, and clicks _total indicates a number of times Means all clicks, clicks _kwd is the number of times that an arbitrary content click occurs after inputting a specific input query or an associated search word, and clicks _contents is the total number of clicks of the target content exposed as a search result.

At this time, the " clicks " is calculated within the result of counting the past from a certain period to the present. On the other hand, when the " number of clicks " is replaced with " number of users who click ", the value due to the repeat action of the same user can be removed.

The target user information extracting unit 220 generates a profile vector of the target user. For example, the target user information extractor 220 adds the main subject profile vectors of the contents that have been searched for a predetermined period at the target user's terminal, thereby generating a profile vector of the target user. The target user information extracting unit 220 extracts the contents retrieved by the user based on the retrieval record of the target user terminal for a predetermined period and the records obtained by selecting and inputting the contents provided as the retrieval result along with the retrieval records, The inflow search word can be grasped, and the keyword profiles corresponding to the inflow search word identified are identified. In addition, if the incoming query is too large, only the top few influent queries with high scalar values of the vector terms may be configured as the target user profile vector. For example, the target user profile vector may be configured as the profile vector shown in FIG.

Meanwhile, as a further embodiment, the target user information extracting unit 220 may extract an associated search word of the content, generate a related-word profile vector, and configure a target user profile vector including the input query and the associated query .

The inverse index generator 230 may generate inverse index information in which the candidate contents are sorted based on the influential search words on the main keyword profile vector or the related search words on the related word profile vector.

Specifically, the inverse index generator 230 filters the contents generated within a predetermined period from the current time as candidate contents. In this case, the preset period may be shorter than the period set for searching for the content viewed by the target user in order to derive the profile vector of the target user. At this time, the inverse index generator 230 may filter the candidate contents by setting the predetermined period differently for each category of contents. For example, content that is important in terms of time, such as news, can be filtered by content within the last few hours. In the case of other contents such as a blog or social media, content of a day or week . ≪ / RTI >

Then, the inverse index generator 230 indexes the main word profile vector of the candidate content and the associative profile vector. The profile vector of the candidate content and the associated profile vectors may have already been generated by the content search term extraction unit 210. The inverse index generation unit 230 may generate an inverse search term on the main profile profile vector and an association And generates the inverse index information by constructing a set of candidate contents corresponding to each search term for each search term (i.e., an influential search term and an associated search term). In this case, the inverse index information is generated by arranging the candidate contents in the descending order of the vector anti-scalar value of the subject keyword profile vector in each set.

For example, referring to FIG. 5, when an incoming search word LG twins is extracted, a keyword profile vector for an incoming search term LG twins may be indexed for each candidate content. The contents (hereinafter referred to as "A content") titled "KIA.LG wild card match ticket sold out in 35 minutes" (hereafter referred to as A content) and the content titled "Glossy jumper for winning [SS photo] LG twins and wild card" Can include "LG Twins" and "Yang Sansum" as influent search words, respectively. In this case, the A content may have a value of 0.6 for the main word profile vector for the LG Twins, and the B content may have a value of 0.5 for the main word profile vector for the LG Twins. The inverse index generation unit 230 generates indexes of the incoming query terms by sequentially sorting the contents having a high value based on the incoming query term " LG Twins " can do. In addition, it is also possible to generate the related term inverse index information by performing a similar process for the related term.

At this time, in the contents sorted for each influential search word or each associated search term, the inverse index generator 230 selects the candidate contents ranked only from the highest ranking to a predetermined rank, thereby calculating the number of candidate contents included in the inverse index information It can also be filtered.

The recommended content providing unit 240 calculates a score related to the degree of interest between the incoming search word and the associated search word of the candidate content included in the influent search word and inverse index information extracted from the log information of the target user terminal, And provides it to the terminal of the target user. Specifically, the score of interest between the target user's profile vector extracted from the main subject profile vectors of the content searched at the terminal of the target user, the main subject profile vector of the candidate content included in the inverse index information, and the association profile vector is calculated Recommended content can be selected.

First, the recommended content providing unit 240 internalizes the profile profile vector of the candidate content included in the profile vector of the target user, the candidate profile content vector of the candidate content, and the profile profile vector of each candidate content included in the related keyword inverse index information, Calculate the score of the candidate content. The score at this time may be a score regarding the above interest.

Specifically, the recommended content providing unit 240 extracts a keyword having a scalar value of a vector value on the profile vector of the target user that is not equal to zero (i.e., an influential search word). For example, the subject keywords in the target user's profile vector in FIG. 6 all extract all the influent search terms in the profile vector of the target user because the vector value is greater than zero. Then, the recommended content providing unit 240 retrieves the candidate content corresponding to the input search word and the associated search term in the inverse index information corresponding to the keyword of the target user's profile vector, thereby generating a candidate content set Lt; / RTI > For example, in FIG. 6, it is judged whether the search word corresponding to the search term " LG Twins " is included in the influent term index information or the related term index information in the profile vector of the target user. And extracts the sorted candidate contents as one set. Then, the value of the profile vector of the target user and the profile vector of each content in the candidate content set are discarded. In this case, an inner product value is calculated for each candidate content, and this can be a score.

Through this process, scores of candidate contents related to the keyword and the score of the candidate contents related to the keyword are separately calculated by the influent keyword and the related keyword, so that two groups of keyword-based recommendation content and associative word recommendation content are derived .

The recommended content providing unit 240 may rank the candidate content items in descending order of the score of each candidate content item and provide a predetermined number of candidate contents to the user terminal 100 as recommendation contents according to the ranking value .

At this time, the recommended content providing unit 240 may alternatively provide the contents related to the main language and the related words alternately, or may provide a set of contents related to the main language to the top of the list provided to the user terminal 100 , And may provide a subset of the content related to the associated word.

Hereinafter, a content recommendation method based on an influential search term related to a content according to an embodiment of the present invention will be described in detail.

The following method is performed by the above-described server 200, so that even if there is an omission in the following, the above description is omitted.

The server 200 first extracts an entry search word and an associated search term of content from search click logs of all users connected to the server 200 for each content (S110).

The server 200 generates a main keyword profile vector for the incoming query and an associated keyword profile vector for the associated keyword (S120).

In addition, the server 200 generates a profile vector of the target user using the contents retrieved by the target user and the subject keyword profile vectors of the contents, which are left in the search record related to the target user and the search result content click record accompanying the search record (S130).

Next, the server 200 selects the candidate contents, and generates the index information based on the input keyword and the related keyword extracted from the candidate contents (S140). Specifically, the contents can be classified according to each search word, and the profile vector values for the search terms can be sorted and sorted in descending order.

Then, the server 200 selects recommended content matching the target user's interest based on the inverse index information and the profile vector of the target user, and provides the selected recommended content to the target user terminal 100 (S150). At this time, the profile vector of each content in the inverse index information and the profile vector of the target user are deduced, and it can be recommended that the inner product result score is best matched with the user's interest in the order of highest score.

One embodiment of the present invention may also be embodied in the form of a recording medium including instructions executable by a computer, such as program modules, being executed by a computer. Computer readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. In addition, the computer readable medium may include both computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

While the methods and systems of the present invention have been described in connection with specific embodiments, some or all of those elements or operations may be implemented using a computer system having a general purpose hardware architecture.

It will be understood by those skilled in the art that the foregoing description of the present invention is for illustrative purposes only and that those of ordinary skill in the art can readily understand that various changes and modifications may be made without departing from the spirit or essential characteristics of the present invention. will be. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. For example, each component described as a single entity may be distributed and implemented, and components described as being distributed may also be implemented in a combined form.

The scope of the present invention is defined by the appended claims rather than the detailed description and all changes or modifications derived from the meaning and scope of the claims and their equivalents are to be construed as being included within the scope of the present invention do.

100: user terminal 200: server
210: content search term extraction unit 220: target user information extraction unit
230: an inverse indexing unit 240: a recommended content providing unit

Claims (17)

A content recommendation method based on an influential search term for content, which is performed by a content recommendation apparatus,
(a) extracting an influential search word for each content based on a search record of users who have accessed the content recommendation apparatus and search for the content, and a selective input record of the search result content;
(b) generating reverse index information by sorting the candidate contents based on the identified input query word after identifying influent search words of candidate contents; And
(c) a score related to the degree of interest between the input search word extracted from the search record of the target user's terminal and the selection input record of the search result content and the influential search word of the candidate content included in the inverse index information, Selecting candidate content among the candidate contents and providing the selected candidate contents to the terminal of the target user,
Wherein the influent search word for the content is a search window input keyword received by the content recommendation device before receiving a user input for the link information of the content.
The method according to claim 1,
(a-1) generating a main language profile vector for the incoming query after the step (a); And
(a-2) extracting at least one influent keyword inputted by the target user from search records within a predetermined period of time of the target user's terminal and selective input records for the search result content, Further comprising generating a profile vector of the target user configured with a profile vector. 3. The method of claim 2,
The profile vector of the target user may include:
Wherein the input profile includes a plurality of influent search terms and a subject profile vector corresponding to each influent search term. 3. The method of claim 2,
The step (b) includes the steps of: sorting candidate contents including the influent search word for each influent search word of the candidate contents, and generating inverse index information sorted on the basis of a scalar value of a main word profile vector of the influent search word Lt; / RTI >
In the step (c), a score of the candidate contents is calculated based on the profile vector of the target user and the main language profile vector of the candidate content included in the inverse index information, thereby selecting the recommended contents among the candidate contents, To a user's terminal,
Wherein the main keyword profile vector is calculated and generated based on the number of clicks of the content generated after inputting the input query word.
5. The method of claim 4,
The step (b)
(b-1) filtering the contents generated within a predetermined period from the current time among the entire contents with candidate contents, and filtering the candidate contents by setting the predetermined period differently for each category of contents Based content recommendation method. 5. The method of claim 4,
The step (b)
(b-2) extracting main subject profile vectors of the candidate content, identifying influent search terms for the extracted subject keyword profile vector, constructing a set of candidate contents corresponding to the influent search word for each of the influent search words, And arranging the index information in a higher order as the scalar value of the main word profile vector of the influent search word is higher in each set
Wherein the content recommendation method is based on an inflow search term for content. The method according to claim 6,
The step (b-2)
And selecting candidate content ranked from a highest ranking to a predetermined rank for each entry query, and filtering the inverse index information. 5. The method of claim 4,
The step (c)
(c-1) calculating a score of the candidate content by inserting a subject keyword profile vector of the candidate content included in the profile vector of the target user and the inverse index information, Way. 9. The method of claim 8,
The step (c-1)
Extracting an influent keyword of a subject profile vector having a non-zero scalar value among at least one subject profile vector included in the profile vector of the target user, extracting an influential search word in the inverse index information corresponding to the extracted influent keyword, Collecting a candidate content set of influent search terms belonging to the profile vector of the target user by searching for the candidate content set; And
And calculating a score of each candidate content by subtracting a subject keyword profile vector of candidate contents belonging to the candidate content set and a profile vector of the target user. 9. The method of claim 8,
The step (c)
(c-2) ranking the candidate contents in descending order of scores, and providing a predetermined number of candidate contents as the recommended contents to the user terminal according to a ranking value; . A content recommendation method based on an influential search term for content, which is performed by a content recommendation apparatus,
(a) extracting an influential search word and an associated search term for each content based on a search record of users who have connected to the content recommendation apparatus and search for the content, and a selection input record of the search result content;
(b) generating inverse index information by identifying influent search terms and associated search terms of candidate content, and then sorting the candidate content based on the identified influent search terms and associated search terms; And
(c) an input search word and an associated search term extracted from a selective input record of search results and content of a search result of a target user, and a score related to an interest between an influential search word and an associated search word of candidate content included in the inverse index information, Selecting candidate content among the candidate contents and providing the selected candidate contents to the terminal of the target user by calculating based on the search words and the intra-vector related to the related search words,
Wherein the input search word for the content is a search window input keyword received by the content recommendation device immediately before receiving a user input for the link information of the content,
Wherein the related search term for the content is a search word input keyword received from the content recommendation apparatus before receiving the user input for the link information of the content, How to recommend content.
12. The method of claim 11,
(a-1) generating a main word profile vector for the incoming search word and a related word profile vector for the associated keyword after the step (a); And
(a-2) extracting at least one influent keyword inputted by the target user from search records within a predetermined period of time of the target user's terminal and selective input records for the search result content, Further comprising generating a profile vector of the target user configured with a profile vector. 13. The method of claim 12,
Wherein the step (b) comprises: arranging candidate contents including the input search word and candidate contents including the associated search term for each of the identified influent search words and each associated search term, And generating inverse index information sorted based on a scalar value of an association profile vector,
The step (c) may include calculating a score of the candidate content based on the profile vector of the target user, the main language profile vector of the candidate content included in the inverse index information, and the association profile profile vector, And providing the selected terminal to the terminal of the target user,
Wherein the keyword profile vector is calculated and generated on the basis of the number of clicks of the content generated after inputting the input query word and the associative word profile vector is calculated based on the number of clicks of the content generated after inputting the related keyword The content recommendation method based on an inflow search term for the content. 14. The method of claim 13,
The step (c)
(c-1) calculating a score of the candidate content related to a main keyword by inserting a main keyword profile vector of the candidate content included in the profile vector of the target user and the inverse index information, And calculating a score of the candidate content with respect to the association word by interpolating the association word profile vector of the candidate content included in the information. 15. The method of claim 14,
The step (c)
(c-2) ranking the scores of the candidate contents in descending order, providing the recommendation contents according to a ranking value, and alternately providing contents related to the main language and contents related to the association word, And providing contents related to the related word in a rearranged manner. The method of claim 1, wherein the contents are related to the related words. A content recommendation apparatus based on an influent search term for content,
A content query extracting unit for extracting an influential search word for each content based on a search record of users who have connected to the content recommendation apparatus and search for content and a selection input record of the search result content;
An inverse index generating unit for generating inverse index information by sorting the candidate contents based on the identified influent search word after identifying influent search words of candidate contents; And
Based on the in-vocabulary extracted from the search record of the terminal of the target user and the selection input record of the search result content and the score of the interest between the influent search word of the candidate content included in the inverse index information, And a recommendation contents providing unit for selecting recommended contents among the candidate contents and providing the selected contents to the terminal of the target user,
Wherein the influent search word for the content is a search window input keyword received by the content recommendation apparatus before receiving a user input for link information of the content.
A content recommendation apparatus based on an influent search term for content,
A content query extracting unit for extracting an influential search word and an associated search term for each content based on a search record of users who have connected to the content recommendation apparatus and search for content and a selection input record of the search result content;
An inverse index generating unit for generating inverse index information by sorting the candidate contents based on the identified influent search terms and related search terms after identifying influent search terms and related search terms of candidate contents; And
The input search term and the related search term extracted from the selective input record of the search result of the target user terminal and the search result contents, and the score of the interest between the influent search term and the related search term of the candidate content included in the inverse index information, And a recommended content providing unit for selecting the recommended content among the candidate contents and providing the selected content to the terminal of the target user,
Wherein the input search word for the content is a search window input keyword received by the content recommendation device immediately before receiving a user input for the link information of the content,
Wherein the related search term for the content is a search word input keyword received from the content recommendation apparatus before receiving the user input for the link information of the content, Content recommendation apparatus.

Images