CN109388760A - Recommend label acquisition method, media content recommendations method, apparatus and storage medium - Google Patents
Recommend label acquisition method, media content recommendations method, apparatus and storage medium Download PDFInfo
- Publication number
- CN109388760A CN109388760A CN201710654487.9A CN201710654487A CN109388760A CN 109388760 A CN109388760 A CN 109388760A CN 201710654487 A CN201710654487 A CN 201710654487A CN 109388760 A CN109388760 A CN 109388760A
- Authority
- CN
- China
- Prior art keywords
- label
- content
- user
- media content
- recommendation
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
This application discloses a kind of recommendation label acquisition methods, it include: to obtain the label data of each sample content in multiple sample contents and launch data, user behavior data associated with the multiple sample content is obtained, the label data of each sample content includes the label for including in each sample content;For each of label data label, determine the quality score of the label, the quality score be used to characterize the label as content tab can recommendation;And quality score is met at least one label of predetermined condition as recommendation label, formation recommendation tag set.Disclosed herein as well is media content recommendations method, corresponding device and storage mediums.
Description
Technical field
This application involves Internet technical fields, more particularly to recommend label acquisition method, media content recommendations method, dress
It sets and storage medium.
Background technique
With the development of internet technology, people can pass through the various types of texts of network reading.Label (tag) be with
The very strong keyword of text relevant, it can be briefly described and be classified to content of text.In media content supplying system
In, it is the basis that subsequent article is recalled and recommended that the tag of media content, which extracts work,.The tag of media content is accurately extracted,
The core point for refining text receives more and more attention.
Summary of the invention
Present application example provides a kind of recommendation label acquisition method, comprising:
It obtains the label data of each sample content in multiple sample contents and launches data, obtain and the multiple sample
The associated user behavior data of content, the label data of each sample content include the mark for including in each sample content
Label;
For each of label data label, following processing is executed:
According to the dispensing data of each sample content comprising the label, the user's acceptance of the label is determined;
According to the user behavior data associated with each sample content comprising the label, the user of the label is determined
Interest-degree parameter;And
According to the user's acceptance and the user interest degree parameter, the quality score of the label, the quality are determined
Score for characterize the label as content tab can recommendation;
And
At least one label that quality score is met predetermined condition forms as label is recommended and recommends tag set.
Optionally, wherein the dispensing data include exposure data and/or click data;
The dispensing data of each sample content of the basis comprising the label, determine the user's acceptance of the label, comprising:
Obtain the exposure data and/or click data of each sample content comprising the label;
According to the exposure data and/or the click data of each sample content comprising the label, the label is determined
The user's acceptance.
Optionally, wherein the label data of each sample content includes: at least one label and its in the sample content
Weight;
The method further includes: for each of candidate tag set label, from including the label
Weight of the label in each sample content is extracted in the label data of each sample content;
Wherein, the exposure data and/or click data of each sample content of the basis comprising the label, determining should
The user's acceptance of label, comprising: according in the weight of the label in each sample content, each sample comprising the label
The exposure data and/or the click data held, determines the user's acceptance.
Optionally, wherein the user's acceptance is determined using following formula (1):
Wherein, N is the number of the sample content comprising the label, and i is i-th of content in N number of sample content,
tagweightiFor weight of the label in i-th of content, hit_muni is the click volume of i-th of content, and post_muni is
The light exposure of i-th of content.
Optionally, wherein the user interest degree parameter of the identified label includes the click volume of the label, the label
At least one of in subscription amount and the volumes of searches of the label.
Optionally, wherein the quality score of the determination label includes: by the click volume to the label, the label
Subscription amount and the label volumes of searches at least one of and the user's acceptance be weighted summation, obtain the quality
Scoring.
Optionally, the method further includes:
It include the recommendation label in statistics fixed time period for any recommendation label in the recommendation tag set
Content to be pushed quantity;
Extract the recommendation label that the content to be pushed quantity meets predetermined condition;
The Annual distribution of the quantity of content to be pushed of the statistics comprising the recommendation label chosen;
The recommendation label that the Annual distribution is unsatisfactory for predetermined condition is deleted from the recommendation tag set.
Optionally, the method further includes:
For multiple media contents to be pushed, at least one keyword of each media content to be pushed is extracted;
According to the recommendation tag set and at least one keyword of each media content to be pushed, determine each
At least one label of a media content to be pushed.
Optionally, wherein at least one label of each media content to be pushed of the determination includes:
Word frequency of each keyword in the media content to be pushed is obtained, in the multiple keyword
Any keyword executes following processing:
When there is label corresponding with the keyword in the recommendation tag set, the first scoring of the keyword is set
It is set to the first preset value;
When label corresponding with the keyword is not present in the recommendation tag set, by the first scoring of the keyword
It is set as the second preset value;
The second scoring of the keyword is determined according to word frequency of the keyword in the media content to be pushed;
It is scored according to first scoring and described second and determines the third scoring of the keyword;
Third scoring is met into the keyword of predetermined condition as the label of the media content.
Optionally, described for each of label data label, execute following processing, comprising:
Candidate tag set is determined according to the label data of each sample content;
For each of candidate tag set label, the processing is executed.
Present application example additionally provides a kind of media content recommendations method, comprising:
For multiple media contents to be pushed, at least one keyword of each media content to be pushed is extracted;
The recommendation tag set and each media content to be pushed obtained according to the method for claim 1
At least one keyword, determine at least one label of each media content to be pushed;
The media content recommendations request that applications client is sent is received, includes the application in media content recommendations request
The user identifier of client;
The interest tags of the user are determined according to the user identifier;
Using in its label exist label corresponding with the interest tags of the user media content to be pushed as
Alternative media content;
For each alternative media content, according at least one label of the alternative media content and the interest of the user
Label determines the matching degree of the alternative media content and the interest tags of the user;
Matching degree is met into the alternative media content of predetermined condition as the media content recommended;
The information of the media content of the recommendation is returned into the applications client.
Present application example additionally provides a kind of recommendation label acquisition device, comprising:
Acquiring unit is obtained to obtain the label data of each sample content in multiple sample contents and launch data
User behavior data associated with the multiple sample content, the label data of each sample content include each sample
The label for including in content;
Score unit, to:
For each of candidate tag set label, following processing is executed:
According to the dispensing data of each sample content comprising the label, the user's acceptance of the label is determined;
According to the user behavior data associated with each sample content comprising the label, the user of the label is determined
Interest-degree parameter;And
According to the user's acceptance and the user interest degree parameter, the quality score of the label, the quality are determined
Score for characterize the label as content tab can recommendation;And
Recommend tag determination unit, at least one label quality score to be met to predetermined condition is marked as recommendation
Label form and recommend tag set.
Optionally, described device further comprises cleaning unit, to:
It include the recommendation label in statistics fixed time period for any recommendation label in the recommendation tag set
Content to be pushed quantity;
Extract the recommendation label that the content to be pushed quantity meets predetermined condition;
The Annual distribution of the quantity of content to be pushed of the statistics comprising the recommendation label chosen;
The recommendation label that the Annual distribution is unsatisfactory for predetermined condition is deleted from the recommendation tag set.
Present application example additionally provides a kind of media content recommendations device, comprising:
Tag extraction unit extracts each media content to be pushed to be directed to multiple media contents to be pushed
At least one keyword;According to the method for claim 1 obtain the recommendation tag set and each wait pushing
Media content at least one keyword, determine at least one label of each media content to be pushed;
Request reception unit, to receive the media content recommendations request of applications client transmission, the media content recommendations
It include the user identifier of the applications client in request;
Media content selection unit, to:
The interest tags of the user are determined according to the user identifier;
Using in its label exist label corresponding with the interest tags of the user media content to be pushed as
Alternative media content;
For each alternative media content, according at least one label of the alternative media content and the interest of the user
Label determines the matching degree of the alternative media content and the interest tags of the user;
Matching degree is met into the alternative media content of predetermined condition as the media content recommended;
Information transmitting unit, the information of the media content of the recommendation is returned to the applications client.
Present application example additionally provides a kind of computer readable storage medium, is stored with computer-readable instruction, can make
At least one processor executes method as described above.
Using above scheme provided by the present application, it can help to choose in multiple candidate keywords of media content and more close
Suitable label, and then preferably carry out the recommendation of media content.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is the system architecture figure that present application example is related to;
Fig. 2 is the flow chart that one example of the application recommends label acquisition method;
Fig. 3 A is the page figure of one example news of the application and its display label;
Fig. 3 B is the page figure of news under one example label theme of the application;
Fig. 3 C is the page figure of one example news home of the application;
Fig. 4 is the flow chart of one example media content recommendations method of the application;
Fig. 5 is the structural schematic diagram that one example of the application recommends label acquisition device;
Fig. 6 is the structural schematic diagram of one example media content recommendations device of the application;And
Fig. 7 is that the calculating equipment in present application example forms structural schematic diagram.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that the described embodiment is only a part of the embodiment of the present invention, instead of all the embodiments.Based on this
Embodiment in invention, every other reality obtained by those of ordinary skill in the art without making creative efforts
Example is applied, shall fall within the protection scope of the present invention.
Present applicant proposes recommendation label acquisition method, media content recommendations method, apparatus and storage medium, this method can
Applied in system architecture shown in FIG. 1.As shown in Figure 1, the system architecture includes: application (APP) client 101, push letter
Platform 102 and pushed information provider client 105 are ceased, these entities can be communicated by internet 106, wherein pushing
Information platform 102 includes application server 103, user's access database of record 104 and processing platform 107.
The application server 103 in the access pushed information platform 102 of applications client 101 can be used in terminal user, than
Such as: browsing news or article.When user accesses application server 103 using applications client 101, applications client
101 can be reported to the access behavior of user application server 103, and application server 103 is by the access behavioral data of user
User is stored in access in database of record 104.While 101 report of user of applications client accesses behavior, applications client
101 can issue information push request to pushed information platform 102, and pushed information platform 102 can will be pushed with the information and be asked
The media content to match is asked to be pushed to applications client 101.By pushed information provider client 105, pushed information is mentioned
The material for the media content that it to be pushed can be uploaded to pushed information platform 102 by supplier, specifically, be uploaded to application service
Device 103, to generate the media content for push accordingly.After application server receives the media content, by the media
Content is sent to the processing platform 107 in pushed information platform 102, and processing platform 107 extracts the label of the media content, will
The label of the media content of extraction is stored in user and accesses in database of record 104.During media content push, application
It include the mark of user in the information push request that application server 103 of the client 101 into pushed information platform 102 issues
Know, the history access record found in database of record 104 with the user can be accessed in user according to the mark of the user,
Determine that the interest tags of the user, application server 103 access in user record simultaneously according to the history access record of the user
The label of media content is extracted in database 104, and then will be to according to the label determination of the interest tags of user and media content
The media content that applications client 101 pushes.
For processing platform 107 when extracting the label of media content, it includes distributed text that user, which accesses in database of record 104,
Part system, such as HDFS, the relevant user behavior data of media content to save to push;It further include relational data
Base management system, such as mysql, to save the label data of media content extraction;It further include Key-Value database, example
Such as redis, to save statistical data relevant to label.Processing platform 107 is accessed in database of record 104 from user and is obtained
The media content pushed extracts user behavior number relevant to the media content pushed from distributed file system
According to extracting the extraction label data of the media content pushed from Relational DBMS, pushed away according to described
The media content passed through, the label data for pushing media content, user relevant to the media content pushed
Behavioral data forms and recommends tag set.The label of media content is extracted according to the recommendation tag set of formation.For example,
When extracting the label of article, by obtaining multiple keywords after the operation such as being segmented to article text, removing stop words, for
Multiple keyword carries out various dimensions scoring, for example, word frequency, as a dimensions, whether keyword is to recommend tally set
Word in conjunction is as another dimensions.Here, label is recommended to be alternatively referred to as high-quality label.
In some instances, when carrying out various dimensions scoring to above-mentioned multiple keywords, label is recommended to score as one
Dimension, establishes the library label (tag), and recommending tag set is the subset of entire tag library, use recommended keywords as one
When dimensions, when a keyword, which appears in, to be recommended in tag set, a relatively high score value is got to the keyword,
When the keyword, which does not appear in, to be recommended in tag set, a relatively low score value is got to the keyword.For mentioning
Recommendation tag set is taken, in some instances, on the basis of existing tag library, the system in article corpus (i.e. sample content)
The statistical nature of each label in tag library, such as TF-IDF feature are counted, after label is sorted according to statistical nature, determines and recommends
Candidate word in tag set.Meanwhile in conjunction with a part of artificial staking-out work, is determined after calibration and recommend tag set.The extraction
Recommend the technical solution of tag set, the evaluative dimension used is relatively simple, the bad control of accuracy to label is recommended, simultaneously
The artificial mark work in part is also needed, the problems such as higher cost, efficiency are lower, accuracy is unstable is brought.
In order to solve the above technical problems, the application proposes a kind of recommendation label acquisition method, is applied to processing platform 107,
Recommend tag set for obtaining, as shown in Fig. 2, method includes the following steps:
Step 201: obtain in multiple sample contents the label data of each sample content and launch data, obtain with it is described
Multiple associated user behavior datas of sample content, the label data of each sample content include in each sample content
The label for including.
Sample content can be obtained from some existing databases, such as is accessed in database of record 104 and obtained from user
It takes the label data of sample content and sample content and launches data.For example, can be extracted in HDFS and sample content phase
The dispensing data of pass extract the label data of sample content in mysql, to save the label data of media content extraction.
For example, accessing the article for obtaining and pushing in database of record 104 from user, and push away when the sample content is article
The label data and dispensing data of the article passed through.It include what a sample content was extracted in the label data of the sample content
The weight of label and each label in the sample content.It is described launch data include sample content exposure data and/
Or user is to the click data of the sample content.For example, obtaining what each news was extracted when the sample content is news
Label, weight of each label in the news, the dispensing data of each news, exposure data including news and/or
Click data.It is described before institute, it, can also be from some existing data in the label data and dispensing data for obtaining sample content
Library, such as user access and obtain user behavior data associated with the multiple sample content, example in database of record 104
Such as, user behavior data relevant to sample content can be extracted in HDFS.The user behavior data includes: user to sample
Content shows the tag that the click data of tag, user include to sample content to the subscription data of sample content display tag, user
Search data.For example, the application server 103 in Fig. 1 is news media's server when the sample content is news,
In the displayed page of a news as shown in Figure 3A, showing in the bottom of news has the label 1, label 2 and label of the news
3, wherein label 1, label 2 and label 3 are the label in the label that news shown in Fig. 3 A is extracted, and the label of the displaying can
Think that the news extracts all or part in label.The mark 301 for clicking label 1, shows page figure as shown in Figure 3B,
In the page figure, news relevant to label 1 is illustrated.In the page figure described in Fig. 3 B, while showing there is subscription control
302, user can subscribe to label 1 by clicking subscription control 302, to can show user in the personal homepage of client
Article mark under the label of subscription and the subscription label.Meanwhile in the news home shown in Fig. 3 C, there is search control
Part 303 and corresponding input frame 304, user interior input label, click control 303 can carry out to input in input frame 304
The search of label.Above-mentioned user clicks the behavior of the mark 301 of label 1, clicks the behavior of control 302 and in input frame 304
Interior input label and the behavior for clicking control 303, can all be recorded in the cookie of user, and user is in access news media clothes
When business device, the cookie is carried, thus the behavioral data of the available user of news media's server, and it is stored in user's visit
It asks in database of record 104.
Step 202: for each of label data label, executing following processing: according to including the label
The dispensing data of each sample content, determine the user's acceptance of the label.
The user's acceptance of the label is determined according to the light exposure of the sample content comprising a label and/or click volume,
Wherein acceptance level of the user's acceptance characterization user to the sample content for being extracted the label.Wherein, it is extracted the mark
The click volume of the sample content of label is bigger, illustrates that user is higher to the acceptance of the label, meanwhile, it is extracted the sample of the label
The light exposure of this content is bigger, illustrates that user is higher to the acceptance of the label.For example, being extracted the new of label " power of chaotic state "
Article is heard, compared to the news article of " refreshing work ghost power " is extracted, click volume is higher, illustrates user for label " power of chaotic state "
Acceptance it is higher.
Step 203: according to the user behavior data associated with each sample content comprising the label, determining the mark
The user interest degree parameter of label.
The user behavior data includes: that user shows that the click data of tag, user are aobvious to sample content to sample content
Show the search data for the tag that the subscription data of tag, user include to sample content.Behavioral data of the user for a label
Reflect user to the interested degree of the label.For some keywords for being related to the people's livelihood, such as: school district room, room rate, two tires
Deng user often shows higher interest-degree for this class keywords.According to related to each sample content comprising the label
The user behavior data of connection can determine that the user interest degree parameter of the label, the user interest degree parameter include mark
Click volume, subscription amount and the volumes of searches of label.
Step 204: according to the user's acceptance and the user interest degree parameter, determine the quality score of the label,
The quality score be used for characterize the label as content tab can recommendation.
According to the user's acceptance and user interest degree parameter of a label, the quality score of the label is determined, scoring is got over
High label illustrates that the label is more suitable for the label of sample content, is extracted the label of the sample content, is pushing the sample
When this content, it is easier to be clicked by user.
Step 205: at least one label that quality score is met predetermined condition forms as label is recommended and recommends label
Set.
The predetermined condition can be, and label is ranked up according to its quality score, and the top n label in sequence is made
To recommend label.Or one preset threshold of setting, quality score is greater than the label of the preset threshold as recommendation label.Choosing
The recommendation label taken, which is formed, recommends tag set.
Using recommendation label acquisition method provided by the present application, candidate tally set is formed according to the label that sample content is extracted
It closes, for each of candidate tag set label, by label weight, in conjunction with the label condition being related in user behavior
And the multi-dimensional datas such as input situation of the sample content comprising the label, various dimensions scoring is carried out to label, to mention
The high label of significance level is taken out as recommendation label.The process for extracting recommendation label is an automated procedure, does not need manually to mark
Note, reduces costs, while completion that can be more efficient, as a result stable, and can quickly update.
In some instances, the dispensing data include exposure data and/or click data, in above-mentioned steps 204,
The dispensing data for executing each sample content of the basis comprising the label, when determining the user's acceptance of the label, including with
Lower step:
S101: the exposure data and/or click data of each sample content comprising the label are obtained;
S102: according to the exposure data and/or the click data of each sample content comprising the label, determining should
The user's acceptance of label.
Multiple sample contents may include the label of identical extraction, thus, in candidate tag set, including one
The sample content of label is there are multinomial, for example, for label i, the sample content comprising the label are as follows: in sample content 1, sample
Hold 2, sample content 3 ... sample content N.Determine the label i's according to the light exposure of N number of sample content and click volume
User's acceptance.
In some instances, the label data of each sample content includes: at least one label and its in the sample content
In weight;The method further includes: for each of candidate tag set label, from including the label
Weight of the label in each sample content is extracted in the label data of each sample content;Wherein, in above-mentioned steps S102,
The exposure data and/or click data for executing each sample content of the basis comprising the label, determine the use of the label
When the acceptance of family, comprising the following steps: S201: according to the weight of the label in each sample content, including the label
The exposure data and/or the click data of each sample content, determine the user's acceptance.
Choose recommend label when, in addition to need to refer to include label sample content dispensing data, choose user more
Acceptable label, while the content of text of sample content is also needed to refer to, select the label that can embody sample content.
The label data of sample content includes weight of the label in the sample content that the sample content includes, and the label is at this
Weight in sample content can obtain in several ways, can be extracted according to the TF-IDF feature of label in sample content
The weight of label
In some instances, the user's acceptance is determined using following formula (1):
Wherein, N is the number of the sample content comprising the label, and i is i-th of content in N number of sample content,
tagweightiFor weight of the label in i-th of sample content, hit_muni is the click volume of i-th of sample content,
Post_muni is the light exposure of i-th of sample content.
In some instances, wherein the user interest degree parameter of the identified label includes the click volume of the label, is somebody's turn to do
At least one of in the subscription amount of label and the volumes of searches of the label.
Wherein, the user behavior data associated with each sample content comprising the label includes user to described each
The click behavioral data of aobvious label outside sample content, user to the subscription of the label accordingly and user searches the label
Rope data.For example, as shown in Figure 3A, the bottom of article shows there is label, which is institute when the sample content is article
The all or part for stating the label of article extraction can show the page as shown in Figure 3B when user clicks the mark 301 of label 1
Figure, records the behavioral data that user clicks label 1 at this time, and click volume corresponding with label 1 adds 1.The page figure shown in Fig. 3 B
In, i.e., in the theme page relevant to label 1, user clicks control 302, subscribes to label 1, records user at this time and clicks control
302 behavioral data, while the corresponding subscription amount of label 1 adds 1.In the homepage of the client shown in Fig. 3 C, user can be with
The input label 1 in input frame 304, while control 304 is clicked, the search to label 1 is realized, at this point, record user searches for mark
The behavioral data of label 1, while the volumes of searches of label 1 adds 1.By the behavioral data of the user of record, the point an of label is determined
The amount of hitting, subscription amount and volumes of searches.
It in some instances, include following step executing the quality score of the determination label in above-mentioned steps 206
It is rapid:
S201: pass through at least one in the subscription amount of click volume, the label and the volumes of searches of the label to the label
And the user's acceptance is weighted summation, obtains the quality score.
After user's acceptance, click volume, subscription amount and the volumes of searches for getting the label, the wherein click of the label
Amount, subscription amount and volumes of searches are the user interest degree parameter of the label, interest size of the reaction user to the label.It can be by it
In user interest degree parameter and the user's acceptance determine that the quality score of the label or two user interest degrees are joined
Number any combination and the user's acceptance determine the quality score or three user interest degree parameters and the use of the label
Family acceptance determines the quality score of the label.
When three user interest degree parameters and the user's acceptance determine the quality score of the label, using following public affairs
Formula (2) determines the quality score of the label:
tagscore=wh*taghitscore+ws*tagshowscore+wc*tagcollectscore+wsetagsearchscore (2)
Wherein, taghitscoreUser's acceptance, w for the labelhFor the weight of the user's acceptance, tagshowscore
For the click volume of the label, wsFor the weight of the click volume, tagcollectscoreFor the subscription amount of the label, wcFor the subscription
The weight of amount, tagsearchscoreFor the volumes of searches of the label, wseFor the weight of described search amount.
The quality score of each label is calculated by formula (2), it can be by the label in candidate tag set according to mark
The quality score of label is ranked up, using the top n label in sequence as recommendation label.But it mentions by adopting the above technical scheme
The recommendation label taken finds in practice, comes the label of front, have the bigger label of some ranges (such as " China ").
The range of these labels is excessive, the key point of expression content that cannot be relatively good and the point of interest of user, so being not suitable as
Recommend label, it should delete from recommendation tag set.
Thus, the application proposes another example, to delete the excessive recommendation mark of above range in recommending tag set
Label, the method further includes following steps:
S31: including the recommendation mark in statistics fixed time period for any recommendation label in the recommendation tag set
The content to be pushed quantity of label.
The content to be pushed data are the content to be pushed of full dose, for example, counting each when the content is article
It article to be pushed, wherein the article to be pushed had extracted label, for each in the recommendation tag set of extraction
A recommendation label, count every day wait push in article, comprising one recommend label article to be pushed quantity, that is, count
How many article to be pushed under each recommendation label.The article quantity to be pushed under label is recommended to be stored in user by each daily
It accesses in database of record 104 in redis.
S32: the recommendation label that the content to be pushed quantity meets predetermined condition is extracted.
Shown in example also as above, when the content is article, label will be recommended according to corresponding article to be pushed
Quantity is ranked up, and M recommendation label preceding in sequence can be selected and.The recommendation label of front is come, containing there are two types of feelings
Condition wants the large-scale recommendation label removed one is us, and there are also one is recommendation labels more popular in the recent period.
S33: the Annual distribution of the quantity of content to be pushed of the statistics comprising the recommendation label chosen.
Described in step S32 as above, the forward recommendation label of the sequence of selection, it may be possible to the bigger recommendation label of range,
It is also likely to be popular recommendation label, needs to retain popular recommendation label, deletes and recommend label on a large scale, therefore, statistics
For a certain period, the Annual distribution of the quantity of the content to be pushed of the recommendation label comprising selection.Recommend on a large scale
The corresponding Annual distribution of label is relatively uniform, is not in bigger variation;The popular corresponding Annual distribution of recommendation label
Unevenly, it can deviate and be evenly distributed farther out.For either one or two of selected recommendation label, database of record 104 is accessed in user
In redis in extract recommendation label every day media content to be pushed quantity, and then determine the recommendation label one
Annual distribution in the section time, the time point of the quantity for the media content to be pushed that Annual distribution characterization recommends label to include
Cloth.
S34: the recommendation label that the Annual distribution is unsatisfactory for predetermined condition is deleted from the recommendation tag set.
When recommending the Annual distribution of label than more gentle, then confirm that the recommendation label is a wide range of recommendation label, when
Recommend the Annual distribution of label to change greatly, then confirms that the recommendation label is hot recommendation label.It will recommend to mark on a large scale
Label are deleted from recommendation tag set.
Using the example, hot recommendation tag recognition can be come out, the biggish recommendation label of range can be cleared out of
Recommend tag set.
In some instances, the method further includes following steps:
S41: for multiple media contents to be pushed, at least one key of each media content to be pushed is extracted
Word.
In system architecture shown in Fig. 1, pushed information provider client 105 can the media content be pushed it
Material upload to the application server 103 in pushed information platform 102, to generate media content to be pushed accordingly.It is right
In each media content, application server 103 sends it to processing platform 107, is extracted in the media by processing platform 107
The label of appearance.At least one keyword of the media content is extracted first, it can be by the textual portions of the media content by dividing
At least one keyword is obtained after word, removal stop words.
S42: it according to the recommendation tag set and at least one keyword of each media content to be pushed, determines
At least one label of each media content to be pushed.
For a media content, obtained according at least one keyword of the media content and above-mentioned recommendation label
The recommendation tag set that method obtains, determines at least one label of the media content.Specifically, a pass of the media content
Whether keyword, which appears in, is recommended to score as one-dimensional dimensions to the keyword in tag set.For example, can be
Recommend directly to search recommendation label identical with the keyword in tag set, when a keyword is in recommending tag set
There are when corresponding label, a relatively high score value is made a call to the keyword, when a keyword is recommending tag set
In be not present corresponding label when, make a call to a relatively low score value to the keyword.The score value that the keyword is beaten
An item rating as the keyword.Can be in conjunction with other dimensions, such as keyword is in the text of corresponding media content
Word frequency in this content is as a marking dimension, another marking of the available keyword.Multinomial by keyword beats
Divide weighted sum, obtains the comprehensive score of the keyword, be then ranked up keyword according to the comprehensive score, before selection
Label of the L keyword as the media content.
In some instances, in above-mentioned steps S42, described each media content to be pushed of determination is being executed extremely
When a few label, comprising the following steps:
S51: word frequency of each keyword in the media content to be pushed is obtained.
The media content to be pushed includes at least one keyword, counts each keyword in the media content
Textual portions in the frequency that occurs, that is, count the number occurred in the content of text.
For any keyword in the multiple keyword, following processing is executed:
S52: when there is label corresponding with the keyword in the recommendation tag set, the first of the keyword is commented
Set up separately and is set to the first preset value.
Whether first scoring is recommended in tag set to characterize the keyword and appear in, that is, reacting the keyword is
No is to recommend label, when the keyword, which appears in, to be recommended in tag set, sets first for the first scoring of the keyword
Preset value.
S53: when label corresponding with the keyword is not present in the recommendation tag set, by the first of the keyword
Scoring is set as the second preset value;
When the keyword, which does not appear in, to be recommended in tag set, it is pre- that second is set by the first scoring of the keyword
If value.Wherein, the first preset value is a relatively high score value, and the second preset value is a relatively low score value.
S54: the second scoring of the keyword is determined according to word frequency of the keyword in the media content to be pushed.
Whether keyword, which appears in, is recommended as one-dimensional dimensions in tag set, can be in conjunction with other scoring dimensions
Degree, word frequency can be the second dimensions, and corresponding score value is the second scoring, and a keyword goes out in the text of media content
The existing frequency is higher, and second scoring is higher, and the frequency occurred in the text of media content is lower, and second scoring is got over
It is low.
S55: it is scored according to first scoring and described second and determines that the third of the keyword scores.
First scoring and the second scoring can be weighted summation, obtained summed result as the keyword the
Three scorings.Wherein, the weight of the first scoring and the second scoring can rule of thumb be preset.
S56: third scoring is met into the keyword of predetermined condition as the label of the media content.
Keyword can be ranked up according to the scoring of its third, Q keyword is as institute before choosing in ranking results
State the label of media content.It can also be using the keyword for being scored above preset threshold as the label of the media content.
The recommendation tag set of above-mentioned acquisition can select more suitable keyword and make in multiple candidate keywords of media content
For the label of media content.
The application also proposed a kind of media content recommendations method, and this method is applied in pushed information platform 102, such as Fig. 4
It is shown, method includes the following steps:
Step 401: for multiple media contents to be pushed, extracting at least one of each media content to be pushed
Keyword.
In system architecture shown in Fig. 1, pushed information provider client 105 can the media content be pushed it
Material upload to pushed information platform 102, to generate media content to be pushed accordingly.For each media content,
At least one keyword of the media content is extracted, the textual portions of the media content can be passed through into participle, removal stop words
After obtain at least one keyword.
Step 402: according to the described above recommendation tag set for recommending label acquisition method to obtain and each
At least one keyword of media content to be pushed determines at least one label of each media content to be pushed.
For a media content, obtained according at least one keyword of the media content and above-mentioned recommendation label
The recommendation tag set that method obtains, determines at least one label of the media content.Specifically, a pass of the media content
Whether keyword, which appears in, is recommended to score as one-dimensional dimensions to the keyword in tag set.For example, can be
Recommend directly to search recommendation label identical with the keyword in tag set, when a keyword is in recommending tag set
There are when corresponding label, a relatively high score value is made a call to the keyword, when a keyword is recommending tag set
In be not present corresponding label when, make a call to a relatively low score value to the keyword.The score value that the keyword is beaten
An item rating as the keyword.Can be in conjunction with other dimensions, such as keyword is in the text of corresponding media content
Word frequency in this content is as a marking dimension, another marking of the available keyword.Multinomial by keyword beats
Divide weighted sum, obtains the comprehensive score of the keyword, be then ranked up keyword according to the comprehensive score, before selection
Label of the L keyword as the media content.
Wherein, above-mentioned steps 401 and step 402 are executed by processing platform 107, extract media to be pushed to execute
The label of content.
Step 403: receiving the media content recommendations request that applications client is sent, include in media content recommendations request
The user identifier of the applications client.
The application server 103 in the access pushed information platform 102 of applications client 101 can be used in terminal user, than
Such as: browsing news or article.When user accesses application server 103 using applications client 101, applications client
101 can issue media content recommendations request to pushed information platform 102, include 101 institute of applications client in the request
In the mark of user.The cookie of the user can be sent to and answer when accessing application server 103 by applications client 101
With server 103, the history access data of the user are carried in the cookie of user, application server 103 is according to the user's
History access data can determine the interest tags of the user, while can also determine the class for the media content that the user accessed
Not Deng other information, the determining information preservation is accessed in database of record 104 in user.
Step 404: the interest tags of the user are determined according to the user identifier.
Described in step 403 as above, application server 103 can determine the interest of user according to the history access record of user
Label, can be in the Database user access database of record 104 of the application server 103 according to the mark of the user
Obtain the interest tags of user.The classification etc. for the media content that user accessed is obtained at the same time it can also the mark according to user
Other information.
Step 405: will there are the media to be pushed of label corresponding with the interest tags of the user in its label
Content is as alternative media content.
In this step, according to the interest tags of the user of acquisition, media are carried out in media content wait push in full dose
Content is recalled.Wherein, the media content to be pushed is in all media of the transmission of pushed information provider client 105
Hold, for each recommendation request, in the full dose wait push in the media that selection will recommend client in media content
Hold.For example, when being when pushing article, when being recalled wait push article, when including in the label of article wait push media content
When with the interest tags of user, then the article to be pushed is recalled, as candidate article, wherein the interest tags of user can be with
It is multiple.When the interest tags according to user recall when pushing article negligible amounts, can also be accessed according to user
The classification of media content, such as the classification of article that user accessed recall some articles to be pushed again.
Step 406: each alternative media content is directed to, according at least one label of the alternative media content and the use
The interest tags at family determine the matching degree of the alternative media content and the interest tags of the user.
In this step, the matching degree of each alternative media content and the interest tags of user, a candidate matchmaker are calculated
Internal container has multiple labels, meanwhile, user also include it is multiple, determined in alternative media according to the label of alternative media content
The label vector of appearance determines interest tags vector according to the interest tags of user, calculate the label of the alternative media content to
Matching degree between amount and the interest tags vector.
Step 407: matching degree is met into the alternative media content of predetermined condition as the media content recommended.
Alternative media content can be ranked up according to the matching degree being calculated, selected and sorted forward P
Alternative media content is as recommendation media content.It can also be more than the alternative media content of preset threshold using matching degree as recommendation
Media content.
Step 408: the information of the media content of the recommendation is returned into the applications client.
The link of the media content of determining recommendation is sent to applications client, so that applications client is according to the chain
It obtains and takes corresponding media content.For example, the link for recommending news is sent to client for news recommendation, show
Page figure as shown in Figure 3 C illustrates the link for recommending news in the page figure, and user clicks the link of a wherein news,
Show corresponding news pages.
Above-mentioned steps 402-408 is executed by application server 103, the push process to specific media content.
In some instances, described for each of described label data mark executing in above-mentioned steps 201
Label when executing following processing, for the label in the label data, execute following processing:
S61: candidate tag set is determined according to the label data of each sample content.
It include the label that each sample content is extracted in the label data of the sample content, by each sample content
The set of the label of extraction is as the candidate tag set, in this process, for identical label in candidate tag set
Carry out duplicate removal processing.For the label of extraction each single item sample content, the textual portions of the sample content can be segmented,
Some keywords are obtained after removal stop words, the scoring of various dimensions are then carried out to the keyword, then by high pass of scoring
Label of the keyword as the sample content.The dimensions may include the word frequency that keyword occurs in content of text,
Whether keyword, which appears in, is recommended in tag set.Recommendation label acquisition method provided by the present application recommends label for obtaining
Set, while extracting recommendation tag set is the process regularly updated, in this step, can use existing recommendation mark
Label set is used as one-dimensional dimensions, to extract the label of sample content.
S62: for each of candidate tag set label, the quality score of the label is calculated.
Its quality score is calculated to each of candidate tag set label according to above-mentioned method.
Present invention also provides a kind of recommendation label acquisition device 500, as shown in Figure 5, comprising:
Acquiring unit 501 is obtained to obtain the label data of each sample content in multiple sample contents and launch data
User behavior data associated with the multiple sample content is taken, the label data of each sample content includes each sample
The label for including in this content;
Score unit 502, to:
For each of label data label, following processing is executed:
According to the dispensing data of each sample content comprising the label, the user's acceptance of the label is determined;
According to the user behavior data associated with each sample content comprising the label, the user of the label is determined
Interest-degree parameter;And
According to the user's acceptance and the user interest degree parameter, the quality score of the label, the quality are determined
Score for characterize the label as content tab can recommendation;And
Recommend tag determination unit 503, quality score to be met at least one label of predetermined condition as recommendation
Label forms and recommends tag set.
In some instances, described device further comprises cleaning unit 504, to:
It include the recommendation label in statistics fixed time period for any recommendation label in the recommendation tag set
Content to be pushed quantity;
Extract the recommendation label that the content to be pushed quantity meets predetermined condition;
The Annual distribution of the quantity of content to be pushed of the statistics comprising the recommendation label chosen;
The recommendation label that the Annual distribution is unsatisfactory for predetermined condition is deleted from the recommendation tag set.
Present invention also provides a kind of media content recommendations devices 600, as shown in Figure 6, comprising:
Tag extraction unit 601 is extracted in each media to be pushed to be directed to multiple media contents to be pushed
At least one keyword held;According to the method for claim 1 obtain the recommendation tag set and each wait pushing away
At least one keyword of the media content sent determines at least one label of each media content to be pushed;
Request reception unit 602, to receive the media content recommendations request of applications client transmission, which is pushed away
Recommend the user identifier in request including the applications client;
Media content selection unit 603, to:
The interest tags of the user are determined according to the user identifier;
Using in its label exist label corresponding with the interest tags of the user media content to be pushed as
Alternative media content;
For each alternative media content, according at least one label of the alternative media content and the interest of the user
Label determines the matching degree of the alternative media content and the interest tags of the user;
Matching degree is met into the alternative media content of predetermined condition as the media content recommended;
Information transmitting unit 604, the information of the media content of the recommendation is returned to the applications client.
Present invention also provides a kind of computer readable storage mediums, are stored with computer-readable instruction, can make at least
One processor executes method as described above.
Fig. 7 shows the composite structural diagram of the calculating equipment where communication link.As shown in fig. 7, the calculating equipment
Including one or more processor (CPU) 702, communication module 704, memory 706, user interface 710, and for interconnecting
The communication bus 708 of these components.
Processor 702 can send and receive data by communication module 704 to realize network communication and/or local communication.
User interface 710 includes one or more output equipments 712 comprising one or more speakers and/or one
Or multiple visual displays.User interface 710 also includes one or more input equipments 714 comprising such as, keyboard, mouse
Mark, voice command input unit or loudspeaker, touch screen displays, touch sensitive tablet, posture capture camera or other inputs are pressed
Button or control etc..
Memory 706 can be high-speed random access memory, such as DRAM, SRAM, DDR RAM or other deposit at random
Take solid storage device;Or nonvolatile memory, such as one or more disk storage equipments, optical disc memory apparatus, sudden strain of a muscle
Deposit equipment or other non-volatile solid-state memory devices.
The executable instruction set of 706 storage processor 702 of memory, comprising:
Operating system 716, including the program for handling various basic system services and for executing hardware dependent tasks;
It is this to apply journey including the various application programs for obtaining recommendation label and media content recommendations using 718
Sequence can be realized the process flow in above-mentioned each example, for example may include that label acquisition device 500 or media content is recommended to push away
Some or all of recommend in device 600 unit or module.Recommend label acquisition device 500 or media content recommendations device 600
In each unit at least one unit can store machine-executable instruction.Processor 702 is by executing memory 706
Machine-executable instruction in middle each unit at least one unit, and then can be realized in above-mentioned each unit or module at least
The function of one module.
It should be noted that step and module not all in above-mentioned each process and each structure chart be all it is necessary, can
To ignore certain steps or module according to the actual needs.Each step execution sequence be not it is fixed, can according to need into
Row adjustment.The division of each module is intended merely to facilitate the division functionally that description uses, and in actual implementation, a module can
It is realized with point by multiple modules, the function of multiple modules can also be realized by the same module, these modules can be located at same
In a equipment, it can also be located in different equipment.
Hardware module in each embodiment can in hardware or hardware platform adds the mode of software to realize.Above-mentioned software
Including machine readable instructions, it is stored in non-volatile memory medium.Therefore, each embodiment can also be presented as software product.
In each example, hardware can be by special hardware or the hardware realization of execution machine readable instructions.For example, hardware can be with
Permanent circuit or logical device (such as application specific processor, such as FPGA or ASIC) specially to design are used to complete specifically to grasp
Make.Hardware also may include programmable logic device or circuit by software provisional configuration (as included general processor or other
Programmable processor) for executing specific operation.
In addition, each example of the application can pass through the data processor by data processing equipment such as computer execution
To realize.Obviously, data processor constitutes the application.In addition, being commonly stored data processing in one storage medium
Program is by directly reading out storage medium or the storage by program being installed or being copied to data processing equipment for program
It is executed in equipment (such as hard disk and/or memory).Therefore, such storage medium also constitutes the application, and present invention also provides one
Kind non-volatile memory medium, wherein being stored with data processor, this data processor can be used for executing in the application
State any one of method example example.
The corresponding machine readable instructions of Fig. 7 module can make operating system operated on computer etc. described herein to complete
Some or all of operation.Non-volatile computer readable storage medium storing program for executing can be set in the expansion board in insertion computer
In the memory set or write the memory being arranged in the expanding element being connected to a computer.It is mounted on expansion board or expansion
Opening up CPU on unit etc. can be according to instruction execution part and whole practical operations.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (15)
1. a kind of recommendation label acquisition method characterized by comprising
It obtains the label data of each sample content in multiple sample contents and launches data, obtain and the multiple sample content
Associated user behavior data, the label data of each sample content include the label for including in each sample content;
For each of label data label, following processing is executed:
According to the dispensing data of each sample content comprising the label, the user's acceptance of the label is determined;
According to the user behavior data associated with each sample content comprising the label, the user interest of the label is determined
Spend parameter;And
According to the user's acceptance and the user interest degree parameter, the quality score of the label, the quality score are determined
For characterize the label as content tab can recommendation;
And
At least one label that quality score is met predetermined condition forms as label is recommended and recommends tag set.
2. according to the method described in claim 1, wherein, the dispensing data include exposure data and/or click data;
The dispensing data of each sample content of the basis comprising the label, determine the user's acceptance of the label, comprising:
Obtain the exposure data and/or click data of each sample content comprising the label;
According to the exposure data and/or the click data of each sample content comprising the label, the institute of the label is determined
State user's acceptance.
3. according to the method described in claim 2, wherein, the label data of each sample content include: at least one label and
Its weight in the sample content;
The method further includes: for each of candidate tag set label, from the various kinds comprising the label
Weight of the label in each sample content is extracted in the label data of this content;
Wherein, the exposure data and/or click data of each sample content of the basis comprising the label, determines the label
User's acceptance, comprising: according to the weight of the label in each sample content, each sample content comprising the label
The exposure data and/or the click data, determine the user's acceptance.
4. according to the method described in claim 3, wherein, determining the user's acceptance using following formula (1):
Wherein, N is the number of the sample content comprising the label, and i is i-th of content in N number of sample content,
tagweightiFor weight of the label in i-th of content, hit_muni is the click volume of i-th of content, and post_muni is
The light exposure of i-th of content.
5. according to the method described in claim 1, wherein, the user interest degree parameter of the identified label includes the label
At least one of in the volumes of searches of click volume, the subscription amount of the label and the label.
6. according to the method described in claim 5, wherein, the quality score of the determination label includes: by the label
Click volume, in the volumes of searches of the subscription amount of the label and the label at least one of and the user's acceptance be weighted and ask
With obtain the quality score.
7. according to the method described in claim 1, further comprising:
It include the recommendation label wait push away in statistics fixed time period for any recommendation label in the recommendation tag set
Send content quantity;
Extract the recommendation label that the content to be pushed quantity meets predetermined condition;
The Annual distribution of the quantity of content to be pushed of the statistics comprising the recommendation label chosen;
The recommendation label that the Annual distribution is unsatisfactory for predetermined condition is deleted from the recommendation tag set.
8. according to the method described in claim 1, further comprising:
For multiple media contents to be pushed, at least one keyword of each media content to be pushed is extracted;
According to the recommendation tag set and at least one keyword of each media content to be pushed, determine that each is waited for
At least one label of the media content of push.
9. according to the method described in claim 8, wherein, at least one of each media content to be pushed of the determination is marked
Label include:
Word frequency of each keyword in the media content to be pushed is obtained, for any in the multiple keyword
Keyword executes following processing:
When there is label corresponding with the keyword in the recommendation tag set, set the first scoring of the keyword to
First preset value;
When label corresponding with the keyword is not present in the recommendation tag set, the first scoring of the keyword is arranged
For the second preset value;
The second scoring of the keyword is determined according to word frequency of the keyword in the media content to be pushed;
It is scored according to first scoring and described second and determines the third scoring of the keyword;
Third scoring is met into the keyword of predetermined condition as the label of the media content.
10. according to the method described in claim 1, wherein, described for each of label data label, execution is such as
Lower processing, comprising:
Candidate tag set is determined according to the label data of each sample content;
For each of candidate tag set label, the processing is executed.
11. a kind of media content recommendations method characterized by comprising
For multiple media contents to be pushed, at least one keyword of each media content to be pushed is extracted;
According to the method for claim 1, the recommendation tag set and each media content to be pushed obtained extremely
A few keyword, determines at least one label of each media content to be pushed;
The media content recommendations request that applications client is sent is received, includes the application client in media content recommendations request
The user identifier at end;
The interest tags of the user are determined according to the user identifier;
Using the media content to be pushed that there is label corresponding with the interest tags of the user in its label as candidate
Media content;
For each alternative media content, according at least one label of the alternative media content and the interest mark of the user
Label determine the matching degree of the alternative media content and the interest tags of the user;
Matching degree is met into the alternative media content of predetermined condition as the media content recommended;
The information of the media content of the recommendation is returned into the applications client.
12. a kind of recommendation label acquisition device characterized by comprising
Acquiring unit, to obtain the label data of each sample content in multiple sample contents and launch data, acquisition and institute
The associated user behavior data of multiple sample contents is stated, the label data of each sample content includes each sample content
In include label;
Score unit, to:
For each of label data label, following processing is executed:
According to the dispensing data of each sample content comprising the label, the user's acceptance of the label is determined;
According to the user behavior data associated with each sample content comprising the label, the user interest of the label is determined
Spend parameter;And
According to the user's acceptance and the user interest degree parameter, the quality score of the label, the quality score are determined
For characterize the label as content tab can recommendation;And
Recommend tag determination unit, quality score to be met at least one label of predetermined condition as recommendation label, shape
At recommendation tag set.
13. device according to claim 12 further comprises cleaning unit, to:
It include the recommendation label wait push away in statistics fixed time period for any recommendation label in the recommendation tag set
Send content quantity;
Extract the recommendation label that the content to be pushed quantity meets predetermined condition;
The Annual distribution of the quantity of content to be pushed of the statistics comprising the recommendation label chosen;
The recommendation label that the Annual distribution is unsatisfactory for predetermined condition is deleted from the recommendation tag set.
14. a kind of media content recommendations device characterized by comprising
Tag extraction unit extracts each media content to be pushed extremely to be directed to multiple media contents to be pushed
A few keyword;The recommendation tag set obtained according to the method for claim 1 and each matchmaker to be pushed
At least one keyword held in vivo determines at least one label of each media content to be pushed;
Request reception unit, to receive the media content recommendations request of applications client transmission, media content recommendations request
In include the applications client user identifier;
Media content selection unit, to:
The interest tags of the user are determined according to the user identifier;
Using the media content to be pushed that there is label corresponding with the interest tags of the user in its label as candidate
Media content;
For each alternative media content, according at least one label of the alternative media content and the interest mark of the user
Label determine the matching degree of the alternative media content and the interest tags of the user;
Matching degree is met into the alternative media content of predetermined condition as the media content recommended;
Information transmitting unit, the information of the media content of the recommendation is returned to the applications client.
15. a kind of computer readable storage medium, which is characterized in that be stored with computer-readable instruction, at least one can be made
Processor executes such as the described in any item methods of claim 1-11.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710654487.9A CN109388760B (en) | 2017-08-03 | 2017-08-03 | Recommendation label obtaining method, media content recommendation method, device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710654487.9A CN109388760B (en) | 2017-08-03 | 2017-08-03 | Recommendation label obtaining method, media content recommendation method, device and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109388760A true CN109388760A (en) | 2019-02-26 |
CN109388760B CN109388760B (en) | 2023-04-07 |
Family
ID=65413774
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710654487.9A Active CN109388760B (en) | 2017-08-03 | 2017-08-03 | Recommendation label obtaining method, media content recommendation method, device and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109388760B (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109889891A (en) * | 2019-03-05 | 2019-06-14 | 腾讯科技(深圳)有限公司 | Obtain the method, apparatus and storage medium of target media file |
CN109918662A (en) * | 2019-03-04 | 2019-06-21 | 腾讯科技(深圳)有限公司 | A kind of label of e-sourcing determines method, apparatus and readable medium |
CN110209958A (en) * | 2019-05-31 | 2019-09-06 | Oppo广东移动通信有限公司 | Thematic production method, device, terminal and storage medium |
CN110263188A (en) * | 2019-05-29 | 2019-09-20 | 深圳市元征科技股份有限公司 | Media data methods of marking, device and equipment |
CN110348993A (en) * | 2019-06-28 | 2019-10-18 | 北京淇瑀信息科技有限公司 | Wind is discussed and select model workers determination method, determining device and the electronic equipment of type label |
CN112781604A (en) * | 2019-11-08 | 2021-05-11 | 逸驾智能科技有限公司 | Method, apparatus, device and computer-readable storage medium for navigation |
CN113343024A (en) * | 2021-08-04 | 2021-09-03 | 北京达佳互联信息技术有限公司 | Object recommendation method and device, electronic equipment and storage medium |
CN113761361A (en) * | 2021-07-29 | 2021-12-07 | 深圳市思为软件技术有限公司 | House property information recommendation method and terminal equipment |
CN114491283A (en) * | 2022-04-02 | 2022-05-13 | 浙江口碑网络技术有限公司 | Object recommendation method and device and electronic equipment |
CN114866603A (en) * | 2022-04-19 | 2022-08-05 | 北京安锐卓越信息技术股份有限公司 | Information pushing method and device, electronic equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TW200919210A (en) * | 2007-07-18 | 2009-05-01 | Steven Kays | Adaptive electronic design |
US20120191716A1 (en) * | 2002-06-24 | 2012-07-26 | Nosa Omoigui | System and method for knowledge retrieval, management, delivery and presentation |
CN103886090A (en) * | 2014-03-31 | 2014-06-25 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user favorites |
CN104537115A (en) * | 2015-01-21 | 2015-04-22 | 北京字节跳动科技有限公司 | Method and device for exploring user interests |
CN105512326A (en) * | 2015-12-23 | 2016-04-20 | 成都品果科技有限公司 | Picture recommending method and system |
CN106686063A (en) * | 2016-12-07 | 2017-05-17 | 乐视控股(北京)有限公司 | Information recommendation method and apparatus, and electronic device |
-
2017
- 2017-08-03 CN CN201710654487.9A patent/CN109388760B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120191716A1 (en) * | 2002-06-24 | 2012-07-26 | Nosa Omoigui | System and method for knowledge retrieval, management, delivery and presentation |
TW200919210A (en) * | 2007-07-18 | 2009-05-01 | Steven Kays | Adaptive electronic design |
CN103886090A (en) * | 2014-03-31 | 2014-06-25 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user favorites |
CN104537115A (en) * | 2015-01-21 | 2015-04-22 | 北京字节跳动科技有限公司 | Method and device for exploring user interests |
CN105512326A (en) * | 2015-12-23 | 2016-04-20 | 成都品果科技有限公司 | Picture recommending method and system |
CN106686063A (en) * | 2016-12-07 | 2017-05-17 | 乐视控股(北京)有限公司 | Information recommendation method and apparatus, and electronic device |
Non-Patent Citations (2)
Title |
---|
KYUNGHYE YOON ET AL.: "An exploratory study of library website accessibility for visually impaired users", 《LIBRARY & INFORMATION SCIENCE RESEARCH》 * |
李春城: "个性化推荐系统在牛赞网中的应用", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109918662A (en) * | 2019-03-04 | 2019-06-21 | 腾讯科技(深圳)有限公司 | A kind of label of e-sourcing determines method, apparatus and readable medium |
CN109889891A (en) * | 2019-03-05 | 2019-06-14 | 腾讯科技(深圳)有限公司 | Obtain the method, apparatus and storage medium of target media file |
CN110263188A (en) * | 2019-05-29 | 2019-09-20 | 深圳市元征科技股份有限公司 | Media data methods of marking, device and equipment |
CN110263188B (en) * | 2019-05-29 | 2023-03-28 | 深圳市元征科技股份有限公司 | Media data scoring method, device and equipment |
CN110209958B (en) * | 2019-05-31 | 2022-03-15 | Oppo广东移动通信有限公司 | Thematic making method, device, terminal and storage medium |
CN110209958A (en) * | 2019-05-31 | 2019-09-06 | Oppo广东移动通信有限公司 | Thematic production method, device, terminal and storage medium |
CN110348993A (en) * | 2019-06-28 | 2019-10-18 | 北京淇瑀信息科技有限公司 | Wind is discussed and select model workers determination method, determining device and the electronic equipment of type label |
CN110348993B (en) * | 2019-06-28 | 2023-12-22 | 北京淇瑀信息科技有限公司 | Determination method and determination device for label for wind assessment model and electronic equipment |
CN112781604A (en) * | 2019-11-08 | 2021-05-11 | 逸驾智能科技有限公司 | Method, apparatus, device and computer-readable storage medium for navigation |
CN112781604B (en) * | 2019-11-08 | 2024-02-09 | 逸驾智能科技有限公司 | Method, apparatus, device and computer readable storage medium for navigation |
CN113761361A (en) * | 2021-07-29 | 2021-12-07 | 深圳市思为软件技术有限公司 | House property information recommendation method and terminal equipment |
CN113343024B (en) * | 2021-08-04 | 2021-12-07 | 北京达佳互联信息技术有限公司 | Object recommendation method and device, electronic equipment and storage medium |
CN113343024A (en) * | 2021-08-04 | 2021-09-03 | 北京达佳互联信息技术有限公司 | Object recommendation method and device, electronic equipment and storage medium |
CN114491283A (en) * | 2022-04-02 | 2022-05-13 | 浙江口碑网络技术有限公司 | Object recommendation method and device and electronic equipment |
CN114866603A (en) * | 2022-04-19 | 2022-08-05 | 北京安锐卓越信息技术股份有限公司 | Information pushing method and device, electronic equipment and storage medium |
CN114866603B (en) * | 2022-04-19 | 2024-05-03 | 北京安锐卓越信息技术股份有限公司 | Information pushing method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN109388760B (en) | 2023-04-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109388760A (en) | Recommend label acquisition method, media content recommendations method, apparatus and storage medium | |
CN109697629B (en) | Product data pushing method and device, storage medium and computer equipment | |
US8725717B2 (en) | System and method for identifying topics for short text communications | |
US8380727B2 (en) | Information processing device and method, program, and recording medium | |
US8615707B2 (en) | Adding new attributes to a structured presentation | |
CN109753601B (en) | Method and device for determining click rate of recommended information and electronic equipment | |
CN107295361B (en) | A kind of content delivery method | |
US20120066073A1 (en) | User interest analysis systems and methods | |
JP2015191655A (en) | Method and apparatus for generating recommendation page | |
CN105847985A (en) | Video recommendation method and device | |
US10713291B2 (en) | Electronic document generation using data from disparate sources | |
CN107273489A (en) | Content delivery method, electronic equipment and computer-readable storage medium | |
JP2013531289A (en) | Use of model information group in search | |
CN103885951A (en) | Graphics and text information releasing and generating method and graphics and text information releasing and generating device | |
CN103279513A (en) | Method for generating content label and method and device for providing multi-media content information | |
CN102708174A (en) | Method and device for displaying rich media information in browser | |
JP5494126B2 (en) | Document recommendation system, document recommendation device, document recommendation method, and program | |
CN104503988B (en) | searching method and device | |
CN109168047A (en) | Video recommendation method, device, server and storage medium | |
Wang et al. | Bilateral correspondence model for words-and-pictures association in multimedia-rich microblogs | |
CN103279504A (en) | Searching method and device based on ambiguity resolution | |
US8346799B1 (en) | Recent content rank adjustment with inverted decay | |
CN112579854A (en) | Information processing method, device, equipment and storage medium | |
CN112825089A (en) | Article recommendation method, article recommendation device, article recommendation equipment and storage medium | |
CN111061954A (en) | Search result sorting method and device and storage medium |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |