CN112818230B - Content recommendation method, device, electronic equipment and storage medium - Google Patents
Content recommendation method, device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN112818230B CN112818230B CN202110129782.9A CN202110129782A CN112818230B CN 112818230 B CN112818230 B CN 112818230B CN 202110129782 A CN202110129782 A CN 202110129782A CN 112818230 B CN112818230 B CN 112818230B
- Authority
- CN
- China
- Prior art keywords
- content
- tag
- combination
- combinations
- label
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 230000003993 interaction Effects 0.000 claims abstract description 57
- 230000004044 response Effects 0.000 claims abstract description 8
- 230000006399 behavior Effects 0.000 claims description 54
- 230000002452 interceptive effect Effects 0.000 claims description 8
- 230000001960 triggered effect Effects 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 3
- 238000003058 natural language processing Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 11
- 238000004590 computer program Methods 0.000 description 10
- 238000004891 communication Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 7
- 230000011218 segmentation Effects 0.000 description 6
- 238000013145 classification model Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000013522 software testing Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000009960 carding Methods 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000153 supplemental effect Effects 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 239000008207 working material Substances 0.000 description 1
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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The disclosure provides a content recommendation method, a content recommendation device, electronic equipment and a storage medium, and relates to the field of natural language processing, in particular to the field of intelligent recommendation. The specific implementation scheme is as follows: in response to the interaction behavior for the target content in the content database, determining M tag combinations of the target content according to N tags of the target content; updating the real-time weight of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations; and under the condition that a content recommendation request is received, determining recommended content in a content database according to the current real-time weight of the tag combinations in the candidate tag combination set. According to the embodiment of the disclosure, the recommended content can be determined more accurately and more closely to the user preference.
Description
Technical Field
The present disclosure relates to the field of natural language processing, and in particular, to the field of intelligent recommendation.
Background
With the development of the internet and the rapid expansion of content, related content needs to be recommended to users in more and more scenes. For example, with the development of enterprises, document knowledge deposited inside the enterprises is more and more, and in order to enable knowledge to flow inside the enterprises efficiently, a knowledge recommendation system needs to be built, so that knowledge initiative searching is realized. How to more accurately recommend related content to a user is a hotspot problem in the field of intelligent recommendation.
Disclosure of Invention
The disclosure provides a content recommendation method, a content recommendation device, electronic equipment and a storage medium.
According to an aspect of the present disclosure, there is provided a content recommendation method including:
in response to the interaction behavior for the target content in the content database, determining M tag combinations of the target content according to N tags of the target content; wherein N is an integer greater than or equal to 2, and M is a positive integer;
updating the real-time weight of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations;
and under the condition that a content recommendation request is received, determining recommended content in a content database according to the current real-time weight of the tag combinations in the candidate tag combination set.
According to another aspect of the present disclosure, there is provided a content recommendation apparatus including:
the interaction response module is used for responding to the interaction behavior aiming at the target content in the content database and determining M label combinations of the target content according to N labels of the target content; wherein N is an integer greater than or equal to 2, and M is a positive integer;
the weight updating module is used for updating the real-time weight of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations;
And the recommending module is used for determining recommended content in the content database according to the current real-time weight of the tag combinations in the candidate tag combination set under the condition of receiving the content recommending request.
According to another aspect of the present disclosure, there is provided an electronic device including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of the embodiments of the present disclosure.
According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method of any of the embodiments of the present disclosure.
According to another aspect of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the method in any of the embodiments of the present disclosure.
According to the technology disclosed by the disclosure, when a user performs interactive behaviors on target content in a content database, determining tag combinations of the target content, and updating real-time weights of tag combinations in candidate tag combinations according to types of the interactive behaviors and the tag combinations of the target content. Because the weight used for determining the recommended content is updated in units of tag combinations, the tag combinations can more accurately characterize the user preferences, and the weight is updated in real time based on the type of interaction behavior, the recommended content can be determined more accurately and more closely to the user preferences based on the weight.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.
Drawings
The drawings are for a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:
FIG. 1 is a schematic diagram of a content recommendation method provided by one embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a content recommendation method provided by another embodiment of the present disclosure;
FIG. 3 is a schematic diagram of one example application of a content recommendation system in an embodiment of the present disclosure;
FIG. 4 is a schematic diagram of a content recommendation device provided by one embodiment of the present disclosure;
FIG. 5 is a schematic diagram of a content recommendation device provided in another embodiment of the present disclosure;
fig. 6 is a block diagram of an electronic device for implementing a content recommendation method of an embodiment of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
In the field of intelligent recommendation, in order to intuitively embody the matching degree between the recommended content and the user requirement, the recommendation should generally have interpretability. To achieve interpretive of recommendations, content recommendations are typically made based on tags.
In an exemplary manner, individual tags are determined for the user and the content in a tag hierarchy, the individual tags of the user and the individual tags of the content are matched, and a recommendation is made based on the matching results. In this way, the correlation between the user and the recommended content is high. However, with the expansion of content and the expansion of knowledge boundaries, the cost of maintaining a tag system based on information such as multidimensional knowledge types and topics is high, and the determination of an accurate single tag for a user and content is difficult.
Another exemplary embodiment maintains a plurality of flat simple tab systems corresponding to a plurality of dimensions, respectively, and determines tabs in the plurality of dimensions for the user and the content, respectively. The matching content is determined based on each tag of the user, respectively. For example, if the user's tag includes A, B and C, the user is recommended the content of tag a, the content of tag B, and the content of tag C. As such, despite ease of maintenance, expanding the tagging architecture, and determining tags for users and content, drift in the correlation between recommended content and users is caused.
The solution provided by the embodiments of the present disclosure is mainly used for solving at least one of the above problems.
Fig. 1 is a schematic diagram of a content recommendation method according to an embodiment of the present disclosure. As shown in fig. 1, the method includes:
step S110, in response to the interaction behavior for the target content in the content database, determining M label combinations of the target content according to N labels of the target content; wherein N is an integer greater than or equal to 2, and M is a positive integer;
step S120, updating the real-time weight of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations;
step S130, under the condition that a content recommendation request is received, determining recommended content in a content database according to the current real-time weight of the tag combinations in the candidate tag combination set.
The content recommendation method may be performed by an electronic device, such as a server or a user device of a content recommendation system, for example. The content recommendation system may be integrated in an information sharing App (Application), a social App, an enterprise office App, etc.
In the disclosed embodiments, content may refer to various carriers of information, such as knowledge documents, articles, advertisements, and the like. The content database may be a collection of at least one content, such as a collection of individual content in a content recommendation system. The target content may be any one of the contents in the content database. The target content may also be a specific content in a content database, such as content distributed over a specific period of time or some type of content, etc.
For example, the electronic device may determine the tags for each content in the content database based on a multi-dimensional tag hierarchy. The process of determining the tags for content may also be referred to as content tagging.
In some embodiments, the tagging system may be a preconfigured set of tags. Correspondingly, for each content in the content database, the electronic device can respectively select a label from a plurality of label sets corresponding to a plurality of dimensions as the label of the content.
Alternatively, the electronic device may construct a tag system by extracting keywords from each content of the content database, identifying entities, or receiving a manually induced tag set, etc.
Alternatively, the electronic device may tag the content using a deep learning model such as a classifier, a similarity prediction model, or the like. For example, the content is input into a deep learning model, and a corresponding tag is output from the deep learning model. The electronic device may also tag the content based on preset rules.
Taking the content recommendation system of the enterprise office App as an example, various exemplary ways of constructing and tagging content are provided:
example one:
in this example, the plurality of dimensions includes a key entity dimension. Wherein a key entity may refer to an entity to which the content relates. For example, key entities may include departments, projects, technical terms, tool software, etc. to which content relates.
The process of constructing a tab system for critical entity dimensions may include: common keywords in the enterprise are mined from the content database as labels in the label system by using an unsupervised keyword extraction algorithm and a NER (Named Entity Recognition ) model. And/or obtain existing structured information in the enterprise office App, such as departments, projects, tools software, etc. in ERP (Enterprise Resource Planning ) as tags in the tagging system. The number of labels contained in the label system of the key entity dimension constructed by the method is large, for example, 1.8 ten thousand, 3 ten thousand and the like.
The process of tagging content with a key entity dimension may include: word segmentation is carried out on all labels in a label system, word segmentation corresponding to each label is obtained, inverted indexes between the labels and the word segmentation are established, namely the indexes are the word segmentation, and the search object is the index system of the labels. And carrying out word segmentation on the title and the text of the content to obtain the segmented words of the content. And searching at least one label in the label system based on the content word segmentation and the inverted index to serve as a candidate label. And determining the similarity between the content and each candidate label by using a similarity prediction model, for example, a SimNet network structure-based prediction model. And determining the label with the similarity exceeding a preset threshold as the label in the key entity dimension corresponding to the content, or determining the label with the maximum similarity as the label in the key entity dimension corresponding to the content.
Example two:
in this example, the plurality of dimensions includes a theme dimension. For example, topics may include machine learning, deep learning, software testing, enterprise culture, and the like.
The process of building a tab hierarchy for a subject dimension may include: and manually combing and summarizing the related knowledge according to the internal knowledge distribution of the enterprise to obtain a plurality of labels in the theme dimension. The electronic equipment receives the input multiple tags, and a tag system of the theme dimension is obtained based on the input multiple tags. The number of tags contained in the tag system for the subject dimension is relatively small, e.g., 50, 83, etc.
The process of tagging content with a subject dimension may include: and obtaining the label on the theme dimension corresponding to the content based on the label system of the theme dimension and the input content by utilizing the multi-classification model. For example, the multi-classification model may be derived from fine-tuning (fine-tune) a pre-trained model, such as the ernie2.0 model, using pairs of sample data. The sample data pair can be a data pair of content and a label obtained based on a preset rule. The preset rules include, for example: the labels of the contents including keywords such as "model", "neural network" or "convolution" are deep learning, and the labels of the contents stored in the enterprise database are working materials.
Example three:
in this example, the plurality of dimensions includes a tendentiousness dimension. Different contents including articles, knowledge documents and the like have different forms and depths, and are suitable for different people to read. For example, trending news is suitable for all people to read, while software testing base knowledge is suitable for primary scholars of the testing direction or people interested in the testing direction to read. Tags in the tendentiousness dimension may refer to tags related to the type of audience to which the content is suitable, such as popular news, domain dynamics, corporate institutes and welfare, technical entry knowledge, technical depth knowledge, product entry knowledge, product depth knowledge, general capabilities, and the like.
The process of building a tagging system in the prone dimension may include: and manually carding and summarizing the related knowledge according to the internal knowledge distribution of the enterprise to obtain a plurality of labels in the tendency dimension. The electronic device receives the input plurality of tags and obtains a tag system of a tendency dimension based on the input plurality of tags. The number of tags contained in the tag system in the prone dimension is relatively small, e.g., 7, 9, etc.
The process of tagging content with a tendentiousness dimension may include: and obtaining the label on the tendency dimension corresponding to the content based on the label system of the tendency dimension and the input content by utilizing the multi-classification model. For example, the multi-classification model may be derived from fine-tuning (fine-tune) a pre-trained model, such as the ernie2.0 model, using pairs of sample data. The sample data pair can be obtained based on manual labeling.
As can be seen from the above exemplary manner, the content in the content database may have a plurality of tags. In an example, according to the step S110, when the interaction behavior for the target content in the content database is detected, N tags of the target content may be arranged and combined to obtain M tag combinations of the target content. Here, each tag combination may include at least two tags such that the tag combination has a higher correlation with the content than a single tag. For example, if the tags of a certain article include A1, B6, and C2, the following 4 tag combinations can be obtained: a first tag combination comprising A1 and B6, a second tag combination comprising A1 and C2, a third tag combination comprising B6 and C2, and a third tag combination comprising A1, B6 and C2.
The interaction behavior may include, for example, browsing, collecting, canceling, feeding back, etc., of the content by the user, or may include presenting, recommending, etc., that the user does not browse after recommending a certain content. According to the above step S120, taking the user triggering the interaction for the target content as the first user as an example, the real-time weight of each candidate tag combination in the candidate tag combination set may be updated for the first user based on the type of the interaction and the M tag combinations of the target content. Wherein the set of candidate tag combinations may include a plurality of tag combinations, which may be referred to as candidate tag combinations.
In some exemplary embodiments, the set of candidate tag combinations for each user is the same. For example, labels in the label system of each dimension may be arranged and combined to obtain a plurality of candidate label combinations. For example, a label system of key entity dimension, topic dimension, and tendentiousness dimension is constructed, the number of labels contained in each label system being 18000, 83, and 9, respectively. 18000×83×9= 13446000 candidate tag combinations are obtained through permutation and combination, and a candidate tag combination set of each user is obtained based on the candidate tag combinations. In these embodiments, the real-time weights of the tag combinations in the candidate tag combination set may be different for each user. For example, for a first tag combination in the set of candidate tag combinations that includes tags A1 and B6, user 1 has a real-time weight of 50 and user 2 has a real-time weight of 60. When the interaction behavior is triggered, the electronic equipment determines the current real-time weight to be updated according to the ID (Identifier) of the user currently using the content recommendation system, and updates the real-time weight. Here, the real-time weights are used to characterize the magnitude of the relevance of the tag combination and the user, and their values can be accumulated during the update process without an upper limit.
In some exemplary embodiments, the candidate tag combination sets are different for each user. For example, a specific candidate tag combination set is configured for the user in advance according to related information of different users, such as occupation, reading habit and the like.
In these embodiments, each user may have its own set of candidate tag combinations. Each tag combination in each candidate tag combination set corresponds to a real-time weight. For example, the candidate tag combination for user 1 has a first tag combination comprising tags A1 and B6 in the set, but does not have a third tag combination comprising B6 and C2. The first tag combination and the third tag combination are present in the candidate tag combination set of user 2 at the same time. The real-time weight of the third tag combination in the candidate tag combination set for user 1 is 50 and the real-time weight of the third tag combination in the candidate tag combination set for user 2 is 60. The electronic device may record a user ID corresponding to each candidate tag combination set. When the interaction behavior is triggered, the electronic equipment determines a candidate label combination set to be updated from a plurality of candidate label combination sets corresponding to a plurality of user IDs according to the ID of the user currently using the content recommendation system, and updates the real-time weight of the candidate label combination in the candidate label combination set.
In practical application, when the real-time weight is updated each time, the real-time weight can be increased or decreased in different magnitudes according to the type of the interaction behavior. For example, browsing the corresponding increment of 1, collecting the corresponding increment of 1.5; the corresponding increasing amplitude of the collection canceling is-1, namely the decreasing amplitude is 1; the magnitude of the feedback dislike increase is-4; the increase amplitude corresponding to the target content recommended but not browsed by the user is x= (-1/120+pos/(120X 15)). Each time content is recommended to a user, the content identification is displayed in 120 pages, 15 content identifications are displayed in each page (the user can browse the content by clicking the content identification), pos represents the position number of the target content in each page, and the value is one of 1 to 15. According to the above formula, the more the target content is displayed in the page, the smaller the magnitude of the real-time weight needs to be reduced. For example, when the number is 15, that is, the last display position of the target content in the page, even if the user does not browse the target content, the real-time weight corresponding to the tag combination of the target content is not affected.
In the embodiment of the disclosure, the real-time weight of the tag combination in the candidate tag combination set is used for determining recommended content in the content database. For example, in the step S130, Y tag combinations with the largest real-time weights may be determined according to the real-time weights of all tag combinations in the candidate tag combination set, and the content corresponding to any one of the Y tag combinations may be determined from the content database as the recommended content, so as to present the recommended content to the user.
It should be noted that, after updating the real-time weight each time, the recommended content is not necessarily determined. The step of determining recommended content may be performed upon receipt of a content recommendation request. The content recommendation request may be generated based on a user operation, for example, when the user switches to a recommendation page in App. The content recommendation request may also be automatically generated, for example, at intervals, i.e., based on a predetermined frequency, to display a popup in the App, in which recommended content is displayed.
It can be seen that according to the content recommendation method provided by the embodiment of the present disclosure, when a user performs an interaction on a target content in a content database, a tag combination of the target content is determined first, and a real-time weight of the tag combination in the candidate tag combination set is updated according to a type of the interaction and the tag combination of the target content. Because the weight used for determining the recommended content is updated in the unit of the tag combination, the tag combination can more accurately represent the user preference, and the weight is updated in real time based on the type of the interaction behavior, the recommended content can be more accurately determined closer to the user preference based on the weight, and the relevance of the recommended content is improved.
For example, if the candidate set of tag combinations must contain all of the M tag combinations, the real-time weights of all of the M tag combinations may be updated. For example, in a case where the candidate tag combination set is obtained by arranging and combining the tags in the plurality of tag systems, the real-time weights of the M tag combinations in the candidate tag combination set may be updated directly.
If the candidate tag combination set does not necessarily contain all of the M tag combinations, the above-described step S120 may be implemented according to the following exemplary embodiment:
updating the real-time weights of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations may include:
and under the condition that the candidate label combination set comprises the ith label combination in M label combinations, updating the real-time weight of the ith label combination according to the type of the interaction behavior, wherein i is a positive integer less than or equal to M.
Optionally, updating the real-time weights of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations may include: in the case where the candidate tag combination set does not include the i-th tag combination described above, the i-th tag combination may not be processed.
For example, the M tag combinations of the target content include a first tag combination composed of tags A1 and B6, and if the candidate tag combination set to be updated also includes the first tag combination, the real-time weight of the first expression combination is updated. If the candidate tag combination set to be updated does not contain the first tag combination, the real-time weight does not need to be updated for the ith tag combination.
Therefore, the number of the label combinations in the candidate label combination set can be controlled, the operation amount of the electronic equipment during real-time online updating can be limited, the operation pressure of the electronic equipment can be reduced, and the updating efficiency can be improved.
Optionally, updating the real-time weights of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations may include:
if the number of the label combinations in the candidate label combination set is smaller than a first preset threshold value under the condition that the candidate label combination set does not contain the jth label combination in the M label combinations, adding the jth label combination in the candidate label combination set, and determining the real-time weight of the jth label combination according to the type of the interaction behavior; wherein j is a positive integer less than or equal to M.
For example, the M tag combinations of the target content include a first tag combination composed of tags A1 and B6, and if the candidate tag combination set to be updated also includes the first tag combination, the real-time weight of the first expression combination is updated. If the candidate label combination set to be updated does not contain the first label combination, but the number of label combinations in the candidate label combination set to be updated is smaller than the first preset threshold, the first label combination can be added in the candidate label combination set, and the real-time weight of the first expression combination can be determined as the initial real-time weight.
In this manner, the number of tag combinations in the set of candidate tag combinations may be limited to within a first preset threshold. In case the number is smaller than a first preset threshold, the real-time weight may be updated. Therefore, the calculation amount of the electronic equipment during real-time online updating is limited, and meanwhile, label combinations related to users can be ignored as little as possible, so that the recommendation accuracy is improved.
Illustratively, as shown in fig. 2, the content recommendation method may further include:
step S210, counting the interaction behaviors of each content in the content database triggered in a preset period of time to obtain the statistical weights of K label combinations; wherein K is a positive integer;
step S220, selecting L label combinations from the K label combinations according to the statistical weights of the K label combinations; wherein L is a positive integer less than or equal to K;
step S230, a candidate label combination set is obtained based on the L label combinations.
Alternatively, the steps S210 to S230 may be performed before the steps S110 to S130, or may be performed after the steps S110 to S130.
Specifically, in some exemplary embodiments, a set of candidate tag combinations may be predetermined, and then the real-time weights of the tag combinations in the set of candidate tag combinations may be updated based on the interaction behavior, and recommended content may be determined for the content recommendation request. In some exemplary embodiments, the candidate tag combination set may also be periodically redetermined during the process of updating the real-time weights in real-time and determining the recommended content. For example, offline calculation is performed at 0 point every day, statistics is performed on all interaction behaviors of the user in the previous month, statistical weights of all tag combinations are obtained, and a candidate tag combination set is redetermined.
Illustratively, when a new candidate tag combination set is obtained, the real-time weight, i.e., the initial real-time weight, of each tag combination in the candidate tag combination set may be a preset value, such as 0 or 1. Alternatively, for each tag combination in the candidate tag combination set, if the tag combination was also in the candidate tag combination set before performing step S210, the real-time weights for the tag combination may follow the real-time weights before re-determining the candidate tag combination set; if the tag combination is not in the candidate tag combination set before step S210 is performed, the initial value of the real-time weight of the tag combination may be a preset value.
In practical application, for a certain user, for example, a first user, the interaction behavior of the first user for each content triggered in a predetermined period of time may be counted, so as to obtain a candidate tag combination set of the first user. In the above step S210, for each of the K tag combinations, the content corresponding to the tag combination may be determined among the respective contents related to the interactive behavior within the predetermined period. And determining the statistical weight of the tag combination according to the type and the times of the interaction behavior of the content in the preset time period. Or traversing each interactive behavior in a preset period, and accumulating the statistical weights of the label combinations related to the interactive behavior in the K label combinations based on the type of the interactive behavior to obtain the final statistical weights.
Alternatively, the K tag combinations may be preconfigured. For example, K tag combinations are obtained by permutation and combination of the respective tags in a plurality of tag systems.
Alternatively, L may be preconfigured, for example, the number of tag combinations to be selected in the step S220 is predetermined, and the tag combinations are selected according to the number, for example, the L tag combinations with the largest statistical weights are selected. In practical application, if Y tag combinations with the largest real-time weights are determined according to the real-time weights of all tag combinations in the candidate tag combination set, and recommended content is determined based on the Y tag combinations, L may be related to Y. For example l=2y.
According to the embodiment shown in fig. 2, the candidate tag combination set may be accurately determined based on the interaction behavior within a predetermined period of time. For example, when the interest of the user is transferred, a candidate label combination set with high correlation with the new interest of the user can be timely captured. The method has the advantages that the number of the tag combinations in the candidate tag combination set is controlled, the calculation amount of the electronic equipment during real-time online updating is limited, the low correlation between the tag combinations which are ignored to be updated and users can be ensured, and the accuracy of recommended content is improved.
In an exemplary embodiment, in step S130, determining the recommended content in the content database according to the real-time weights of the tag combinations in the candidate tag combination set includes:
according to the real-time weight of each tag combination in the candidate tag combination set, determining the interested tag combination from the candidate tag combination set;
and determining the content corresponding to the interesting tag combination in the content database as recommended content.
For example, the Y tag combinations with the greatest real-time weights may be determined as the tag combinations of interest. Further, Y may be equal to 0.5L.
For example, for a user, such as a first user, an interesting tag combination may be determined from a candidate tag combination set according to the real-time weights of the respective tag combinations in the candidate tag combination set of the first user; and determining the content corresponding to the interesting tag combination as recommended content of the first user.
According to the embodiment, the interesting label combination of the user can be accurately determined. Corresponding recommended content is determined based on the interesting tag combinations. Because the content recommendation is performed based on the interested tag combinations, the tag combinations can more accurately represent the user preference, so that the recommended content can be determined more accurately and more closely to the user preference, and the correlation of the recommended content is improved.
In some implementations, corresponding content may be searched from the content database in real-time according to the tag combinations of interest. In some embodiments, some tag combinations, such as tag combinations of interest to more users, may be predetermined, and marked as high-frequency tag combinations, and at least one content corresponding to the high-frequency tag combinations may be screened in advance to form a content queue. And when receiving a content recommendation request, rapidly determining recommended content based on a content queue corresponding to the high-frequency tag combination. Specifically, determining the content corresponding to the interesting tag combination in the content database as the recommended content includes:
under the condition that the interesting tag combination is a high-frequency tag combination, determining a content queue corresponding to the interesting tag combination according to a predetermined mapping relation between the high-frequency tag combination and the content queue;
and selecting the content in the content queue from the content database, and determining the selected content as recommended content.
For example, at least one content queue corresponding to the at least one high-frequency tag combination may be stored in a key-value storage manner.
According to the embodiment, the content queue of the high-frequency tag combination is preset, and when a content recommendation request is received, the recommended content can be quickly selected from the content database by utilizing the mapping relation between the high-frequency tag combination and the content queue, so that the recommendation efficiency is improved.
For example, for a certain tag combination, such as an interesting tag combination or a high frequency tag combination, the process of determining its corresponding content may comprise:
determining at least one label combination of each content according to at least two labels of each content in the content database;
calculating the similarity between each content and the label combination to be matched, such as the interesting label combination or the high-frequency label combination, according to at least one of at least one label combination of each content, the heat of each content and the timeliness of each content; for example, firstly, screening out content containing a label combination to be matched according to at least one label combination of each content, and then calculating the similarity between the content and the label combination to be matched by using a similarity prediction model according to the label, the heat and the timeliness of the screened content;
and determining the content with the similarity meeting the preset condition, such as the content with the similarity larger than a certain threshold value or the P content with the maximum similarity, as the content corresponding to the label combination to be matched. Wherein P is a preset positive integer.
Thus, for each label combination, more accurate content corresponding to the label combination can be obtained.
The disclosed embodiments also provide an exemplary implementation of determining a high frequency tag combination. Specifically, the content recommendation method may further include:
Determining at least one candidate tag combination set corresponding to the at least one user identifier respectively;
at least one high frequency tag combination is determined from the at least one tag combination based on the number of occurrences of each of the at least one pre-configured tag combination in the at least one candidate set of tag combinations.
That is, different users correspond to different sets of candidate tag combinations. And (3) offline counting a candidate label combination set of each user, and if one label combination appears in a plurality of candidate label combination sets, determining that the label combination has higher correlation with more users as a high-frequency label combination.
For example, a tag combination whose number of occurrences is larger than Q may be determined as a high-frequency tag combination, that is, if one tag combination has a higher correlation with at least Q users, the tag combination may be determined as a high-frequency tag combination. Where Q may be a threshold set according to the carrying capacity of the electronic device.
According to the above embodiment, the high-frequency tag combination can be accurately determined. Therefore, the utilization rate of a predetermined content queue is improved, the requirement and the number of times of determining the content corresponding to the tag combination in real time are reduced, the processing pressure of the electronic equipment is reduced, and the recommendation efficiency is improved.
According to embodiments of the present disclosure, it is also possible to assist an administrator of the content database in supplementing the content. Specifically, the content recommendation method may further include:
for each of the at least one high frequency tag combination, adding content corresponding to the high frequency tag combination in the content database if the number of content included in the content queue corresponding to the high frequency tag combination is less than a second preset threshold.
According to the steps, if the label combination interested by most users has less relevant content in the content database, the content can be supplemented in time, so that the better operation of the content recommendation system is promoted. The knowledge management system is used for an enterprise knowledge management system, and can actively find whether the requirement for knowledge in the enterprise is met or not, and more knowledge interested by staff is introduced.
FIG. 3 shows a schematic diagram of one example of an application of the content recommendation system. The application example takes enterprise knowledge recommendation as an example, wherein the content database is a knowledge database. The content is knowledge, otherwise known as document knowledge, knowledge document. As shown in FIG. 3, a user may click on knowledge, collect, cancel collection, and other interactions at the recommendation user interface 310. The real-time interaction behavior triggers the updating of the user interest tag assembly portrayal 320. Here, the user interest tag combination portrayal may be a real-time weight of each tag combination in the user's candidate tag combination set. And the real-time interactive behavior triggers the behavior log generation and landing, the corresponding behavior log will be stored in the behavior log database 360. Using the types and amounts of various interactions of the user recorded in the behavior log database 360, the user's interest tag combination portraits may be periodically rectified offline, i.e., the user's candidate tag combination sets updated. Based on the user interest tag combination portrait 320, an interest tag combination pair 330 may be determined upon receipt of a content recommendation request. The interest tag combination pair may also be called an interest tag combination, and its related knowledge may be referred to as recommended content.
In addition, related knowledge recalls 340 may be conducted periodically. The low coverage knowledge may be supplemented, for example, when the relevant knowledge of the high frequency tag combinations is less, i.e. low knowledge coverage tag combinations 350 are present in the high frequency tag combinations. Based on the supplemental document knowledge 370, the related knowledge recall 340 is completed. And updating the relevant knowledge corresponding to the interest tag combination pair offline based on the recalled relevant knowledge.
It can be seen that according to the content recommendation method provided by the embodiment of the present disclosure, when a user performs an interaction on a target content in a content database, a tag combination of the target content is determined first, and a real-time weight of the tag combination in the candidate tag combination set is updated according to a type of the interaction and the tag combination of the target content. Because the weight used for determining the recommended content is updated in the unit of the tag combination, the tag combination can more accurately represent the user preference, and the weight is updated in real time based on the type of the interaction behavior, the recommended content can be more accurately determined closer to the user preference based on the weight, and the relevance of the recommended content is improved.
As an implementation of the above methods, the present disclosure further provides a content recommendation apparatus. Fig. 4 shows a schematic diagram of a content recommendation device provided in an embodiment of the present disclosure. As shown in fig. 4, the apparatus 400 includes:
An interaction response module 410, configured to determine M tag combinations of the target content according to N tags of the target content in response to interaction behavior with respect to the target content in the content database; wherein N is an integer greater than or equal to 2, and M is a positive integer;
the weight updating module 420 is configured to update real-time weights of tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations;
and the recommending module 430 is configured to determine recommended content in the content database according to the current real-time weight of the tag combinations in the candidate tag combination set when the content recommending request is received.
As shown in fig. 5, in one embodiment, the weight update module 420 includes:
a first updating unit 421, configured to update the real-time weight of the ith tag combination according to the type of the interaction behavior in the case where the candidate tag combination set includes the ith tag combination in the M tag combinations.
In one embodiment, the weight update module 420 includes:
the second updating unit 422 is configured to, if the candidate tag combination set does not include the jth tag combination of the M tag combinations, add the jth tag combination to the candidate tag combination set if the number of tag combinations in the candidate tag combination set is less than the first preset threshold, and determine a real-time weight of the jth tag combination according to the type of the interaction behavior.
As shown in fig. 5, in one embodiment, the apparatus further comprises:
the weight determining module 440 is configured to perform statistics on interaction behaviors triggered in a predetermined period of time for each content in the content database, so as to obtain statistical weights of K tag combinations; wherein K is a positive integer;
the combination selection module 450 is configured to select L tag combinations from the K tag combinations according to the statistical weights of the K tag combinations; wherein L is a positive integer less than or equal to K;
the first set determining module 460 is configured to obtain a candidate tag combination set based on the L tag combinations.
As shown in fig. 5, in one embodiment, recommendation module 430 includes:
a combination screening unit 431, configured to determine a tag combination of interest from the candidate tag combination set according to the real-time weight of each tag combination in the candidate tag combination set;
a content determining unit 432, configured to determine content corresponding to the interesting tag combination in the content database as recommended content.
In one embodiment, the content determining unit 432 is configured to:
under the condition that the interesting tag combination is a high-frequency tag combination, determining a content queue corresponding to the interesting tag combination according to a predetermined mapping relation between the high-frequency tag combination and the content queue;
And selecting the content in the content queue from the content database, and determining the selected content as recommended content.
As shown in fig. 5, in one embodiment, the method further includes:
a second set determining module 470, configured to determine at least one candidate tag combination set corresponding to the at least one user identifier respectively;
a high frequency determination module 480 for determining at least one high frequency tag combination from the at least one tag combination based on a number of occurrences of each of the at least one pre-configured tag combination in the at least one candidate tag combination set.
In one embodiment, the method further comprises:
the content adding module 490 is configured to add, for each of the at least one high-frequency tag combination, content corresponding to the high-frequency tag combination in the content database if the number of content included in the content queue corresponding to the high-frequency tag combination is less than a second preset threshold.
According to embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.
Fig. 6 illustrates a schematic block diagram of an example electronic device 600 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 6, the electronic device 600 includes a computing unit 601 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 602 or a computer program loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 can also be stored. The computing unit 001, ROM 602, and RAM 603 are connected to each other by a bus 604. An input output (I/O) interface 605 is also connected to bus 604.
A number of components in the electronic device 600 are connected to the I/O interface 605, including: an input unit 606 such as a keyboard, mouse, etc.; an output unit 607 such as various types of displays, speakers, and the like; a storage unit 608, such as a magnetic disk, optical disk, or the like; and a communication unit 609 such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the electronic device 600 to exchange information/data with other devices through a computer network, such as the internet, and/or various telecommunication networks.
The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the respective methods and processes described above, such as a content recommendation method. For example, in some embodiments, the content recommendation method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 600 via the ROM 602 and/or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the content recommendation method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the content recommendation method by any other suitable means (e.g. by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel, sequentially, or in a different order, provided that the desired results of the disclosed aspects are achieved, and are not limited herein.
The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.
Claims (16)
1. A content recommendation method, comprising:
in response to interaction behavior for target content in a content database, determining M label combinations of the target content according to N labels of the target content; wherein N is an integer greater than or equal to 2, and M is a positive integer;
updating the real-time weights of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations;
under the condition of receiving a content recommendation request, determining recommended content in the content database according to the current real-time weight of the tag combinations in the candidate tag combination set;
wherein updating the real-time weight of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations comprises:
If the number of the label combinations in the candidate label combination set is smaller than a first preset threshold value under the condition that the candidate label combination set does not contain the jth label combination in the M label combinations, adding the jth label combination in the candidate label combination set, and determining the real-time weight of the jth label combination according to the type of the interaction behavior; wherein j is a positive integer less than or equal to M.
2. The method of claim 1, wherein the updating the real-time weights of the tag combinations in the candidate tag combination set according to the type of interaction behavior and the M tag combinations comprises:
updating the real-time weight of the ith tag combination according to the type of the interactive behavior when the candidate tag combination set contains the ith tag combination in the M tag combinations; wherein i is a positive integer less than or equal to M.
3. The method of any of claims 1-2, further comprising:
counting the interaction behaviors triggered in a preset time period and aiming at each content in the content database to obtain the statistical weights of K label combinations; wherein K is a positive integer;
Selecting L label combinations from the K label combinations according to the statistical weights of the K label combinations; wherein L is a positive integer less than or equal to K;
and obtaining the candidate label combination set based on the L label combinations.
4. The method of any of claims 1-2, the determining recommended content in the content database according to real-time weights of tag combinations in the candidate tag combination set, comprising:
determining interesting tag combinations from the candidate tag combination set according to the real-time weights of the tag combinations in the candidate tag combination set;
and determining the content corresponding to the interesting tag combination in the content database as the recommended content.
5. The method of claim 4, wherein the determining content in the content database corresponding to the tag combination of interest as the recommended content comprises:
under the condition that the interesting tag combination is a high-frequency tag combination, determining a content queue corresponding to the interesting tag combination according to a predetermined mapping relation between the high-frequency tag combination and the content queue;
selecting the content in the content queue from the content database, and determining the selected content as the recommended content.
6. The method of claim 5, further comprising:
determining at least one candidate tag combination set corresponding to the at least one user identifier respectively;
determining the at least one high frequency tag combination from the at least one tag combination according to the occurrence number of each tag combination in the at least one candidate tag combination set, which is preconfigured.
7. The method of claim 5, further comprising:
for each high-frequency tag combination of the at least one high-frequency tag combination, adding content corresponding to the high-frequency tag combination in the content database when the number of content contained in a content queue corresponding to the high-frequency tag combination is smaller than a second preset threshold.
8. A content recommendation device, comprising:
the interaction response module is used for responding to the interaction behavior aiming at the target content in the content database, and determining M label combinations of the target content according to N labels of the target content; wherein N is an integer greater than or equal to 2, and M is a positive integer;
the weight updating module is used for updating the real-time weight of the tag combinations in the candidate tag combination set according to the type of the interaction behavior and the M tag combinations;
The recommending module is used for determining recommended content in the content database according to the current real-time weight of the tag combinations in the candidate tag combination set under the condition of receiving a content recommending request;
wherein, the weight updating module comprises:
a second updating unit, configured to, if the candidate tag combination set does not include a jth tag combination of the M tag combinations, add the jth tag combination to the candidate tag combination set if the number of tag combinations in the candidate tag combination set is smaller than a first preset threshold, and determine a real-time weight of the jth tag combination according to the type of the interaction behavior; wherein j is a positive integer less than or equal to M.
9. The apparatus of claim 8, wherein the weight update module comprises:
a first updating unit, configured to update, when the candidate tag combination set includes an ith tag combination of the M tag combinations, a real-time weight of the ith tag combination according to the type of the interaction behavior; wherein i is a positive integer less than or equal to M.
10. The apparatus of any of claims 8-9, further comprising:
The weight determining module is used for counting the interaction behaviors of each content in the content database triggered in a preset period of time to obtain the statistical weights of K label combinations; wherein K is a positive integer;
the combination selection module is used for selecting L label combinations from the K label combinations according to the statistical weights of the K label combinations; wherein L is a positive integer less than or equal to K;
and the first set determining module is used for obtaining the candidate label combination set based on the L label combinations.
11. The apparatus of any of claims 8-10, the recommendation module comprising:
a combination screening unit, configured to determine a tag combination of interest from the candidate tag combination set according to the real-time weight of each tag combination in the candidate tag combination set;
and the content determining unit is used for determining the content corresponding to the interesting label combination in the content database as the recommended content.
12. The apparatus of claim 11, wherein the content determination unit is configured to:
under the condition that the interesting tag combination is a high-frequency tag combination, determining a content queue corresponding to the interesting tag combination according to a predetermined mapping relation between the high-frequency tag combination and the content queue;
Selecting the content in the content queue from the content database, and determining the selected content as the recommended content.
13. The apparatus of claim 12, further comprising:
a second set determining module, configured to determine at least one candidate tag combination set corresponding to at least one user identifier respectively;
the high-frequency determining module is used for determining the at least one high-frequency label combination from the at least one label combination according to the occurrence times of each label combination in the at least one candidate label combination set, which is preconfigured.
14. The apparatus of claim 13, further comprising:
a content adding module, configured to add, for each high-frequency tag combination of the at least one high-frequency tag combination, content corresponding to the high-frequency tag combination in the content database when the number of content included in the content queue corresponding to the high-frequency tag combination is smaller than a second preset threshold.
15. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.
16. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110129782.9A CN112818230B (en) | 2021-01-29 | 2021-01-29 | Content recommendation method, device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110129782.9A CN112818230B (en) | 2021-01-29 | 2021-01-29 | Content recommendation method, device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112818230A CN112818230A (en) | 2021-05-18 |
| CN112818230B true CN112818230B (en) | 2023-10-20 |
Family
ID=75860489
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110129782.9A Active CN112818230B (en) | 2021-01-29 | 2021-01-29 | Content recommendation method, device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112818230B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113382279B (en) * | 2021-06-15 | 2022-11-04 | 北京百度网讯科技有限公司 | Live broadcast recommendation method, device, equipment, storage medium and computer program product |
| CN114969502B (en) * | 2021-06-21 | 2023-10-27 | 中移互联网有限公司 | Vehicle information exchange method and system and computer readable storage medium |
| CN113706253B (en) * | 2021-08-30 | 2024-10-15 | 康键信息技术（深圳）有限公司 | Real-time product recommendation method and device, electronic equipment and readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102754094A (en) * | 2009-10-29 | 2012-10-24 | 谷歌公司 | Ranking user generated web content |
| CN102880687A (en) * | 2012-09-14 | 2013-01-16 | 北京大学 | Personal interactive data retrieval method and system based on tag technology |
| CN108121770A (en) * | 2017-11-30 | 2018-06-05 | 南京南邮信息产业技术研究院有限公司 | A kind of information sorting device based on mobile terminal big data |
| US10529000B1 (en) * | 2017-02-22 | 2020-01-07 | Udemy, Inc. | System and method for automatically tagging products for an e-commerce web application and providing product recommendations |
| CN111428138A (en) * | 2020-03-26 | 2020-07-17 | 中国建设银行股份有限公司 | Course recommendation method, system, equipment and storage medium |
| US10783167B1 (en) * | 2016-08-23 | 2020-09-22 | Amazon Technologies, Inc. | System for classifying items based on user interactions |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2666336C1 (en) * | 2017-08-01 | 2018-09-06 | Общество С Ограниченной Ответственностью "Яндекс" | Method and system for recommendation of media-objects |
-
2021
- 2021-01-29 CN CN202110129782.9A patent/CN112818230B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102754094A (en) * | 2009-10-29 | 2012-10-24 | 谷歌公司 | Ranking user generated web content |
| CN102880687A (en) * | 2012-09-14 | 2013-01-16 | 北京大学 | Personal interactive data retrieval method and system based on tag technology |
| US10783167B1 (en) * | 2016-08-23 | 2020-09-22 | Amazon Technologies, Inc. | System for classifying items based on user interactions |
| US10529000B1 (en) * | 2017-02-22 | 2020-01-07 | Udemy, Inc. | System and method for automatically tagging products for an e-commerce web application and providing product recommendations |
| CN108121770A (en) * | 2017-11-30 | 2018-06-05 | 南京南邮信息产业技术研究院有限公司 | A kind of information sorting device based on mobile terminal big data |
| CN111428138A (en) * | 2020-03-26 | 2020-07-17 | 中国建设银行股份有限公司 | Course recommendation method, system, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112818230A (en) | 2021-05-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11663254B2 (en) | System and engine for seeded clustering of news events | |
| CN109684538A (en) | A kind of recommended method and recommender system based on individual subscriber feature | |
| WO2021098648A1 (en) | Text recommendation method, apparatus and device, and medium | |
| CN112818230B (en) | Content recommendation method, device, electronic equipment and storage medium | |
| CN109948121A (en) | Article similarity method for digging, system, equipment and storage medium | |
| CN103838756A (en) | Method and device for determining pushed information | |
| CN109471978B (en) | Electronic resource recommendation method and device | |
| CN113239275B (en) | Information pushing method, device, electronic equipment and storage medium | |
| CN104077407A (en) | System and method for intelligent data searching | |
| CN107609192A (en) | The supplement searching method and device of a kind of search engine | |
| CN111475725A (en) | Method, apparatus, device, and computer-readable storage medium for searching for content | |
| CN115423555A (en) | Commodity recommendation method and device, electronic equipment and storage medium | |
| CN116049379A (en) | Knowledge recommendation method, knowledge recommendation device, electronic equipment and storage medium | |
| CN110347923B (en) | Traceable fast fission type user portrait construction method | |
| CN115827956A (en) | Data information retrieval method and device, electronic equipment and storage medium | |
| CN113378015B (en) | Search method, search device, electronic apparatus, storage medium, and program product | |
| CN112528644A (en) | Entity mounting method, device, equipment and storage medium | |
| CN114491232B (en) | Information query method and device, electronic equipment and storage medium | |
| CN116304236A (en) | User portrait generation method and device, electronic equipment and storage medium | |
| CN116070028A (en) | Page-based content information pushing method, device, equipment and storage medium | |
| CN113269600B (en) | Information sending method and device | |
| CN116186198A (en) | Information retrieval method, information retrieval device, computer equipment and storage medium | |
| CN114528378A (en) | Text classification method and device, electronic equipment and storage medium | |
| CN115080824A (en) | Target word mining method and device, electronic equipment and storage medium | |
| CN113722593A (en) | Event data processing method and device, electronic equipment and 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 |