KR102206001B1 - Apparatus and method for recommending e-books based on user behavior - Google Patents

Abstract

Disclosed is an electronic book recommendation apparatus and method based on user behavior. A preference for each e-book may be calculated based on behavior information of users who use the e-book, and similar users may be extracted by calculating the similarity between the recommended user and other users using the behavior information. It is possible to recommend an electronic book that is related to the user to be recommended by using the preference among the extracted electronic books used by similar users.

Description

E-book recommendation device and method based on user behavior {APPARATUS AND METHOD FOR RECOMMENDING E-BOOKS BASED ON USER BEHAVIOR}

The present invention relates to a device and method for recommending electronic books based on user behavior for recommending electronic books using behavior information of users who use electronic books. In particular, similar users using the similarity calculated using user behavior information The present invention relates to a device and method for recommending electronic books based on user behavior that extracts and recommends electronic books with high preference among electronic books used by similar users.

In general, there are methods for recommending e-books based on the purchase history of e-book users and meta information of products. For example, a user who has purchased a specific e-book recommends other e-books that have been purchased a lot, or a user who clicks a specific e-book recommends other e-books that have been clicked a lot. Mainly.

Association Rule Mining is a technology that recommends items with high possibility of simultaneous purchase or click at the same time for each product using purchase records of multiple users, and determines how many products to check the association between the length of the rule. I can. For example, if the length of the rule is 2, it can be checked to mean that two products, that is, an association rule between product A and product B, are calculated.

However, since the existing e-book recommendation method using such association rule mining only checks the purchase history and recommends it, it is not possible to check whether the user actually read the e-book after purchasing it. In addition, it is difficult to determine whether or not the user gave a good evaluation of the purchased e-book if the user does not use separate rating information. Therefore, if a user makes a recommendation based on a purchase history without knowing whether or not the user has read the purchased e-book or not, there is a concern that the quality of the recommendation may deteriorate.

Therefore, in order to recommend suitable e-books to individual users, data related to various behaviors left while reading e-books are collected for each user, and preference for e-books is determined based on the collected data. Therefore, there is an urgent need for an e-book recommendation technology based on user behavior that can recommend related books.

Korean Patent Publication No. 10-2013-0115016A, published on October 21, 2013 (Name: Method and apparatus for providing feedback related to e-books in terminals)

It is an object of the present invention to provide e-book preference information on content that e-book users prefer to use together with whether or not to actually use the e-book purchased.

In addition, it is an object of the present invention to recommend an electronic book that is more suitable to a user by using preference information collected through a user's actions when recommending an electronic book that is related to an electronic book user.

The electronic book recommendation apparatus according to the present invention for achieving the above object comprises: a preference calculating unit for calculating a preference for each electronic book of users based on behavior information of users using the electronic book; A similar user extracting unit that calculates a similarity between the recommendation target user and other users by using the behavior information of the users, and extracts one or more similar users with respect to the recommendation target user based on the calculated similarity; And an e-book recommendation unit that recommends at least one e-book among e-books used by the similar user to the recommendation target user based on the calculated preference.

In this case, the similar user extracting unit may acquire location information on the action information performed by the recommendation target user and other users, respectively.

In this case, the location information may include a start position and an end position of the highlight when the action information is a highlight, and when the action information is a screen capture, may include a position of the first letter and the last letter of the screen capture.

In this case, the similar user extraction unit may obtain distance information by using the location information and calculate a similarity degree based on the obtained distance information.

At this time, the similar user extraction unit replaces the location information obtained from the action information performed by the recommended target user and the location information obtained from the action information performed by the other user, respectively, with a line segment, and the distance using each replaced line segment. Information can be obtained.

At this time, the similar user extracting unit acquires the distance of the overlapped area as the distance information if there is an overlapping area between the replaced line segments, and if there is no overlapping area between the substituted line segments, between the respective line segments The distance between the two closest points of may be obtained as the distance information.

In this case, the similar user extraction unit may calculate the similarity by using an average value of distance information obtained between the recommendation target user and other users when the location information obtained from the action information performed by the recommendation target user is two or more.

In this case, the e-book recommendation device may further include an action information collection unit that collects the action information including one or more of the number of times of reading the e-book, the number of highlights, the content of the highlights, the number of screen captures, and the content of the screen capture.

In this case, the preference calculator may set the number of readings as the limit value if the number of readings is greater than or equal to a preset limit value.

In this case, the preference calculator may calculate a value obtained by summing two or more of the number of readings, the number of highlights, and the number of screen captures as the preference.

In this case, the preference calculator may assign a weight to each of the number of browsing, the number of highlights, and the number of screen captures, and calculate a value obtained by summing two or more of the number of browsing, the number of highlights, and the number of screen captures to which the weight is assigned. .

In this case, a weight for the number of views may be relatively lower than a weight for the number of highlights or a weight for the number of screen captures.

In addition, the e-book recommendation method according to the present invention includes: calculating user preferences for each e-book based on behavior information of users using e-books; Calculating a degree of similarity between the user to be recommended and other users by using the user's behavior information; Extracting one or more similar users for the recommendation target user based on the calculated similarity; And recommending one or more e-books from among e-books used by the similar user to the recommendation target user based on the calculated preference.

In this case, calculating the degree of similarity may include obtaining location information for action information performed by the recommendation target user and another user, respectively; Obtaining distance information by using the obtained location information; And calculating a similarity based on the obtained distance information.

In this case, the obtaining of the distance information may include replacing the location information on the action information performed by the recommendation target user and the location information on the action information performed by the other user with line segments, respectively; And acquiring distance information by using each of the replaced line segments.

At this time, the step of obtaining distance information using each line segment is to obtain the distance of the overlapping area as distance information when there is an area where each replaced line segment overlaps, and an area where each line segment overlaps exists. If not, the distance between the two closest points between each line segment may be obtained as distance information.

In this case, calculating the preference may include assigning weights to the number of views, the number of highlights, and the number of screen captures among the collected behavior information; And calculating a value obtained by summing two or more of the number of viewing times, the number of highlights, and the number of screen captures to which a weight is assigned, as the preference, wherein the weight for the number of viewing is a weight for the number of highlights or the number of screen captures. It can be given relatively lower than the weight for

According to the present invention, when recommending an e-book, it is possible to recommend a new e-book that is more relevant by using a preference based on an actual user's behavior value rather than a purchase history.

In addition, the present invention obtains preference information by applying a weight of an appropriate value for each user's actions, and then prioritizes e-books according to the obtained preference information to make recommendations, so that users can more easily select recommended books. Can help.

In addition, the present invention can recommend e-books of a wider category by obtaining e-books to be recommended from various similar users.

1 is a block diagram showing an electronic book recommendation device according to an embodiment of the present invention.
2 is a flowchart illustrating an electronic book recommendation method according to an embodiment of the present invention.
3 is an operation flow diagram showing in detail an e-book recommendation method according to an embodiment of the present invention.
4 is a flowchart illustrating in detail a process of calculating a preference among e-book recommendation methods according to an embodiment of the present invention.
5 is a diagram for explaining an example of determining similarity by using highlights.
6 is a diagram for describing an example of obtaining distance information for calculating a similarity.
7 is a diagram illustrating an e-book recommendation process according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, in the following description and the accompanying drawings, detailed descriptions of known functions or configurations that may obscure the subject matter of the present invention will be omitted. In addition, it should be noted that the same components are indicated by the same reference numerals as possible throughout the drawings.

The terms or words used in the present specification and claims described below should not be construed as being limited to a conventional or dictionary meaning, and the inventor is appropriate as a concept of terms for describing his own invention in the best way. It should be interpreted as a meaning and concept consistent with the technical idea of the present invention on the basis of the principle that it can be defined. Therefore, the embodiments described in the present specification and the configurations shown in the drawings are only the most preferred embodiments of the present invention, and do not represent all the technical spirit of the present invention, and thus various alternatives that can be substituted for them at the time of application It should be understood that there may be equivalents and variations. In addition, terms such as first and second are used to describe various elements, and are used only for the purpose of distinguishing one element from other elements, and are not used to limit the elements.

1 is a block diagram showing an electronic book recommendation device according to an embodiment of the present invention.

Referring to FIG. 1, an electronic book recommendation device 100 according to an embodiment of the present invention includes a behavior information collection unit 110, a preference calculation unit 120, a similar user extraction unit 130, and an electronic book recommendation unit ( 140) may be included.

The behavior information collection unit 110 may collect behavior information including one or more of the number of browsing, the number of highlights, the highlight content, the number of screen captures, and the screen capture content among actions that users who use the electronic book leave while reading the electronic book. have. At this time, the number of reading is the number of times the user has read the e-book, the number of highlights is the number of highlighted portions while reading the e-book, the highlight content is the text content of the highlighted portion, and the number of screen captures is the e-book. In addition, the number of times the screen of the e-book is captured and the content of the screen capture may include text content of the e-book for the portion where the screen is captured.

The preference calculation unit 120 may calculate a preference for each e-book of users based on behavior information left by users who use the e-book while reading the e-book. For example, a score may be calculated for each number of behavior information applied to an electronic book read by a user, and the summed value may be calculated as a preference for the corresponding electronic book.

In this case, if the number of times of reading the electronic book in the behavior information is more than a preset limit value, the number of reading may be set as the limit value. The limit value can be set to an appropriate value as the number of times required to read the book completely. For example, the average number of times users who have read the book can be calculated and set as the limit value. You can set the limit value of the number of readings to about 20 in advance, and if the total number of readings acquired while reading the e-book is more than the limit value, you can set the number of readings to the limit value of 20.

According to the reading habits of each user who purchases the e-book, the user can divide the e-book into several times, or read all e-books with a single reading. Even for such an electronic book, since there may be a large difference in the number of times to be viewed for each user, an appropriate limit value can be set to reduce an error in calculating preference.

In this case, a value obtained by summing two or more of the number of electronic books reading, the number of highlights, and the number of screen captures among the behavior information may be calculated as a preference. For example, each time the number of times of reading, the number of highlights, and the number of screen captures occurs, a preference score is given, and a value obtained by summing the preference scores given when the electronic books are all read may be calculated as a preference.

In this case, a weight is assigned to each of the number of times of reading the e-book, the number of highlights, and the number of screen captures, and a value obtained by adding at least two of the number of readings, the number of highlights, and the number of screen captures to which the weight is assigned may be calculated as a preference. For example, among the weights of 1, a weight of 0.5 is assigned to a highlight, a weight of 0.4 to the number of screen captures, and a weight of 0.1 to the number of views may be assigned to calculate preference.

The weight can be given to calculate the preference by placing weight on the corresponding behavior information if there is behavior information judged as an important factor in calculating the preference of an e-book. For example, if the number of highlights is determined as an important factor for calculating a preference among the number of times the electronic book is viewed, the number of highlights, and the number of screen captures, the preference can be calculated with a high weighting weight on the number of highlights.

In this case, a weight for the number of views may be given relatively lower than a weight for the number of highlights or a weight for the number of screen captures. When the number of viewing is compared with the number of highlights and the number of screen captures, since it is a criterion for determining a preference, which is not important, a weight may be given relatively low.

The similar user extraction unit 130 calculates the similarity between the recommended user who will recommend the e-book and other users who use the e-book by using the collected user's behavior information, and based on the calculated similarity, You can extract one or more similar users for. For example, for a specific electronic book, behavior information of a recommended user and behavior information of other target users may be compared, and similar users with high similarity calculated as a result of the comparison may be extracted. The similarity can be calculated by assigning points to overlapping portions between the recommended user and other users among the highlight content and screen capture content in the behavior information, and summing the points assigned to the e-book.

In this case, the similar user extraction unit 130 may obtain location information indicating where the action information is located in the e-book for action information performed by each of the recommendation target user and the other user in the e-book. For example, when the contents of an e-book are arranged and displayed as a line segment, the start point and the end point of a portion where highlights and screen captures performed by users are located may be displayed by connecting them with a line segment. Position information for highlights and screen captures can be expressed numerically by assigning numerical values for each letter or word included in the line segment indicating the contents of the electronic book in order from the starting point to the ending point of the electronic book. In this case, if the behavior information is a highlight, the start position and the end position of the highlight may be included, and if the behavior information is a screen capture, the position of the first letter and the last letter of the captured screen may be included.

In addition, the similar user extracting unit 130 may obtain distance information using the location information, and calculate a similarity between the recommended target user and other users based on the obtained distance information. The distance information may be a numerical representation of the degree of similarity between the recommended user and other users' actions performed by using location information, and it can be determined that the higher the value of the distance information, the higher the similarity between the recommended user and other users. have.

At this time, the location information obtained from the action information performed by the recommended user and the location information obtained from the action information performed by another user may be replaced with a line segment, and distance information may be obtained using each replaced line segment. .

In addition, if there is an overlapping area between each replaced line segment, the distance of the overlapped area is obtained as distance information, and if there is no overlapping area between each substituted line segment, the distance between the two closest points between each line segment is determined. Can be obtained with distance information. For example, when the highlight content of the recommended user is replaced with a line segment and the highlight content of another user is replaced with a line segment, if there is an area where each line segment overlaps, the distance of the overlapping area is obtained as distance information. I can. Similarly, in the case of screen capture, when the screen capture content of the recommended user is replaced with a line segment and the screen capture content of another user is replaced with a line segment, if there is an area where each line segment overlaps, the distance of the overlapping area is Can be obtained with distance information.

If there is no area where each replaced line segment overlaps, the distance between the two closest points between each line segment may be obtained as distance information. For example, if the start and end positions of the segment A where the highlight of the recommendation target user is replaced are 1 and 10, and the start and end positions of the segment B where the highlight of the other user is replaced are 15 and 30, between the two segments. 5, which is the distance between 10 and 15 closest to, can be obtained as distance information.

In this case, if the distance of the area where the line segments overlap is obtained as positive distance information, the distance between the two closest points between each line segment may be obtained as negative distance information.

In addition, the similar user extraction unit 130 may calculate the similarity by using the average value of distance information obtained between the recommendation target user and other users when the location information obtained from the action information performed by the recommendation target user is two or more. . For example, if the location information on two highlights performed by the recommended user is A and B, and location information on the highlight performed by another user is C, distance information and location information B of location information A and C Distance information of and C is obtained, and the average of each distance information can be calculated as a degree of similarity.

The e-book recommendation unit 140 may recommend one or more e-books among e-books used by similar users to a user to be recommended based on the calculated preference. For example, among the electronic books used by similar user A of the recommendation target user, an electronic book not used by the recommendation target user and having high preference may be recommended. If several similar users are extracted in this way, the electronic books can be recommended in the order of similar users with high similarity and in the order of high preference among the books viewed by the similar user.

In this way, not only the purchase history of the e-book, but also the preference information of the users who read the e-book through the similar user information extracted by calculating the e-book preference and similarity obtained by collecting the user's behavior information You can recommend e-books.

2 is a flowchart illustrating an electronic book recommendation method according to an embodiment of the present invention.

Referring to FIG. 2, the e-book recommendation method according to an embodiment of the present invention may calculate user preferences for each e-book based on behavior information that users who use e-books leave while reading e-books (S210). ). For example, a score may be calculated for each number of behavior information applied to an electronic book read by a user, and the summed value may be calculated as a preference for the corresponding electronic book.

In this case, if the number of times the user reads the e-book among the behavior information performed while reading the e-book is checked and the number of reads is greater than or equal to a preset limit value, the number of reading may be set to the same value as the limit value. In this case, the limit value is a value of the number of times of reading that is generally required until all of the electronic books are read, and an appropriate value may be set to reduce an error when calculating a preference using the number of readings. For example, it is possible to calculate the average number of times that users who have viewed the corresponding e-book and set a limit value for the e-book. If the average number of times the users who read the e-book is 20 times, the limit of the number of times is set to 20, and if a specific user reads a total of 30 times while reading the e-book, the number of times a particular user reads is the limit value. Can be modified to 20. As described above, according to the reading habits of the user who purchased the e-book, the e-book may be divided and read several times, or all e-books may be read through the first reading. Therefore, even for the same e-book, since there may be a large difference in the number of times to be viewed for each user, an appropriate limit value for the number of times to be viewed can be set to reduce an error in calculating preference.

In addition, a value obtained by summing two or more of the number of electronic books reading, the number of highlights, and the number of screen captures among the behavior information collected from the user who uses the e-book may be calculated as a preference. For example, each time a user reads an e-book and the number of reading, highlighting, and screen capture increases, a preference score is assigned to each item, and the preference scores given when the user reads all of the e-book You can add up to calculate your preference.

In addition, a weight can be assigned to each of the number of times that the e-book is viewed, the number of highlights, and the number of screen captures among the behavior information collected from the user who uses the e-book. The summed value can be calculated as a preference. For example, a weight of 0.5 is given to a highlight that is considered to be an important factor in calculating preference among the weights of total 1, a weight of 0.4 to the number of screen captures and a weight of 0.1 to the number of views are applied, and then the weight assigned to each value is applied. Preference can be calculated by summing the values. If there is behavior information judged to be a more important factor in calculating the user's preference for an e-book, a weight can be given to calculate the preference with a high weight on the corresponding behavior information.

In this case, a weight for the number of readings among the behavior information collected from a user who uses the e-book may be relatively lower than a weight for the number of highlights and a weight for the number of screen captures. The reason is that the number of browsing in the behavior information can be relatively low weight because it may be behavior information that is relatively unreliable as a criterion for determining preference when compared with the number of highlights in the behavior information and the number of screen captures. .

In addition, the e-book recommendation method according to an embodiment of the present invention may calculate a similarity between the recommendation target user and other users by using behavior information collected from a user using the e-book (S220).

At this time, with respect to action information performed on the e-book by the recommendation target user and the other user, location information indicating where the action information is located in the e-book may be obtained. For example, when an electronic book is represented as a line segment, the start point and the end point of a place where highlights and screen captures performed by users are located may be connected with a line segment to display the electronic book on a line segment. Position information for highlights and screen captures can be expressed numerically by assigning numerical values for each letter or word included from the start point to the end point of the e-book on the line segment representing the contents of the e-book.

In this case, if the behavior information is a highlight, the start position and the end position of the highlight may be included, and if the behavior information is a screen capture, the position of the first letter and the last letter of the captured screen may be included.

In addition, distance information may be obtained by using the location information obtained from the behavior information, and a similarity between the recommended target user and other users may be calculated based on the obtained distance information. The distance information may be a numerical representation of the degree of similarity between the recommended user and other users' actions performed by using location information, and it can be determined that the higher the value of the distance information, the higher the similarity between the recommended user and other users. have.

At this time, the location information obtained from the action information performed by the recommended user and the location information obtained from the action information performed by another user may be replaced with a line segment, and distance information may be obtained using each replaced line segment. .

In addition, if there is an overlapping area between each line segment with the behavior information replaced, the distance of the overlapped area can be obtained as distance information. If there is no overlapping area between each line segment with the behavior information substituted, between each line segment The distance between the two closest points of can be obtained as distance information. For example, if the position of the line segment replacing the highlight of the recommended user is 1 to 100, and the position of the line segment replacing the highlight of another user is 50 to 150, the overlapping area between each line segment, 50, is positive. Can be obtained with distance information of. If the position of the line segment replacing the highlight of another user is 200 to 250, 100, which is the distance between the two closest points between each line segment, may be obtained as negative distance information. Also, in the case of screen capture, when the screen capture content of the recommended user is replaced with a line segment and the screen capture content of another user is replaced with a line segment, if there is an area where each line segment overlaps, the distance of the overlapping area You can obtain as much as distance information.

In addition, when there are two or more location information obtained from the action information performed by the recommendation target user, the similarity may be calculated using the average value of distance information obtained between the recommendation target user and other users. For example, if the location information of the line segment that replaced the two highlights performed by the recommended user is A and B, and the location information of the line segment that replaced the highlight performed by another user is C, the location information of A and C By obtaining distance information and obtaining distance information of the location information B and C, an average value obtained by calculating the average of the obtained distance information may be calculated as a degree of similarity.

In addition, the e-book recommendation method according to an embodiment of the present invention may extract one or more similar users with respect to the recommended user based on the calculated similarity (S230). For example, by setting a reference value for the degree of similarity, 5 similar users may be extracted in the order of higher similarity among users having a degree of similarity equal to or greater than the reference value. In this case, the reference value for the degree of similarity or the number of similar users to be extracted may be changed according to circumstances.

In addition, the e-book recommendation method according to an embodiment of the present invention may recommend one or more e-books from among e-books used by similar users to a user to be recommended based on a preference for each e-book calculated from a user using the e-book. Yes (S240). In this case, among the electronic books used by similar users of the recommendation target user and not used by the recommendation target user, the electronic books may be recommended in the order of the highest preference. In addition, if several similar users are extracted, similar users may be listed in the order of high similarity, and electronic books may be recommended in the order of preference among electronic books used by similar users higher among the listed similar users.

3 is an operation flow diagram showing in detail an e-book recommendation method according to an embodiment of the present invention.

Referring to FIG. 3, the method for recommending an electronic book according to an embodiment of the present invention may collect behavior information of users who use the electronic book (S310).

In this case, the behavior information may include one or more of the number of times the electronic book is viewed, the number of highlights, the content of highlights, the number of screen captures, and the content of screen captures.

In addition, the e-book recommendation method according to an embodiment of the present invention may calculate user preferences for each e-book (S320). For example, a score may be calculated for each number of behavior information applied to an electronic book read by a user, and the summed value may be calculated as a preference for the corresponding electronic book.

In addition, in the method for recommending an electronic book according to an embodiment of the present invention, location information may be obtained from information about actions performed by users in order to extract a similar user to a user to be recommended (S330).

Also, in the method for recommending an electronic book according to an embodiment of the present invention, distance information for calculating a similarity may be obtained by using the obtained location information (S340). The distance information may be a numerical representation of the degree of similarity between the recommended user and other users' actions performed by using location information, and it can be determined that the higher the value of the distance information, the higher the similarity between the recommended user and other users. have.

At this time, the location information obtained from the action information performed by the recommended user and the location information obtained from the action information performed by another user may be replaced with a line segment, and distance information may be obtained using each replaced line segment. .

In addition, if there is an overlapping area between each replaced line segment, the distance of the overlapped area is obtained as distance information, and if there is no overlapping area between each substituted line segment, the distance between the two closest points between each line segment is determined. Can be obtained with distance information. For example, when the highlight content of the recommended user is replaced with a line segment and the highlight content of another user is replaced with a line segment, if there is an area where each line segment overlaps, the distance of the overlapping area is obtained as distance information. I can. Similarly, in the case of screen capture, when the screen capture content of the recommended user is replaced with a line segment and the screen capture content of another user is replaced with a line segment, if there is an area where each line segment overlaps, the distance of the overlapping area is Can be obtained with distance information.

In this case, if there is no area where each of the replaced line segments overlaps, the distance between the two closest points between each line segment may be obtained as distance information.

In this case, if the distance of the area where the line segments overlap is obtained as positive distance information, the distance between the two closest points between each line segment may be obtained as negative distance information.

In addition, in the method for recommending an electronic book according to an embodiment of the present invention, a similarity degree between a user to be recommended and another user may be calculated using the obtained distance information (S350). For example, a value of the obtained distance information may be calculated as a similarity, or if there are several values of the acquired distance information, an average value of the distance information may be calculated as a similarity.

In addition, the electronic book recommendation method according to an embodiment of the present invention may extract a similar user similar to the recommendation target user by using the calculated similarity (S360). For example, five similar users may be extracted in the order of high similarity, or similar users whose similarity is higher than the reference value may be extracted by creating a reference value for the similarity.

In addition, the e-book recommendation method according to an embodiment of the present invention may recommend one or more e-books among the extracted e-books used by similar users to a recommendation target user (S370).

4 is an operation flowchart showing in detail a process of calculating a preference among e-book recommendation methods according to an embodiment of the present invention.

Referring to FIG. 4, in the process of calculating a preference among e-book recommendation methods according to an embodiment of the present invention, behavior information of e-book users may be collected (S410).

In this case, the behavior information may include one or more of the number of views, the number of highlights, the highlight content, the number of screen captures, and the screen capture content.

In addition, in the process of calculating the preference of the e-book recommendation method according to an embodiment of the present invention, it may be determined whether the number of readings in the collected behavior information is greater than or equal to a preset limit value (S415). The limit value can be set to an appropriate value as the number of times required to read the book completely. For example, the average number of times users who have read the book can be calculated and set as the limit value. You can set the limit value of the number of readings to about 20 in advance, and if the total number of readings acquired while reading the e-book is more than the limit value, you can set the number of readings to the limit value of 20.

As a result of the determination in step S415, if the number of reading is greater than or equal to the limit value, the number of reading may be set as the limit value (S420).

If the number of times of reading is less than the limit value as a result of the determination in step S415, a weight may be assigned to each of the number of times of reading the e-book, the number of highlights, and the number of screen captures (S430). The weight can be given to calculate the preference by placing weight on the corresponding behavior information if there is behavior information judged as an important factor in calculating the preference of an e-book. For example, if the number of highlights is determined as an important factor for calculating a preference among the number of times the electronic book is viewed, the number of highlights, and the number of screen captures, the preference can be calculated with a high weighting weight on the number of highlights.

In this case, a weight for the number of views may be given relatively lower than a weight for the number of highlights or a weight for the number of screen captures. When the number of viewing is compared with the number of highlights and the number of screen captures, since it is a criterion for determining a preference, which is not important, a weight may be given relatively low.

In addition, in the process of calculating the preference of the e-book recommendation method according to an embodiment of the present invention, the number of reading, the number of highlights, and the number of screen captures to which a weight is assigned may be summed. For example, among the weights of 1, a weight of 0.5 is assigned to a highlight, a weight of 0.4 to the number of screen captures, and a weight of 0.1 to the number of views are assigned, and each score may be summed to calculate preference.

In addition, in the process of calculating the preference among the e-book recommendation method according to an embodiment of the present invention, a value obtained by summing the number of times of reading, the number of highlights, and the number of screen captures of the e-book may be calculated as a preference. As for the preference, as the sum of the number of times of reading, the number of highlights, and the number of screen captures of the e-book is larger, it may be determined that the preference for the e-book is higher.

5 is a diagram for explaining an example of determining similarity by using highlights.

Referring to FIG. 5, for user 1, user 2, and user 3, behavior information and purchase history of e-book 1 to e-book 5 can be checked, respectively.

User 1 performed 2 highlighting actions when using e-book 1, and User 2 and 3 performed 1 highlighting action for e-book 1, respectively.

When checking the purchase history, User 1 purchased and used e-books 1, 2 and 5, User 2 purchased and used all e-books 1 to 5, and User 3 used e-books 1, 2, 4, and 5. You can see what you purchased and used.

If, based on the highlight information of the e-book 1, a book is recommended to the user 1 by using the e-book recommendation method according to the present invention, highlight information on the e-book 1 of the users may be obtained first.

If the highlight of user 1 in e-book 1 is that the contents of e-book 1 in Figure 5 are the entire contents of the book, the location information of the first highlight is from the first letter to the 25th letter, and the location information of the second highlight is from the 132th letter. It can be seen that it is up to the 145th letter.

In addition, the location information of the highlight of the user 2 in e-book 1 is from the first letter to the 85th letter, and the location information of the highlight of the user 3 in the e-book 1 is from the 105th letter to the 119th letter.

At this time, if the similarity between User 1, User 2, and User 3 is calculated, the distance information obtained by using the location information of the highlights of User 1 and User 2 is a distance equal to the distance from the distance information 25 for the overlapping area. The information is -47, and the average value of the distance information, -11, may be the degree of similarity between User 1 and User 2.

In addition, the distance information obtained by using the highlight location information of User 1 and User 3 is -80 and -13, and the average value of the distance information is -80 and -13 because there is no overlapping area. May be similarity.

Therefore, it can be determined that User 2 is a similar user than User 3, and among the e-books User 1 has not used, among the e-books User 2 has not used, an e-book with a higher preference of User 2 is given to User 1. I can recommend it.

In the drawing shown in FIG. 5, since only highlight information can be checked, similarity is calculated using only highlight information, but distance information for a screen capture can be obtained and used for similarity calculation.

6 is a diagram for describing an example of obtaining distance information for calculating a similarity.

Referring to FIG. 6, distance information for calculating the similarity may be obtained by representing highlight information as a line segment as shown in FIG. 6.

In order to calculate the similarity between User A and User B, the behavior information is represented as a line segment.When there is no area where the line segments overlap as in Case 1, there is an area where the line segments overlap as in Case 2, and 3 As in case 1, there may be a case where one line segment includes another line segment.

In the case of case 1, the closest distance between each line segment can be obtained as distance information. In this case, the distance apart can be expressed as a number, and can be expressed as either a negative value or a positive value.

In case 2, the distance of the area where each line segment overlaps may be obtained as distance information. In this case, the distance of the overlapping region may be expressed as a numerical value, and may be expressed as either a positive value or a negative value.

In the case of case 3, as in the case of case 2, the distance of the area where each line segment overlaps may be obtained as distance information.

7 is a diagram illustrating an e-book recommendation process according to an embodiment of the present invention.

Referring to FIG. 7, information on actions performed by a recommended user while using an electronic book may be collected.

Using information necessary for calculating preference and information for extracting similar users from among the behavior information collected from users, preference information for each e-book of the user to be recommended can be obtained, and similar users to the user to be recommended can be extracted. .

In addition, information on actions performed while using the e-book used by the recommended user may be collected from the extracted similar users.

Among behavior information collected from similar users, preferences for e-books may be calculated using information necessary for calculating preferences, and information on the calculated preferences may be stored as preference information for each e-book.

The e-book recommendation device may recommend an e-book with a high preference to the recommendation target user by using the preference information for each e-book of a similar user with respect to the recommendation target user.

In addition, in order to recommend an electronic book to a similar user of the user to be recommended, another similar user to the similar user may be extracted and the electronic book may be recommended.

In this way, by collecting behavior information of users who use e-books using the e-book recommendation device and method, it is possible to recommend e-books suitable for the preferences of users by acquiring preference information for each e-book and similar user information. .

The electronic book recommendation method according to the present invention may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Including magneto-optical media, and any type of hardware device specially configured to store and execute program instructions such as ROM, RAM, flash memory, etc. Examples of the program instructions may include not only machine language codes such as those produced by a compiler but also high-level language codes that can be executed by a computer using an interpreter or the like. These hardware devices may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

As described above, the apparatus and method for recommending e-books based on user behavior according to the present invention are not limited to the configuration and method of the embodiments described as described above, but the embodiments may be variously modified. All or part of each of the embodiments may be selectively combined and configured.

According to the present invention, the user's preference for each e-book is calculated based on the behavior information of users who use the e-book, and similar users similar to the user to be recommended are extracted, You can recommend books. Through such a reliable e-book recommendation service, users can improve the income of e-book sellers by purchasing and using a wider variety of e-books, and further, by using the information and recommendation history of users, You can obtain recommended book information.

100: electronic book recommendation device 110: behavior information collection unit
120: preference calculation unit 130: similar user extraction unit
140: e-book recommendation department

Claims (18)

A preference calculation unit that calculates preferences of users for each e-book based on behavior information of users using e-books;
A similar user extracting unit that calculates a similarity between the recommendation target user and other users using the behavior information of the users, and extracts one or more similar users to the recommendation target user based on the calculated similarity; And
An e-book recommendation unit that recommends at least one e-book among e-books used by the similar user to the recommended user based on the calculated preference
Including,
The similar user extracting unit acquires location information for action information performed by the recommendation target user and another user, respectively, obtains distance information using the location information, and calculates the similarity based on the obtained distance information. And
The location information, when the action information is a highlight, includes a start position and an end position of the highlight, and when the action information is a screen capture, the electronic book includes a position of a first letter and a location of a last letter of the screen capture. Recommended device. delete delete delete delete delete The method according to claim 1,
The similar user extraction unit
When the location information obtained from the action information performed by the recommendation target user is two or more, the electronic book recommendation device is characterized by calculating a similarity by using an average value of distance information obtained between the recommendation target user and another user. The method according to claim 1,
The e-book recommendation device
An electronic book recommendation device, further comprising an action information collection unit that collects the action information including at least one of the number of times of reading the e-book, the number of highlights, the content of the highlights, the number of screen captures, and the content of the screen capture. delete The method of claim 8,
The preference calculation unit
And calculating a value obtained by summing two or more of the number of readings, the number of highlights, and the number of screen captures as the preference. The method of claim 8,
The preference calculation unit
An electronic book recommendation device, characterized in that a weight is assigned to each of the number of reading, the number of highlights, and the number of screen captures, and a value obtained by summing at least two of the number of readings, the number of highlights, and the number of screen captures to which the weight is assigned is calculated as a preference. . delete delete delete delete delete delete delete

Images