WO2015129983A1 - Dispositif et procédé destinés à recommander un film en fonction de l'exploration distribuée de règles d'association imprécises - Google Patents

Dispositif et procédé destinés à recommander un film en fonction de l'exploration distribuée de règles d'association imprécises Download PDF

Info

Publication number
WO2015129983A1
WO2015129983A1 PCT/KR2014/010252 KR2014010252W WO2015129983A1 WO 2015129983 A1 WO2015129983 A1 WO 2015129983A1 KR 2014010252 W KR2014010252 W KR 2014010252W WO 2015129983 A1 WO2015129983 A1 WO 2015129983A1
Authority
WO
WIPO (PCT)
Prior art keywords
movie
fuzzy
rating
list
association
Prior art date
Application number
PCT/KR2014/010252
Other languages
English (en)
Korean (ko)
Inventor
김민성
Original Assignee
에스케이플래닛 주식회사
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 에스케이플래닛 주식회사 filed Critical 에스케이플래닛 주식회사
Publication of WO2015129983A1 publication Critical patent/WO2015129983A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G06N5/048—Fuzzy inferencing

Definitions

  • the present invention relates to a movie recommendation apparatus and method based on distributed fuzzy association rule mining for recommending a movie through linguistic information by converting movie rating information of a user into linguistic information.
  • the present invention relates to a movie recommendation apparatus and method based on distributed fuzzy association rule mining that generates a list of related movies obtained by applying to fuzzy association rule mining and recommends a movie to a user who is recommended using the generated list.
  • Movie recommendation through the general movie recommendation algorithm is based on the user's purchase history of the movie and is recommended in the form of 'movie watched by people who saw this movie' through association rule mining. Or, it is possible to recommend by extracting a 'movie similar to this movie' by calculating the similarity between movies.
  • the user's rating information for the movie is converted into linguistic information, and the user obtains an association with the movie rating information of the user through distributed fuzzy association rule mining based on the linguistic information, and uses the obtained association relationship.
  • a film recommendation technique based on distributed fuzzy association rule mining that can recommend a suitable film for the user.
  • An object of the present invention is to recommend a movie suitable for the user's preference tendency by replacing the rating information that the user leaves with respect to the movie by linguistic evaluation information and recommending a movie related to the substituted evaluation information.
  • an object of the present invention is to provide a more reliable movie recommendation function to users by efficiently processing a large amount of movie rating information using a data processing method suitable for a distributed framework.
  • Movie recommendation apparatus for achieving the above object, the data acquisition unit for obtaining the first rating data including the rating for the movie, converts the obtained first rating data into second rating data, An association list generator for generating a related movie list by applying the second rating data to Fuzzy Association Rule Mining and a movie recommendation unit for recommending a movie using the related movie list to the recommendation target user.
  • the association list generation unit assigns a rating to one or more of the fuzzy membership functions including the triangular membership function, the trapezoidal membership function, and the Gaussian membership function, thereby fuzzy membership.
  • a value may be obtained, and the language label according to the acquired fuzzy affiliation value may be replaced with a rating to be converted into second rating data.
  • the association list generator may generate one or more of fuzzy reliability and fuzzy correlation using fuzzy association rule mining, and generate an associated movie list based on at least one of the generated fuzzy reliability and fuzzy correlation.
  • the association list generation unit generates the association combination unit for generating the association combination for each movie by combining the converted second rating data according to the fuzzy association rule, and generates a rating history for each movie by arranging the second rating data for each movie, It may include a fuzzy support calculator for calculating the fuzzy support for each movie by using the generated rating history for each movie.
  • the fuzzy support calculator may calculate the fuzzy support for each movie using a reference value obtained by normalizing the fuzzy belonging value.
  • the association list generation unit combines at least two or more of the fuzzy membership functions to calculate the association combination fuzzy support for the association combination for each movie, and the fuzzy reliability using one or more of the per-movie fuzzy support and the calculated association combination fuzzy support. Can be calculated.
  • the association list generator may calculate the fuzzy correlation using one or more of square values of the fuzzy support for each movie, the fuzzy reliability, and the fuzzy support for each movie.
  • the movie recommendation unit may determine the ranking of the related movie list generated according to a predetermined importance level, and recommend the movies in the order of the determined related movie list having the highest ranking.
  • the list of related movies may include one or more of the title, genre, director, country, production year and image of the movie.
  • the movie recommendation method comprises the steps of: obtaining input data including a rating for a movie; converting the obtained input data into second rating data; and applying the converted second rating data to fuzzy association rule mining. Generating a related movie list and recommending a movie using the related movie list generated to the recommendation target user.
  • the step of generating an associative movie list assigns a rating to at least one of fuzzy membership functions including a triangular membership function, a trapezoidal membership function, and a Gaussian membership function.
  • the method may include obtaining a fuzzy belonging value and converting the language label according to the acquired fuzzy belonging value into a second rating data by substituting the rating for the language label.
  • generating the related movie list may include generating one or more of fuzzy reliability and fuzzy correlation using association rule mining and generating the related movie list based on at least one of the generated fuzzy reliability and fuzzy correlation. It may include the step.
  • the generating of the related movie list may include generating one or more of fuzzy reliability and fuzzy correlation using fuzzy association rule mining and generating the related movie list based on one of the generated fuzzy reliability and fuzzy correlation. It may include the step.
  • the generating of the associative movie list may include combining at least two or more of the fuzzy membership functions to calculate the associative combination fuzzy support for the associative combination by film, and at least one of the associative fuzzy support and the calculated associative combination fuzzy support Computing the fuzzy reliability using may include.
  • the generating of the related movie list may include calculating a fuzzy correlation using at least one of square values of fuzzy support for each movie, fuzzy reliability, and fuzzy support for each movie.
  • the recommending of the movie may include determining a ranking of the related movie list generated according to a predetermined importance level, and recommending the movies in the order of the determined related movie list having the highest ranking.
  • the rating of the user to recommend the movie corresponds to the preference of the user to be recommended I can recommend a movie to say.
  • a list of various recommended movies may be generated and provided based on various directions between linguistic information.
  • FIG. 1 is a block diagram showing a movie recommendation apparatus according to an embodiment of the present invention.
  • FIG. 2 is a block diagram illustrating an association list generator of the movie recommendation apparatus of FIG. 1.
  • FIG. 3 is a flowchart illustrating a movie recommendation method according to an embodiment of the present invention.
  • FIG. 4 is a flowchart illustrating a process of generating a related movie list according to an embodiment of the present invention.
  • FIG. 5 is a diagram illustrating an example of first rating data of users of a movie.
  • 6A to 6C illustrate a fuzzy membership function for generating second rating data according to the present invention.
  • FIG. 7 is a diagram illustrating first rating data illustrated in FIG. 5 as fuzzy membership values and second rating data using the fuzzy membership function of FIG. 6A.
  • FIG. 1 is a block diagram showing a movie recommendation apparatus according to an embodiment of the present invention.
  • the movie recommendation apparatus 100 may include a data acquirer 110, an association list generator 120, and a movie recommender 130.
  • the data acquirer 110 may obtain first rating data including a user's rating for the movie.
  • the first rating data may include information that can be obtained from the input log, that is, a user ID for identifying a user and a movie ID for distinguishing a movie, and may include a rating in the form of a number.
  • the user ID, the movie ID, and the rating may be extracted from another type of input log, the first rating data may be obtained.
  • the first rating data may be represented in the form of 'User1 (m1, r_1_1), (m3, r_1_3), ..., (m100, r_1_100)'.
  • User1 may display a user ID, m1, m3, and m100 are movie IDs, and r_1_1, r_1_3, and r_1_100 may be ratings, and a rating given by the nth user to the mth movie may be displayed in the form of r_n_m.
  • association list generating unit 120 converts the obtained first rating data into second rating data and generates the related movie list by applying the converted second rating data to Fuzzy Association Rule Mining. can do.
  • fuzzy association rule mining is a technique that applies fuzzy theory to association rule mining, and can be expressed by representing the degree to which each object belongs to the group as a function of membership from binary logic that each object belongs to or does not belong to a certain set. . Therefore, the association between the products consumed based on the user's log, such as association rule mining, can be calculated to calculate the association between the products registered in the market or store.
  • Such a fuzzy association rule is mainly calculated using a single machine, and since a large amount of recommendation requires distributed processing of such logic, in the present invention, the fuzzy association rule mining is more effectively based on a distributed framework. You can use Hadoop's MapReduce to calculate.
  • MapReduce can be solved by defining ⁇ key, value> in each step of mapper and reducer.
  • ⁇ key, value> is a data pair that is the basic unit of data processing, and key and value can be defined as any structure or class to process complex data.
  • a fuzzy membership value is obtained by assigning a rating to at least one of fuzzy membership functions including a triangular membership function, a trapezoidal membership function, and a Gaussian membership function.
  • the language label according to the acquired fuzzy affiliation value may be converted into second rating data by substituting with a rating.
  • the fuzzy membership value can be obtained with a value between 0 and 1 with respect to the language label, and the value of the language label with a large fuzzy membership value is obtained.
  • each fuzzy membership function corresponds to a language label such as 'no', 'normal', or 'good', and the range of the fuzzy membership function and the number of fuzzy membership functions may be separately specified.
  • association list generator 120 may generate at least one of fuzzy reliability and fuzzy correlation using fuzzy association rule mining, and generate a related movie list based on at least one of the generated fuzzy reliability and fuzzy correlation. Can be.
  • association list generating unit 120 may generate the association combination for each movie by combining the converted second rating data according to the fuzzy association rule, and generates a rating history for each movie by arranging the second rating data for each movie.
  • the fuzzy support for each movie may be calculated using the generated rating history for each movie.
  • the association association for each movie may generate the combination with a length of a preset association rule according to the fuzzy association rule. For example, if the association rule has a length of 2, create a movie combination for the movies m1 and m3 with 'm1, m3, user1, r_1_1, r_1_3', and gather the rating information of each user's movie for that movie combination. Data such as 'm1, m3, (user1, r_1_1, r_1_3), (uesr7, r_7_1, r_7_3), ..., (userN, r_N_1, r_N_3)' may be collected. At this time, an expression such as r_n_m may be interpreted as a rating given by the nth user to the mth movie.
  • the rating history for each movie is, for example, by collecting the second rating data of the form of 'User1 (m1, r_1_1), (m3, r_1_3), ..., (m100, r_1_100) for each movie' m1, (user1, r_1_1), (user1, r_2_1), ..., (userN, r_N_1) '.
  • the fuzzy support degree for each movie may be calculated using a reference value obtained by normalizing the fuzzy belonging value.
  • each fuzzy belonging value may be normalized using Equation 1 below.
  • t i [a j ] can represent the i th record value in transaction DB
  • fuzzy support for each movie may be calculated using the reference value obtained by normalizing the fuzzy membership degree.
  • fuzzy support can be calculated using Equation 2 below.
  • Is the membership value of the value a j for the lth membership function Is the normalized default value, t i [a j ] may represent the fuzzy support calculated by substituting the i-th record value and FS ⁇ A, X> in the transaction DB
  • At least two or more of the fuzzy membership functions may be combined to calculate the associated combination fuzzy support for the associated association for each movie, and the fuzzy reliability may be calculated using one or more of the per-movie fuzzy support and the calculated associated combination fuzzy support.
  • the associated combination fuzzy support may be represented in the form of 'm1, m2, MF_1, MF_2, FS (m1, MF1, m2, MF_2)', where m is a film, MF is a fuzzy membership function, and FS is a fuzzy support Can be represented.
  • fuzzy reliability may be calculated using Equation 3 below.
  • the fuzzy correlation may be calculated using one or more of squared values of the fuzzy support for each movie, the fuzzy reliability, and the fuzzy support for each movie.
  • fuzzy correlation may be calculated using Equation 4 below.
  • Is the membership value of the value a j for the lth membership function Is the normalized default value, t i [a j ] is the i th record value in transaction DB
  • the movie recommendation unit 130 may recommend a movie to the recommendation user by using the related movie list. For example, it may be recommended to a user in the order of movies more appropriate to the user by showing to the target audience users in the order of highly related movies among the movies included in the related movie list.
  • the ranking of the created related movie list may be determined according to a predetermined importance level, and the movies may be recommended in the order of the determined highest related movie list.
  • the relation between the movies is 'good-> good' or 'no-> good'. While it can be used to attract more users, the relationship of 'like-> dislike' and 'dislike-> dislike' may be suitable for the purpose of inducing curiosity rather than direct purchase.
  • the relation that is connected to 'normal' may be difficult to be connected to a function on the direct recommendation service, which may be unnecessary information in terms of recommendation service. Therefore, the relationship between 'good-> good' or 'no--good' is determined by high priority, and the relationship between 'good-> dislike', 'no-> dislike' and 'normal' is relatively Low priority may be determined.
  • the duplicate movies in the lower list may be deleted based on the ranking of the related movie list. For example, if there is a movie B in the list of movies that says 'user who likes movie A is good' and a movie list that says 'the user who likes movie A does not like', By checking the rankings, you can remove Movie B from the list of movies that you're saying, "I hate Movie A.”
  • the user may recommend a movie that is related to the user's preference by recommending a movie related to the user through the rating information left for the movie.
  • FIG. 2 is a block diagram illustrating an association list generator of the movie recommendation apparatus of FIG. 1.
  • the association list generator 120 of the movie recommendation apparatus of FIG. 1 includes an association combination generator 210 and a fuzzy support calculator 220.
  • the association combination generation unit 210 may generate the association combination for each movie by combining the converted second rating data according to the fuzzy association rule.
  • the association association for each movie may generate the combination with a length of a preset association rule according to the fuzzy association rule. For example, if the association rule has a length of 2, create a movie combination for the movies m1 and m3 with 'm1, m3, user1, r_1_1, r_1_3', and gather the rating information of each user's movie for that movie combination. Data such as 'm1, m3, (user1, r_1_1, r_1_3), (uesr7, r_7_1, r_7_3), ..., (userN, r_N_1, r_N_3)' may be collected. At this time, an expression such as r_n_m may be interpreted as a rating given by the nth user to the mth movie.
  • the fuzzy support calculator 220 may generate a rating history for each movie by arranging the second rating data for each movie, and calculate the fuzzy support for each movie using the generated rating history for each movie.
  • the rating history for each movie may be, for example, collecting second rating data in the form of 'User1 (m1, r_1_1), (m3, r_1_3), ..., (m100, r_1_100)' for each movie and then 'm1'. , (user1, r_1_1), (user1, r_2_1), ..., (userN, r_N_1) '.
  • the fuzzy support degree for each movie may be calculated using a reference value obtained by normalizing the fuzzy belonging value.
  • each fuzzy belonging value may be normalized using Equation 1 described above.
  • fuzzy support for each movie may be calculated using the reference value obtained by normalizing the fuzzy membership degree.
  • fuzzy support may be calculated using Equation 2 described above.
  • FIG. 3 is a flowchart illustrating a movie recommendation method according to an embodiment of the present invention.
  • the movie recommendation method may obtain first rating data including a user's rating for a movie (S310).
  • the first rating data may include information that can be obtained from the input log, that is, a user ID for identifying a user and a movie ID for distinguishing a movie, and may include a rating in the form of a number.
  • the user ID, the movie ID, and the rating may be extracted from another type of input log, the first rating data may be obtained.
  • the first rating data may be represented in the form of 'User1 (m1, r_1_1), (m3, r_1_3), ..., (m100, r_1_100)'.
  • User1 may display a user ID, m1, m3, and m100 are movie IDs, and r_1_1, r_1_3, and r_1_100 may be ratings, and a rating given by the nth user to the mth movie may be displayed in the form of r_n_m.
  • the movie recommendation device converts the obtained first rating data into second rating data and applies the converted second rating data to Fuzzy Association Rule Mining for association.
  • a list of movies may be generated (S320).
  • fuzzy association rule mining is a technique that applies fuzzy theory to association rule mining, and can be expressed by representing the degree to which each object belongs to the group as a function of membership from binary logic that each object belongs to or does not belong to a certain set. . Therefore, the association between the products consumed based on the user's log, such as association rule mining, can be calculated to calculate the association between the products registered in the market or store.
  • Such a fuzzy association rule is mainly calculated using a single machine, and since a large amount of recommendation requires distributed processing of such logic, in the present invention, the fuzzy association rule mining is more effectively based on a distributed framework. You can use Hadoop's MapReduce to calculate.
  • MapReduce can be solved by defining ⁇ key, value> in each step of mapper and reducer.
  • ⁇ key, value> is a data pair that is the basic unit of data processing, and key and value can be defined as any structure or class to process complex data.
  • a fuzzy membership value is obtained by assigning a rating to at least one of fuzzy membership functions including a triangular membership function, a trapezoidal membership function, and a Gaussian membership function.
  • the language label according to the acquired fuzzy affiliation value may be converted into second rating data by substituting with a rating.
  • the fuzzy membership value can be obtained with a value between 0 and 1 with respect to the language label, and the value of the language label with a large fuzzy membership value is obtained.
  • each fuzzy membership function corresponds to a language label such as 'no', 'normal', or 'good', and the range of the fuzzy membership function and the number of fuzzy membership functions may be separately specified.
  • association list generator 120 may generate at least one of fuzzy reliability and fuzzy correlation using fuzzy association rule mining, and generate a related movie list based on at least one of the generated fuzzy reliability and fuzzy correlation. Can be.
  • association list generating unit 120 may generate the association combination for each movie by combining the converted second rating data according to the fuzzy association rule, and generates a rating history for each movie by arranging the second rating data for each movie.
  • the fuzzy support for each movie may be calculated using the generated rating history for each movie.
  • the association association for each movie may generate the combination with a length of a preset association rule according to the fuzzy association rule. For example, if the association rule has a length of 2, create a movie combination for the movies m1 and m3 with 'm1, m3, user1, r_1_1, r_1_3', and gather the rating information of each user's movie for that movie combination. Data such as 'm1, m3, (user1, r_1_1, r_1_3), (uesr7, r_7_1, r_7_3), ..., (userN, r_N_1, r_N_3)' may be collected. At this time, an expression such as r_n_m may be interpreted as a rating given by the nth user to the mth movie.
  • the rating history for each movie is, for example, by collecting the second rating data of the form of 'User1 (m1, r_1_1), (m3, r_1_3), ..., (m100, r_1_100) for each movie' m1, (user1, r_1_1), (user1, r_2_1), ..., (userN, r_N_1) '.
  • the fuzzy support degree for each movie may be calculated using a reference value obtained by normalizing the fuzzy belonging value.
  • each fuzzy belonging value may be normalized using Equation 1 described above.
  • fuzzy support for each movie may be calculated using the reference value obtained by normalizing the fuzzy membership degree.
  • fuzzy support may be calculated using Equation 2 described above.
  • At least two or more of the fuzzy membership functions may be combined to calculate the associated combination fuzzy support for the associated association for each movie, and the fuzzy reliability may be calculated using one or more of the per-movie fuzzy support and the calculated associated combination fuzzy support.
  • the associated combination fuzzy support may be expressed in the form of 'm1, m2, MF_1, MF_2, FS (m1, MF1, m2, MF_2)', where m is a film, MF is a fuzzy membership function, and FS is a fuzzy support Can be represented.
  • fuzzy reliability may be calculated using Equation 3 described above.
  • the fuzzy correlation may be calculated using one or more of squared values of the fuzzy support for each movie, the fuzzy reliability, and the fuzzy support for each movie.
  • fuzzy correlation may be calculated using Equation 4 described above.
  • the movie recommendation method may recommend a movie to the recommendation target user by using the related movie list (S330). For example, it may be recommended to a user in the order of movies more appropriate to the user by showing to the target audience users in the order of highly related movies among the movies included in the related movie list.
  • the ranking of the created related movie list may be determined according to a predetermined importance level, and the movies may be recommended in the order of the determined highest related movie list.
  • the relation between the movies is 'good-> good' or 'no-> good'. While it can be used as an opportunity to attract more users, the relationship of 'like-> dislike' and 'dislike-> dislike' may be suitable for inducing curiosity rather than direct purchase.
  • the relation that is connected to 'normal' is difficult to be connected to a function on the direct recommendation service, and thus may be unnecessary information in terms of service. Therefore, the relationship between 'good-> good' or 'no--good' is determined by high importance ranking, and the relationship between 'good-> dislike', 'no-> dislike' and 'normal' is relatively low. It can be determined by importance ranking.
  • the duplicate movies in the lower list may be deleted based on the ranking of the related movie list. For example, if there is a movie B in the list of movies that says 'user who likes movie A is good' and a movie list that says 'the user who likes movie A does not like', By checking the rankings, you can remove Movie B from the list of movies that you're saying, "I hate Movie A.”
  • a reliable movie recommendation service using a large user log can be provided to users using the movie recommendation service.
  • FIG. 4 is a flowchart illustrating a process of generating a related movie list according to an embodiment of the present invention.
  • second rating data may be obtained by replacing a rating included in the first rating data with a language label (S410).
  • a fuzzy membership value is obtained by assigning a rating to at least one of fuzzy membership functions including a triangular membership function, a trapezoidal membership function, and a Gaussian membership function.
  • the language label according to the acquired fuzzy affiliation value may be converted into second rating data by substituting with a rating.
  • the related second movie combination may be generated by combining the converted second rating data according to the fuzzy association rule (S420).
  • the process of generating a related movie list generates a rating history for each movie by arranging the second rating data for each movie (S430), fuzzy for each movie using the generated rating history for each movie Support can be calculated (S440).
  • the process of generating a list of related movies is based on the fuzzy membership function by using the associated associations and the fuzzy support for each movie generated and calculated in steps S420 and S440, respectively.
  • the association combination fuzzy support for the association association for each movie may be calculated by combining at least two (S450).
  • the process of generating a list of related movies may calculate the fuzzy reliability by using one or more of the fuzzy support for each movie and the calculated associated combination fuzzy support (S460).
  • the fuzzy correlation may be calculated using at least one of square values of fuzzy support for each movie, fuzzy reliability, and fuzzy support for each movie (S470).
  • the process of generating a related movie list may generate a related movie list based on one of the generated fuzzy reliability and fuzzy correlation (S480).
  • FIG. 5 is a diagram illustrating an example of first rating data of users of a movie.
  • 6A to 6C illustrate a fuzzy membership function for generating second rating data according to the present invention.
  • FIG. 7 is a diagram illustrating first rating data illustrated in FIG. 5 as fuzzy membership values and second rating data using the fuzzy membership function of FIG. 6A.
  • FIGS. 5, 6A, 6C, and 7 when the user assigns a rating to a movie as shown in FIG. 5, the fuzzy membership diagram is illustrated in FIG. 7 using the fuzzy membership functions shown in FIGS. 6A through 6C. Value and second rating data may be generated.
  • the user 1 has given a score of 8 to the movie 4.
  • the fuzzy affiliation value for the movie 4 of the user 1 is shown in FIG. 7 as hate: 0.0, normal: 0.33, and good: 0.67. Therefore, in this case, the language label 'good' may be converted into second rating data by substituting eight ratings of the first rating data.
  • the movie recommendation method according to the present invention may be implemented in the form of program instructions that can be executed by various computer means and recorded on a computer readable medium.
  • the computer readable medium may include program instructions, data files, data structures, etc. alone or in combination.
  • Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts.
  • Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks.
  • Magneto-optical media and any type of hardware device specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like.
  • Examples of program instructions may include high-level language code that can be executed by a computer using an interpreter as well as machine code such as produced by a compiler.
  • Such hardware devices may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.
  • the apparatus and method for recommending a movie based on distributed fuzzy association rule mining according to the present invention is not limited to the configuration and method of the embodiments described as described above, and the embodiments may be modified in various ways. All or part of each of the embodiments may be configured to be selectively combined to make it possible.
  • the first rating data including the rating for the movie is converted into second rating data
  • the converted second rating data is applied to fuzzy association rule mining to generate a related movie list
  • the generated related movie list By recommending a movie through the user, it is possible to effectively recommend a movie suitable for the preference of the target user.
  • the data processing method suitable for a distributed framework is used to apply such a recommendation function to a large amount of users, it can be applied to a large-scale rating log data, thereby providing a more reliable service.

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Fuzzy Systems (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Automation & Control Theory (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

L'invention concerne un dispositif et un procédé destinés à recommander un film en fonction de l'exploration distribuée de règles d'association imprécises. Des premières données de classification comprenant une classification de film sont acquises, les premières données de classification acquises sont converties en des secondes données de classification à l'aide d'une fonction d'appartenance imprécise et une liste de films associés est générée par application des secondes données de classification converties à l'exploration de règles d'association imprécises. Un film est recommandé dans l'ordre décroissant des classifications à l'aide de la liste de films associés qui a été générée, permettant ainsi de recommander un film qui est associé à un utilisateur à qui un film doit être recommandé.
PCT/KR2014/010252 2014-02-26 2014-10-29 Dispositif et procédé destinés à recommander un film en fonction de l'exploration distribuée de règles d'association imprécises WO2015129983A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR10-2014-0022934 2014-02-26
KR1020140022934A KR102167593B1 (ko) 2014-02-26 2014-02-26 분산 퍼지 연관 규칙 마이닝에 기반한 영화 추천 장치 및 방법

Publications (1)

Publication Number Publication Date
WO2015129983A1 true WO2015129983A1 (fr) 2015-09-03

Family

ID=54009269

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2014/010252 WO2015129983A1 (fr) 2014-02-26 2014-10-29 Dispositif et procédé destinés à recommander un film en fonction de l'exploration distribuée de règles d'association imprécises

Country Status (2)

Country Link
KR (1) KR102167593B1 (fr)
WO (1) WO2015129983A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107943953A (zh) * 2017-11-24 2018-04-20 福建中金在线信息科技有限公司 榜单推荐方法、装置、电子设备和计算机可读存储介质

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101708254B1 (ko) 2015-09-25 2017-02-21 인하대학교 산학협력단 협업적 필터링과 캐릭터 넷을 이용한 스토리 기반의 영화 추전 시스템 및 방법.
KR20190030194A (ko) 2018-12-31 2019-03-21 씨제이씨지브이 주식회사 상영관 추천에 기반하여 영화와 상영관 예매를 제공하기 위한 방법 및 서버
KR102276353B1 (ko) * 2020-01-22 2021-07-12 공주대학교 산학협력단 퍼지 계층 분석 과정을 이용한 워드투벡터 학습 기반 영화 추천 시스템 및 방법
CN111859117B (zh) * 2020-06-19 2024-03-26 腾讯科技（深圳）有限公司 信息推荐方法、装置、电子设备及可读存储介质
CN111830501B (zh) * 2020-06-28 2023-04-28 中国人民解放军战略支援部队信息工程大学 Hrrp历史特征辅助的信号模糊数据关联方法及系统
KR102433804B1 (ko) * 2021-01-25 2022-08-18 공주대학교 산학협력단 영화 추천 장치
KR20240040527A (ko) 2022-09-21 2024-03-28 서울과학기술대학교 산학협력단 영화 리뷰 작성자의 신뢰성 검증 장치 및 방법

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090014259A (ko) * 2006-02-16 2009-02-09 모바일 컨텐트 네트웍스, 인크. 관련있는 소스를 결정하고, 질의를 하며 다수의 컨텐츠 소스로부터의 결과를 병합하는 방법 및 시스템
KR20090080857A (ko) * 2008-01-22 2009-07-27 에스케이텔레콤 주식회사 소셜 네트워크 기반의 콘텐츠 추천 시스템과 방법 및콘텐츠 제공서버
KR20090124629A (ko) * 2008-05-30 2009-12-03 성균관대학교산학협력단 컨텐츠의 최대 재생 시간과 사용자의 시청시간 혹은청취시간을 고려하는 관심 정보 추천 시스템 및 그 방법
KR20120014458A (ko) * 2010-08-09 2012-02-17 충북대학교 산학협력단 연관 규칙 마이닝을 이용한 주제어 기반 인터넷 정보 검색 방법

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101565339B1 (ko) * 2010-11-03 2015-11-04 네이버 주식회사 집단지성을 이용한 추천 시스템 및 방법
KR20130009360A (ko) 2011-07-15 2013-01-23 박용구 영화추천 서비스 제공방법 및 그 시스템

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090014259A (ko) * 2006-02-16 2009-02-09 모바일 컨텐트 네트웍스, 인크. 관련있는 소스를 결정하고, 질의를 하며 다수의 컨텐츠 소스로부터의 결과를 병합하는 방법 및 시스템
KR20090080857A (ko) * 2008-01-22 2009-07-27 에스케이텔레콤 주식회사 소셜 네트워크 기반의 콘텐츠 추천 시스템과 방법 및콘텐츠 제공서버
KR20090124629A (ko) * 2008-05-30 2009-12-03 성균관대학교산학협력단 컨텐츠의 최대 재생 시간과 사용자의 시청시간 혹은청취시간을 고려하는 관심 정보 추천 시스템 및 그 방법
US20090299943A1 (en) * 2008-05-30 2009-12-03 Sungkyunkwan University Foundation For Corporate Collaboration Concerned information recommendation system and method considering user's watching or listening time and maximum playing time of contents
KR20120014458A (ko) * 2010-08-09 2012-02-17 충북대학교 산학협력단 연관 규칙 마이닝을 이용한 주제어 기반 인터넷 정보 검색 방법

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107943953A (zh) * 2017-11-24 2018-04-20 福建中金在线信息科技有限公司 榜单推荐方法、装置、电子设备和计算机可读存储介质

Also Published As

Publication number Publication date
KR20150101341A (ko) 2015-09-03
KR102167593B1 (ko) 2020-10-19

Similar Documents

Publication Publication Date Title
WO2015129983A1 (fr) Dispositif et procédé destinés à recommander un film en fonction de l'exploration distribuée de règles d'association imprécises
WO2016208916A1 (fr) Procédé et système de gestion d'onglets basée sur le contexte
WO2010117213A2 (fr) Appareil et procédé destinés à fournir des informations en lien avec des programmes de radiodiffusion
WO2014069755A1 (fr) Système et procédé permettant de fournir un service de recommandation de contenu
WO2010119996A1 (fr) Procédé et dispositif de fourniture d'annonces publicitaires à images mobiles
WO2018207975A1 (fr) Système de chaîne de blocs et procédé permettant de générer une chaîne de blocs
WO2015020354A1 (fr) Appareil, serveur et procédé pour fournir un sujet de conversation
WO2011162446A1 (fr) Module et procédé permettant de décider une entité nommée d'un terme à l'aide d'un dictionnaire d'entités nommées combiné avec un schéma d'ontologie et une règle d'exploration
WO2020055071A1 (fr) Système et procédé de classification dynamique de tendances
WO2016093630A1 (fr) Enrichissement sémantique de données de trajectoire
WO2015133856A1 (fr) Procédé et dispositif pour fournir un mot-clé de réponse correcte
WO2016085105A1 (fr) Procédé et dispositif de blocage de publicité
WO2024091080A1 (fr) Procédé de génération de vidéo automatique, et serveur de génération de vidéo automatique
WO2020130418A1 (fr) Procédé d'analyse de risque de chaîne logistique de fournisseurs
WO2017164510A2 (fr) Procédé de marquage de contenu multimédia basé sur des données vocales, et système l'utilisant
WO2022050632A1 (fr) Système de génération automatique multimédia pour générer automatiquement du multimédia approprié pour des données vocales d'utilisateur en utilisant une intelligence artificielle
WO2015080371A1 (fr) Système et procédé de recherche d'image
WO2016129804A1 (fr) Procédé pour générer une page web sur la base de modèles de comportement de consommateurs et procédé d'utilisation de page web
WO2020242089A2 (fr) Procédé de conservation basé sur l'intelligence artificielle et dispositif pour la mise en oeuvre de ce procédé
WO2016186326A1 (fr) Dispositif de fourniture de liste de mots de recherche et procédé associé
WO2016190650A1 (fr) Système pour réaliser un captcha et une publicité utilisant une combinaison de caractères
WO2016195421A1 (fr) Procédé, système et support d'enregistrement lisible par ordinateur non-transitoire pour fournir un profilage de données sur la base d'analyse statistique
WO2012030049A2 (fr) Appareil et procédé de classification de documents similaires par application de valeur seuil dynamique
WO2013089390A1 (fr) Système de fourniture d'informations personnelles sur la base de la création et de la consommation d'un contenu
WO2017222226A1 (fr) Procédé d'enregistrement d'un produit publicitaire sur un contenu d'image, et serveur pour l'exécution du procédé

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14883788

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14883788

Country of ref document: EP

Kind code of ref document: A1