CN116662659B - Media content intelligent recommendation system based on artificial intelligence - Google Patents

Media content intelligent recommendation system based on artificial intelligence Download PDF

Info

Publication number
CN116662659B
CN116662659B CN202310633278.1A CN202310633278A CN116662659B CN 116662659 B CN116662659 B CN 116662659B CN 202310633278 A CN202310633278 A CN 202310633278A CN 116662659 B CN116662659 B CN 116662659B
Authority
CN
China
Prior art keywords
works
recommendation
work
target
updated
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN202310633278.1A
Other languages
Chinese (zh)
Other versions
CN116662659A (en
Inventor
孙金杰
林媚媚
王付琳
孙研冰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangzhou Puff Media Co ltd
Original Assignee
Guangzhou Puff Media Co ltd
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 Guangzhou Puff Media Co ltd filed Critical Guangzhou Puff Media Co ltd
Priority to CN202310633278.1A priority Critical patent/CN116662659B/en
Publication of CN116662659A publication Critical patent/CN116662659A/en
Application granted granted Critical
Publication of CN116662659B publication Critical patent/CN116662659B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/73—Querying
    • G06F16/735—Filtering based on additional data, e.g. user or group profiles
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Multimedia (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

The invention discloses an artificial intelligence based media content intelligent recommendation system, which comprises a data acquisition unit, a preprocessing unit, a recommendation analysis unit and a search recommendation unit. According to the media content intelligent recommendation system based on artificial intelligence, basic information of target users is obtained, the basic information is preprocessed to obtain preference coefficients of the target users, recommendation coefficients of popular works and updated works are calculated, and the generated recommendation coefficients are changed according to different favorites of the users in different periods, so that accuracy of content recommendation can be improved, a good recommendation effect is obtained, all target works are searched out, similarity calculation is conducted, and then recommendation coefficients of all target works are calculated by combining a recommendation analysis unit, so that the users can timely and effectively obtain the wanted works, and experience of the users is affected.

Description

Media content intelligent recommendation system based on artificial intelligence
Technical Field
The invention relates to the technical field of internet data processing, in particular to an artificial intelligence-based media content intelligent recommendation system.
Background
The current short video application program, which is a new video viewing platform, has a large number of short videos and authors, and how to recommend short videos of interest to users from a large number of short videos, has become a technical problem of important attention of technicians.
The related work recommendation system in the current numerous media short videos is used for automatically recommending the user according to the daily browsing behaviors of the user, but when the user searches interested works, the media platform can recommend only works with compact relevance to the client in priority, but because the short videos and authors are more, different authors can send similar short videos of the works to display, so that the recommended works have high repetition rate and related content browsed before repeated pushing cannot be found, other programs in the field cannot be effectively displayed, and the experience of the user is affected.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an intelligent media content recommendation system based on artificial intelligence, which solves the problems in the background technology.
In order to achieve the above purpose, the invention is realized by the following technical scheme: the system comprises a data acquisition unit, a preprocessing unit and a storage unit, wherein the data acquisition unit is used for acquiring basic information of a target user and transmitting the basic information to the preprocessing unit, the basic information comprises a concerned account number and account number type of the target user, collected works, praise works and work types thereof, the account number type is determined according to the work types in all works sent out in a designated account number, and the work types are determined by work labels set by a media platform;
A preprocessing unit for preprocessing the basic information to obtain preference coefficients of the target user, transmitting the preference coefficients to a recommendation analysis unit,
The recommendation analysis unit is used for acquiring the popular works recommended by all the media platforms and the updated works of the attention account numbers according to the preference coefficients, calculating the recommendation coefficients of the popular works and the updated works, sequencing the recommendation coefficients according to the sequence from large to small, and preferentially pushing the popular works and the updated works with the front recommendation coefficients to the user.
Preferably, the specific determination mode of the account number type is as follows: acquiring the number of all works in a designated account, calculating the occupation ratio of the number of the same work type in the number of all works, sequencing the occupation ratio according to the sequence from large to small, and determining the work type with the front occupation ratio as the account type;
Preferably, the specific pretreatment mode of the pretreatment unit is as follows:
Firstly, acquiring the quantity of all concerned accounts, and recording the quantity as concerned quantity;
Acquiring account types of all concerned accounts, and then acquiring the quantity of concerned accounts of the same account type, and recording the quantity as similar concerned quantity;
calculating the ratio of the attention quantity and the attention quantity of each same class, and recording the ratio as the total attention ratio;
Secondly, continuously obtaining the quantity of all collected works and praise works, and recording the quantity as the collection quantity and the praise quantity;
respectively acquiring the types of works of all collected works and praise works, and then acquiring the quantity of all collected works of the same type of works and the quantity of praise works of the same type of works, and recording the quantity as similar collection and similar praise;
Calculating the ratio of the collection amount and the collection amount of the same kind, recording the ratio as the total collection ratio, and calculating the ratio of the praise amount and the praise amount of the same kind, recording the ratio as the total praise ratio;
Thirdly, acquiring the number of newly added attention account numbers, collected works and praise works of a user in a previous period t 1;
Then, the account number type of the newly added attention account number, the newly added collection work and the work types of the praise works in the previous period are obtained, and the quantity of the newly added attention account number of the same account number type, the quantity of the newly added collection work of the same work type and the quantity of the newly added praise works of the same work type are obtained;
Then obtaining a new attention rate I, a new collection rate I and a new praise rate I according to the first step and the second step;
Fourthly, obtaining a designated number of newly added attention account numbers, collected works and praise works of a user in a proximity period;
Then, the account types of the newly added attention account numbers, the newly added collection works and the work types of the praise works are obtained, and the quantity of the newly added attention account numbers of the same account types, the quantity of the newly added collection works of the same work types and the quantity of the newly added praise works of the same work types are obtained;
then obtaining a new attention rate II, a new collection rate II and a new praise rate II according to the first step and the second step;
fifthly, firstly selecting a designated work type, acquiring a related total collection ratio, a related total praise ratio, a related new added collection ratio I, a related new added praise ratio I, a related new added collection ratio II and a related new added praise ratio II, and marking the two as ZS, ZD and XS 1、XD1、XS2、XD2 respectively;
Then selecting an account number type containing a designated work type, acquiring a related total attention ratio, a new attention ratio I and a new attention ratio II, and marking the account number type as ZG and XG 1、XG2 respectively;
then by the formula:
Obtaining a preference coefficient Y of the appointed work, wherein beta 1、β2、β3 is a preset proportionality coefficient;
And sixthly, acquiring preference coefficients of other work types in a fifth step, and sequencing the preference coefficients in a descending order.
Preferably, the specific analysis mode of the recommended analysis unit is as follows:
S1, acquiring updated works of popular works and attention accounts recommended by all media platforms, and simultaneously acquiring work access amounts, work types and browsing behaviors of the popular works and the updated works, wherein the browsing behaviors are divided into non-browsing behaviors which are not checked on the popular works and the updated works and browsed behaviors which are checked on the popular works and the updated works;
S2, recommending and assigning all the hot works and updated works of different work types, wherein the recommended assignments of the hot works and the updated works of the corresponding work types are sequentially 0.1, 0.2, 0.3 according to the arrangement sequence of preference coefficients, and the recommended assignment marks of the hot works and the updated works are TF i, i=1, 2, 3 and represent the quantity of all the hot works and the updated works, and the larger the preference coefficient is, the larger the recommended assignments of the hot works and the updated works of the corresponding work types are;
S3, substituting the hot works and all the works of the updated works into a value assignment interval of corresponding interval values according to the access quantity of the hot works and the updated works, taking specific values of the value assignment interval as the access values of the hot works and the updated works, and marking the access values of the hot works and the updated works as FF i, wherein the interval values in the value assignment interval and the specific values thereof are preset values, the interval values in the value assignment interval represent the set interval of the access quantity, the specific values are sequentially 0.11, 0.12, 0.13 and the number of the access values according to the size of the divided interval values, and the larger the interval values in the value assignment interval are, namely the larger the access values of the hot works and the updated works corresponding to the types of the works are;
S4, performing browsing assignment on all popular works and updated works according to browsing behaviors of users, wherein the browsing assignment comprises the following specific steps: the unbrown behavior is assigned to "1", the browsed behavior is assigned to "0", and the browsed assignment of all popular works and updated works is marked as LF i;
S5, obtaining recommendation coefficients of all popular works and updated works through a formula TJ i＝(TFi*α1+FFi*α2)*LFi, wherein alpha 1、α2 is a preset proportionality coefficient;
S6, sequencing the recommendation coefficients of all the popular works and updated works according to the sequence from large to small, and preferentially pushing the popular works and updated works with the front recommendation coefficients to a user;
Preferably, the method further comprises:
The searching recommendation unit is used for inputting target titles in a search box of the media platform by a user, searching to obtain all target works, calculating according to similarity of all target works, calculating recommendation coefficients of all target works by combining the recommendation analysis unit, sequencing the recommendation coefficients according to the sequence from big to small, and pushing target works with the recommendation coefficients ahead to the user in priority.
Preferably, the specific recommendation mode of the search recommendation unit is as follows:
x1, acquiring all target works, and optionally selecting a group of target works as reference works;
Then, in the reference work, intercepting video frames in the reference work at intervals of standard time t to obtain a plurality of video frames Z1 j, j=1, 2,3, and the number of acquired video frames I is represented;
X2, obtaining a plurality of groups of video frames II from the rest target works, and respectively carrying out similarity calculation on the video frames II and Z1 1 to obtain comparison similarity values X 1 of the video frames and Z1 1, wherein the similarity calculation mode is that the similarity is calculated through a histogram algorithm and a hash algorithm;
the similarity value is then compared with a preset contrast value X 0:
When the video frame two includes X 1≥X0 of the video frame two, after the video frame two, acquiring a plurality of video frames three Z3 j+1 at intervals of a standard time t, and performing similarity calculation on the video frames Z1 2、Z13、Z14 and Z3 2、Z33、Z34 and the video frames X 2、X3、X4 and X 2、X3、X4 and X 0 respectively:
If u continuous comparison similarity values are all larger than a preset comparison value, acquiring corresponding target works, and using the corresponding target works as comparison works, wherein the comparison works are represented as other target works with similarity with the reference works higher than a set value;
If the u continuous comparison similarity values are not greater than the preset comparison value, acquiring corresponding target works, and using the corresponding target works as residual reference works, wherein Yu Can works are represented as other target works with similarity with the reference works being lower than a set value;
When the X 1≥X0 of the video frame II is not included in the video frames II, a corresponding target work is obtained and is used as a rest parameter work;
X4, processing the residual ginseng works again in the mode of the steps X1 and X2 to obtain comparison works corresponding to the residual ginseng works, other residual ginseng works and the comparison works corresponding to the other residual ginseng works;
X5, firstly calculating the recommendation coefficients of the reference works and all the comparison works through a recommendation analysis unit, then selecting a group of target works with the maximum recommendation coefficients as first selection works, simultaneously calculating the recommendation coefficients of the rest reference works and the corresponding comparison works, and selecting a group of target works with the maximum recommendation coefficients as second selection works, third selection works;
calculating the recommendation coefficients of the first referral work, the second referral work, the third referral work and the third referral work by a recommendation analysis unit;
X6, sorting the first, second and third referral works according to the recommendation coefficient, and pushing the target works with the front recommendation coefficient to the user in a priority mode;
And then, acquiring recommendation coefficients of the residual target works, sequencing the recommendation coefficients according to the sequence from large to small, sequentially pushing the residual target works to a user according to the sequence, and pushing the residual target works to the user after the residual target works are placed in the first selecting work, the second selecting work, the third selecting work and the third selecting work.
Advantageous effects
The invention provides an artificial intelligence based media content intelligent recommendation system. Compared with the prior art, the method has the following beneficial effects:
According to the method, basic information of the target user is obtained, the basic information is preprocessed, preference coefficients of the target user are obtained, then updated works of popular works and concerned accounts recommended by all media platforms are obtained according to the preference coefficients, recommendation coefficients of the popular works and the updated works are calculated according to all data of the target user in the early stage and all data of the recent part, and the generated recommendation coefficients are changed according to different favorites of the user on the works in different periods, so that accuracy of content recommendation can be improved, and good recommendation effects are obtained;
According to the invention, through searching out all target works and calculating the similarity, then calculating the recommendation coefficients of all target works by combining the recommendation analysis unit, and recommending target works with low content similarity preferentially, a user can timely and effectively obtain the wanted works, and the situation that other works cannot be effectively displayed due to too high repeatability of the searched works and too many repeated works are arranged before the user can be avoided, so that the experience of the user is influenced.
Drawings
Fig. 1 is a system block diagram of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, the present invention provides the following technical solutions;
as an embodiment of the present invention, an artificial intelligence based media content intelligent recommendation system includes:
The data acquisition unit is used for acquiring basic information of a target user and transmitting the basic information to the preprocessing unit, wherein the basic information comprises a concerned account number and account number type of the target user, collected works, praise works and work types thereof, the account number type is determined according to the work types in all works sent out in a designated account number, and the work types are determined by work labels set by a media platform;
The determination mode of the account number type is as follows: acquiring the number of all works in a designated account, calculating the occupation ratio of the number of the same work type in the number of all works, sequencing the occupation ratio according to the sequence from large to small, and determining the work type with the front occupation ratio as the account type;
The preprocessing unit is used for preprocessing the basic information to obtain preference coefficients of the target user, and transmitting the preference coefficients to the recommendation analysis unit, and the specific mode is as follows:
Firstly, acquiring the quantity of all concerned accounts, and recording the quantity as concerned quantity;
Acquiring account types of all concerned accounts, and then acquiring the quantity of concerned accounts of the same account type, and recording the quantity as similar concerned quantity;
calculating the ratio of the attention quantity and the attention quantity of each same class, and recording the ratio as the total attention ratio;
Secondly, continuously obtaining the quantity of all collected works and praise works, and recording the quantity as the collection quantity and the praise quantity;
respectively acquiring the types of works of all collected works and praise works, and then acquiring the quantity of all collected works of the same type of works and the quantity of praise works of the same type of works, and recording the quantity as similar collection and similar praise;
Calculating the ratio of the collection amount and the collection amount of the same kind, recording the ratio as the total collection ratio, and calculating the ratio of the praise amount and the praise amount of the same kind, recording the ratio as the total praise ratio;
Thirdly, acquiring the number of newly added attention account numbers, collected works and praise works of a user in a previous period t 1;
Then, the account number type of the newly added attention account number, the newly added collection work and the work types of the praise works in the previous period are obtained, and the quantity of the newly added attention account number of the same account number type, the quantity of the newly added collection work of the same work type and the quantity of the newly added praise works of the same work type are obtained;
Then obtaining a new attention rate I, a new collection rate I and a new praise rate I according to the first step and the second step;
Fourthly, obtaining a designated number of newly added attention account numbers, collected works and praise works of a user in a proximity period;
Then, the account types of the newly added attention account numbers, the newly added collection works and the work types of the praise works are obtained, and the quantity of the newly added attention account numbers of the same account types, the quantity of the newly added collection works of the same work types and the quantity of the newly added praise works of the same work types are obtained;
then obtaining a new attention rate II, a new collection rate II and a new praise rate II according to the first step and the second step;
fifthly, firstly selecting a designated work type, acquiring a related total collection ratio, a related total praise ratio, a related new added collection ratio I, a related new added praise ratio I, a related new added collection ratio II and a related new added praise ratio II, and marking the two as ZS, ZD and XS 1、XD1、XS2、XD2 respectively;
Then selecting an account number type containing a designated work type, acquiring a related total attention ratio, a new attention ratio I and a new attention ratio II, and marking the account number type as ZG and XG 1、XG2 respectively;
then by the formula:
Obtaining a preference coefficient Y of the appointed work, wherein beta 1、β2、β3 is a preset proportionality coefficient;
Sixthly, obtaining preference coefficients of other work types according to the mode of the fifth step, and sequencing the preference coefficients in a sequence from small to large;
The recommendation analysis unit is used for acquiring the popular works recommended by all media platforms and updated works of the attention account numbers according to the preference coefficients, calculating the recommendation coefficients of the popular works and the updated works, sequencing the recommendation coefficients according to the sequence from large to small, and preferentially pushing the popular works and the updated works with the front recommendation coefficients to users, wherein the recommendation analysis unit comprises the following specific modes:
S1, acquiring updated works of popular works and attention accounts recommended by all media platforms, and simultaneously acquiring work access amounts, work types and browsing behaviors of the popular works and the updated works, wherein the browsing behaviors are divided into non-browsing behaviors which are not checked on the popular works and the updated works and browsed behaviors which are checked on the popular works and the updated works;
S2, recommending and assigning all the hot works and updated works of different work types, wherein the recommended assignments of the hot works and the updated works of the corresponding work types are sequentially 0.1, 0.2, 0.3 according to the arrangement sequence of preference coefficients, and the recommended assignment marks of the hot works and the updated works are TF i, i=1, 2, 3 and represent the quantity of all the hot works and the updated works, and the larger the preference coefficient is, the larger the recommended assignments of the hot works and the updated works of the corresponding work types are;
S3, substituting the hot works and all the works of the updated works into a value assignment interval of corresponding interval values according to the access quantity of the hot works and the updated works, taking specific values of the value assignment interval as the access values of the hot works and the updated works, and marking the access values of the hot works and the updated works as FF i, wherein the interval values in the value assignment interval and the specific values thereof are preset values, the interval values in the value assignment interval represent the set interval of the access quantity, the specific values are sequentially 0.11, 0.12, 0.13 and the number of the access values according to the size of the divided interval values, and the larger the interval values in the value assignment interval are, namely the larger the access values of the hot works and the updated works corresponding to the types of the works are;
S4, performing browsing assignment on all popular works and updated works according to browsing behaviors of users, wherein the browsing assignment comprises the following specific steps: the unbrown behavior is assigned to "1", the browsed behavior is assigned to "0", and the browsed assignment of all popular works and updated works is marked as LF i;
S5, obtaining recommendation coefficients of all popular works and updated works through a formula TJ i＝(TF i*α1+FFi*α2)*LFi, wherein alpha 1、α2 is a preset proportionality coefficient;
S6, sequencing the recommendation coefficients of all the popular works and updated works according to the sequence from large to small, and preferentially pushing the popular works and updated works with the front recommendation coefficients to a user;
basic information of a target user is obtained, the basic information is preprocessed to obtain preference coefficients of the target user, then updated works of popular works and attention accounts recommended by all media platforms are obtained according to the preference coefficients, recommendation coefficients of the popular works and the updated works are calculated according to all data of the target user in the early stage and all data of the recent part, and the generated recommendation coefficients are changed according to different favorites of the user on the works in different periods, so that accuracy of content recommendation can be improved, and a good recommendation effect is obtained;
As a second embodiment of the present invention, this embodiment differs from the first embodiment in that this embodiment further includes, on the basis of the first embodiment:
The searching recommendation unit is used for inputting target titles in a search box of the media platform by a user, searching to obtain all target works, calculating according to similarity of all target works, calculating recommendation coefficients of all target works by combining the recommendation analysis unit, sequencing the recommendation coefficients according to the sequence from big to small, and preferentially pushing target works with the front recommendation coefficients to the user, wherein the specific mode is as follows:
x1, acquiring all target works, and optionally selecting a group of target works as reference works;
Then, in the reference work, intercepting video frames in the reference work at intervals of standard time t to obtain a plurality of video frames Z1 j, j=1, 2,3, and the number of acquired video frames I is represented;
X2, obtaining a plurality of groups of video frames II from the rest target works, and respectively carrying out similarity calculation on the video frames II and Z1 1 to obtain comparison similarity values X 1 of the video frames and Z1 1, wherein the similarity calculation mode is that the similarity is calculated through a histogram algorithm and a hash algorithm;
the similarity value is then compared with a preset contrast value X 0:
When the video frame two includes X 1≥X0 of the video frame two, after the video frame two, acquiring a plurality of video frames three Z3 j+1 at intervals of a standard time t, and performing similarity calculation on the video frames Z1 2、Z13、Z14 and Z3 2、Z33、Z34 and the video frames X 2、X3、X4 and X 2、X3、X4 and X 0 respectively:
If u continuous comparison similarity values are all larger than a preset comparison value, acquiring corresponding target works, and using the corresponding target works as comparison works, wherein the comparison works are represented as other target works with similarity with the reference works higher than a set value;
If the u continuous comparison similarity values are not greater than the preset comparison value, acquiring corresponding target works, and using the corresponding target works as residual reference works, wherein Yu Can works are represented as other target works with similarity with the reference works being lower than a set value;
When the X 1≥X0 of the video frame II is not included in the video frames II, a corresponding target work is obtained and is used as a rest parameter work;
X4, processing the residual ginseng works again in the mode of the steps X1 and X2 to obtain comparison works corresponding to the residual ginseng works, other residual ginseng works and the comparison works corresponding to the other residual ginseng works;
X5, firstly calculating the recommendation coefficients of the reference works and all the comparison works through a recommendation analysis unit, then selecting a group of target works with the maximum recommendation coefficients as first selection works, simultaneously calculating the recommendation coefficients of the rest reference works and the corresponding comparison works, and selecting a group of target works with the maximum recommendation coefficients as second selection works, third selection works;
calculating the recommendation coefficients of the first referral work, the second referral work, the third referral work and the third referral work by a recommendation analysis unit;
X6, sorting the first, second and third referral works according to the recommendation coefficient, and pushing the target works with the front recommendation coefficient to the user in a priority mode;
Then, acquiring recommendation coefficients of the residual target works, sequencing the recommendation coefficients according to the sequence from large to small, sequentially pushing the residual target works to a user according to the sequence, and pushing the residual target works to the user after the residual target works are placed in the first selecting work, the second selecting work, the third selecting work and the third selecting work;
By searching out all target works and calculating the similarity, then combining a recommendation analysis unit to calculate out recommendation coefficients of all target works, and recommending target works with low content similarity preferentially, a user can timely and effectively obtain wanted works, and the situation that other works cannot be effectively displayed due to too high repeatability of the searched works and too many repeated works are arranged is avoided, so that the user cannot obtain the works wanted to be searched, and further experience of the user is affected;
as an embodiment three of the present invention, this embodiment will implement a fusion implementation of the first and second embodiments.
And all that is not described in detail in this specification is well known to those skilled in the art.
The foregoing describes one embodiment of the present invention in detail, but the disclosure is only a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. All equivalent changes and modifications within the scope of the present invention are intended to be covered by the present invention.

Claims (2)

1. An artificial intelligence based media content intelligent recommendation system, comprising:
The data acquisition unit is used for acquiring basic information of a target user and transmitting the basic information to the preprocessing unit, wherein the basic information comprises a concerned account number and account number type of the target user, collected works, praise works and work types thereof, the account number type is determined according to the work types in all works sent out in a designated account number, and the work types are determined by work labels set by a media platform;
The specific determination mode of the account number type is as follows: acquiring the number of all works in a designated account, calculating the occupation ratio of the number of the same work type in the number of all works, sequencing the occupation ratio according to the sequence from large to small, and determining the work type with the front occupation ratio as the account type;
the preprocessing unit is used for preprocessing the basic information to obtain preference coefficients of the target user, and transmitting the preference coefficients to the recommendation analysis unit;
the specific pretreatment mode of the pretreatment unit is as follows:
Firstly, acquiring the quantity of all concerned accounts, and recording the quantity as concerned quantity;
Acquiring account types of all concerned accounts, and then acquiring the quantity of concerned accounts of the same account type, and recording the quantity as similar concerned quantity;
calculating the ratio of the attention quantity and the attention quantity of each same class, and recording the ratio as the total attention ratio;
Secondly, continuously obtaining the quantity of all collected works and praise works, and recording the quantity as the collection quantity and the praise quantity;
respectively acquiring the types of works of all collected works and praise works, and then acquiring the quantity of all collected works of the same type of works and the quantity of praise works of the same type of works, and recording the quantity as similar collection and similar praise;
Calculating the ratio of the collection amount and the collection amount of the same kind, recording the ratio as the total collection ratio, and calculating the ratio of the praise amount and the praise amount of the same kind, recording the ratio as the total praise ratio;
Thirdly, acquiring the number of newly added attention account numbers, collected works and praise works of a user in a previous period t 1;
Then, the account number type of the newly added attention account number, the newly added collection work and the work types of the praise works in the previous period are obtained, and the quantity of the newly added attention account number of the same account number type, the quantity of the newly added collection work of the same work type and the quantity of the newly added praise works of the same work type are obtained;
Then obtaining a new attention rate I, a new collection rate I and a new praise rate I according to the first step and the second step;
Fourthly, obtaining a designated number of newly added attention account numbers, collected works and praise works of a user in a proximity period;
Then, the account types of the newly added attention account numbers, the newly added collection works and the work types of the praise works are obtained, and the quantity of the newly added attention account numbers of the same account types, the quantity of the newly added collection works of the same work types and the quantity of the newly added praise works of the same work types are obtained;
then obtaining a new attention rate II, a new collection rate II and a new praise rate II according to the first step and the second step;
fifthly, firstly selecting a designated work type, acquiring a related total collection ratio, a related total praise ratio, a related new added collection ratio I, a related new added praise ratio I, a related new added collection ratio II and a related new added praise ratio II, and marking the two as ZS, ZD and XS 1、XD1、XS2、XD2 respectively;
Then selecting an account number type containing a designated work type, acquiring a related total attention ratio, a new attention ratio I and a new attention ratio II, and marking the account number type as ZG and XG 1、XG2 respectively;
then by the formula:
Obtaining a preference coefficient Y of the appointed work, wherein beta 1、β2、β3 is a preset proportionality coefficient;
Sixthly, obtaining preference coefficients of other work types according to the mode of the fifth step, and sequencing the preference coefficients in a sequence from small to large;
the recommendation analysis unit is used for acquiring the popular works recommended by all the media platforms and the updated works of the attention account numbers according to the preference coefficients, calculating the recommendation coefficients of the popular works and the updated works, sequencing the recommendation coefficients according to the sequence from large to small, and preferentially pushing the popular works and the updated works with the front recommendation coefficients to the user;
The specific analysis mode of the recommended analysis unit is as follows:
S1, acquiring updated works of popular works and attention accounts recommended by all media platforms, and simultaneously acquiring work access amounts, work types and browsing behaviors of the popular works and the updated works, wherein the browsing behaviors are divided into non-browsing behaviors which are not checked on the popular works and the updated works and browsed behaviors which are checked on the popular works and the updated works;
S2, recommending and assigning all the hot works and updated works of different work types, wherein the recommended assignments of the hot works and the updated works of the corresponding work types are sequentially 0.1, 0.2, 0.3 according to the arrangement sequence of preference coefficients, and the recommended assignment marks of the hot works and the updated works are TF i, i=1, 2, 3 and represent the quantity of all the hot works and the updated works, and the larger the preference coefficient is, the larger the recommended assignments of the hot works and the updated works of the corresponding work types are;
S3, substituting the hot works and all the works of the updated works into a value assignment interval of corresponding interval values according to the access quantity of the hot works and the updated works, taking specific values of the value assignment interval as the access values of the hot works and the updated works, and marking the access values of the hot works and the updated works as FF i, wherein the interval values in the value assignment interval and the specific values thereof are preset values, the interval values in the value assignment interval represent the set interval of the access quantity, the specific values are sequentially 0.11, 0.12, 0.13 and the number of the access values according to the size of the divided interval values, and the larger the interval values in the value assignment interval are, namely the larger the access values of the hot works and the updated works corresponding to the types of the works are;
S4, performing browsing assignment on all popular works and updated works according to browsing behaviors of users, wherein the browsing assignment comprises the following specific steps: the unbrown behavior is assigned to "1", the browsed behavior is assigned to "0", and the browsed assignment of all popular works and updated works is marked as LF i;
S5, obtaining recommendation coefficients of all popular works and updated works through a formula TJ i＝(TFi*α1+FFi*α2)*LFi, wherein alpha 1、α2 is a preset proportionality coefficient;
S6, sequencing the recommendation coefficients of all the popular works and updated works according to the sequence from large to small, and preferentially pushing the popular works and updated works with the front recommendation coefficients to a user;
the searching recommendation unit is used for inputting target titles in a search box of the media platform by a user, searching to obtain all target works, calculating according to the similarity of all target works, calculating to obtain recommendation coefficients of all target works by combining the recommendation analysis unit, sequencing the recommendation coefficients according to the sequence from large to small, and pushing target works with the front recommendation coefficients to the user in priority.
2. The intelligent media content recommendation system according to claim 1, wherein the specific recommendation mode of the search recommendation unit is as follows:
x1, acquiring all target works, and optionally selecting a group of target works as reference works;
Then, in the reference work, intercepting video frames in the reference work at intervals of standard time t to obtain a plurality of video frames Z1 j, j=1, 2, 3, and the number of acquired video frames I is represented;
X2, obtaining a plurality of groups of video frames II from the rest target works, and respectively carrying out similarity calculation on the video frames II and Z1 1 to obtain comparison similarity values X 1 of the video frames and Z1 1, wherein the similarity calculation mode is that the similarity is calculated through a histogram algorithm and a hash algorithm;
the similarity value is then compared with a preset contrast value X 0:
When the video frame two includes X 1≥X0 of the video frame two, after the video frame two, acquiring a plurality of video frames three Z3 j+1 at intervals of a standard time t, and performing similarity calculation on the video frames Z1 2、Z13、Z14 and Z3 2、Z33、Z34 and the video frames X 2、X3、X4 and X 2、X3、X4 and X 0 respectively:
If u continuous comparison similarity values are all larger than a preset comparison value, acquiring corresponding target works, and using the corresponding target works as comparison works, wherein the comparison works are represented as other target works with similarity with the reference works higher than a set value;
If the u continuous comparison similarity values are not greater than the preset comparison value, acquiring corresponding target works, and using the corresponding target works as residual reference works, wherein Yu Can works are represented as other target works with similarity with the reference works being lower than a set value;
When the X 1≥X0 of the video frame II is not included in the video frames II, a corresponding target work is obtained and is used as a rest parameter work;
X4, processing the residual ginseng works again in the mode of the steps X1 and X2 to obtain comparison works corresponding to the residual ginseng works, other residual ginseng works and the comparison works corresponding to the other residual ginseng works;
X5, firstly calculating the recommendation coefficients of the reference works and all the comparison works through a recommendation analysis unit, then selecting a group of target works with the maximum recommendation coefficients as first selection works, simultaneously calculating the recommendation coefficients of the rest reference works and the corresponding comparison works, and selecting a group of target works with the maximum recommendation coefficients as second selection works, third selection works;
calculating the recommendation coefficients of the first referral work, the second referral work, the third referral work and the third referral work by a recommendation analysis unit;
X6, sorting the first, second and third referral works according to the recommendation coefficient, and pushing the target works with the front recommendation coefficient to the user in a priority mode;
And then, acquiring recommendation coefficients of the residual target works, sequencing the recommendation coefficients according to the sequence from large to small, sequentially pushing the residual target works to a user according to the sequence, and pushing the residual target works to the user after the residual target works are placed in the first selecting work, the second selecting work, the third selecting work and the third selecting work.
CN202310633278.1A 2023-05-31 2023-05-31 Media content intelligent recommendation system based on artificial intelligence Active CN116662659B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310633278.1A CN116662659B (en) 2023-05-31 2023-05-31 Media content intelligent recommendation system based on artificial intelligence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310633278.1A CN116662659B (en) 2023-05-31 2023-05-31 Media content intelligent recommendation system based on artificial intelligence

Publications (2)

Publication Number Publication Date
CN116662659A CN116662659A (en) 2023-08-29
CN116662659B true CN116662659B (en) 2024-05-14

Family

ID=87725564

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310633278.1A Active CN116662659B (en) 2023-05-31 2023-05-31 Media content intelligent recommendation system based on artificial intelligence

Country Status (1)

Country Link
CN (1) CN116662659B (en)

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8732737B1 (en) * 2013-01-25 2014-05-20 Mobitv, Inc. Geographic context weighted content recommendation
CN106028071A (en) * 2016-05-17 2016-10-12 Tcl集团股份有限公司 Video recommendation method and system
CN107943953A (en) * 2017-11-24 2018-04-20 福建中金在线信息科技有限公司 List recommends method, apparatus, electronic equipment and computer-readable recording medium
KR20200030252A (en) * 2018-09-12 2020-03-20 (주)버즈아트 Apparatus and method for providing artwork
CN114265953A (en) * 2021-12-31 2022-04-01 上海艾牌科技集团有限公司 Short video recommendation method, system, device and medium based on label
WO2022105269A1 (en) * 2020-11-17 2022-05-27 北京达佳互联信息技术有限公司 Live broadcast video pushing method and device
CN114564653A (en) * 2020-11-27 2022-05-31 北京达佳互联信息技术有限公司 Information recommendation method and device, server and storage medium
CN114969493A (en) * 2021-02-22 2022-08-30 腾讯科技（北京）有限公司 Content recommendation method and related device
CN115129900A (en) * 2021-03-26 2022-09-30 腾讯科技（深圳）有限公司 Multimedia recommendation method, device, equipment and storage medium

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20210406838A1 (en) * 2020-06-25 2021-12-30 Microsoft Technology Licensing, Llc Recommendations using session relevance and incremental learning

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8732737B1 (en) * 2013-01-25 2014-05-20 Mobitv, Inc. Geographic context weighted content recommendation
CN106028071A (en) * 2016-05-17 2016-10-12 Tcl集团股份有限公司 Video recommendation method and system
CN107943953A (en) * 2017-11-24 2018-04-20 福建中金在线信息科技有限公司 List recommends method, apparatus, electronic equipment and computer-readable recording medium
KR20200030252A (en) * 2018-09-12 2020-03-20 (주)버즈아트 Apparatus and method for providing artwork
WO2022105269A1 (en) * 2020-11-17 2022-05-27 北京达佳互联信息技术有限公司 Live broadcast video pushing method and device
CN114564653A (en) * 2020-11-27 2022-05-31 北京达佳互联信息技术有限公司 Information recommendation method and device, server and storage medium
CN114969493A (en) * 2021-02-22 2022-08-30 腾讯科技（北京）有限公司 Content recommendation method and related device
CN115129900A (en) * 2021-03-26 2022-09-30 腾讯科技（深圳）有限公司 Multimedia recommendation method, device, equipment and storage medium
CN114265953A (en) * 2021-12-31 2022-04-01 上海艾牌科技集团有限公司 Short video recommendation method, system, device and medium based on label

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
An Efficient Personalized Video Recommendation Algorithm Based on Mixed Mode;Wei Zhang等;2019 IEEE International Conferences on Ubiquitous Computing & Communications (IUCC) and Data Science and Computational Intelligence (DSCI) and Smart Computing, Networking and Services (SmartCNS);20200206;第367-373页 *
基于数据融合方法的群推荐***研究;王锐;李军;葛晓滨;叶淳康;;巢湖学院学报;20171125(06);第33-41+98页 *
抖音的算法推荐特点分析;王海燕;;新媒体研究;20181114(20);第25-26+37页 *

Also Published As

Publication number Publication date
CN116662659A (en) 2023-08-29

Similar Documents

Publication Publication Date Title
CN107888950A (en) A kind of method and system for recommending video
CN108460082B (en) Recommendation method and device and electronic equipment
US8893012B1 (en) Visual indicator based on relative rating of content item
CN108629665A (en) A kind of individual commodity recommendation method and system
CN105095431A (en) Method and device for pushing videos based on behavior information of user
CN107844525A (en) A kind of information personalized recommendation method, system and device based on user behavior
CN111914172B (en) Medical information recommendation method and system based on user tags
CN106168980A (en) Multimedia resource recommends sort method and device
CN109168047A (en) Video recommendation method, device, server and storage medium
CN111722766A (en) Multimedia resource display method and device
CN111291191A (en) Radio and television knowledge graph construction method and device
CN103309894A (en) User attribute-based search realization method and system
CN114936301A (en) Intelligent household building material data management method, device, equipment and storage medium
CN107707940A (en) Video sequencing method, device, server and system
CN116883116A (en) Personalized commodity recommendation system driven by user behavior analysis and method thereof
CN108009194A (en) A kind of books method for pushing, electronic equipment, storage medium and device
CN116662659B (en) Media content intelligent recommendation system based on artificial intelligence
CN109299368B (en) Method and system for intelligent and personalized recommendation of environmental information resources AI
CN112765374A (en) Education resource screening system and method for information push
CN110413895B (en) Information pushing method and related device
CN116070028A (en) Page-based content information pushing method, device, equipment and storage medium
CN111932109B (en) User influence evaluation system for mobile short video application
CN112540819B (en) Method for automatically generating recommended detailed page and form page according to query page
CN108109043A (en) A kind of commending system reduces the method for repeating to recommend
CN113657766A (en) Tourist attraction joy index metering method based on tourist multi-metadata

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
TA01 Transfer of patent application right

Effective date of registration: 20240425

Address after: Room 2017, Building 1, No. 202 Wanbo Second Road, Nancun Town, Panyu District, Guangzhou City, Guangdong Province, 510000

Applicant after: Guangzhou Puff Media Co.,Ltd.

Country or region after: China

Address before: Hongxing Dianjin Business Center, No. 208 Pushang Avenue, Jinshan Street, Cangshan District, Fuzhou City, Fujian Province, 350028 (Unit 10, 14th Floor, Building 1, Jinhua Community (Hongxing International Phase II))

Applicant before: Fujian Mojie Culture Development Co.,Ltd.

Country or region before: China

TA01 Transfer of patent application right
GR01 Patent grant
GR01 Patent grant