CN116720894A - Social media advertisement recommendation method based on short-term interests - Google Patents

Social media advertisement recommendation method based on short-term interests Download PDF

Info

Publication number
CN116720894A
CN116720894A CN202310716492.3A CN202310716492A CN116720894A CN 116720894 A CN116720894 A CN 116720894A CN 202310716492 A CN202310716492 A CN 202310716492A CN 116720894 A CN116720894 A CN 116720894A
Authority
CN
China
Prior art keywords
features
text
advertisement
social media
recommendation method
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.)
Pending
Application number
CN202310716492.3A
Other languages
Chinese (zh)
Inventor
王巍
魏家扬
辛国栋
刘扬
黄俊恒
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Weihai Tianzhiwei Network Space Safety Technology Co ltd
Harbin Institute of Technology Weihai
Original Assignee
Weihai Tianzhiwei Network Space Safety Technology Co ltd
Harbin Institute of Technology Weihai
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Weihai Tianzhiwei Network Space Safety Technology Co ltd, Harbin Institute of Technology Weihai filed Critical Weihai Tianzhiwei Network Space Safety Technology Co ltd
Priority to CN202310716492.3A priority Critical patent/CN116720894A/en
Publication of CN116720894A publication Critical patent/CN116720894A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0251—Targeted advertisements
    • G06Q30/0269—Targeted advertisements based on user profile or attribute
    • G06Q30/0271—Personalized advertisement
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/3331—Query processing
    • G06F16/334—Query execution
    • G06F16/3344—Query execution using natural language analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • 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
    • G06F16/9536—Search customisation based on social or collaborative filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0277—Online advertisement
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/01—Social networking
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

The application provides a social media advertisement recommendation method based on short-term interests, which extracts text features; performing similarity calculation on text features and advertisement features which can be identified by named entities in a natural language processing model, and judging whether the text features are matched with the advertisement features or not according to a preset similarity threshold; and judging whether text semantics are added by adopting images or not according to text features which cannot be identified by named entities in the natural language processing model, so as to confirm whether picture information is favorable for screening advertisements or not. The application provides a social media advertisement recommendation method based on short-term interests. The application realizes accurate popularization aiming at short-term interests of users, and the practical range comprises various comprehensive social network platforms mainly containing image-text contents and corresponding mobile PC end applications thereof, which can extract various contents with characteristics for advertisement screening; has wide application prospect.

Description

Social media advertisement recommendation method based on short-term interests
Technical Field
The application belongs to the technical field of advertisement recommendation, and particularly relates to a social media advertisement recommendation method based on short-term interests.
Background
With the development of the internet, people's leisure time is increasingly occupied by the internet, and the advertising ' main battlefield ' is also transferred from the television to the internet. Advertisement recommendation systems are very important business and research areas. Advertisement recommendation according to the big data technology is the current mainstream mode, but advertisement recommendation by the big data technology also has the same disadvantages. On the one hand, with popularization of the internet, more and more users begin to pay attention to privacy protection of network information, and the big data technology cannot extract interesting features of the users. On the other hand, users are likely to come into contact with new areas at any time and briefly generate interest in these new areas. But these briefly generated interests cannot be quickly reflected in the user's interest characteristics, resulting in personalized recommendations based on the user's past information often recommending advertising content that is not of temporary interest to the user. The promotion contents irrelevant to the current contents not only cause the waste of promotion resources, but also influence the use experience of users. The interests of the user are likely to shift within a period of time, but big data is difficult to instantaneously reflect the shift of the interests of the user for advertisement pushing. In fact, a short period of time that the user has just generated interest, often the "window period" in which the advertisement is most likely to be effective.
In social media with broad topic comprehensiveness and faster user interest and attention transfer, the advertisement recommendation system relying on the past information of the user is lower in efficiency, and the recommendation of advertisements should depend more on the short-term interests of the user.
In the prior art, patent number CN201611051650, patent name, a method and system for pushing information are also proposed, however, the method is mainly used in the video field, by capturing the key frames of people contained in the video played by the user, generating corresponding natural language scene description according to the video frames, i.e. images, and extracting the key words from the key frames for selecting and pushing advertisements. The patent can automatically provide product information with high user interest probability when the user evaluates the product information during playing. This approach has considerable problems both from a user perspective and from a technical perspective. From the perspective of a user, the main purpose of watching the video is leisure and entertainment, not television shopping, and any advertisement appearing in the process of watching the video inevitably breaks the continuity of the video watching process and causes the user to generate negative emotion, so that the purpose of popularization cannot be achieved. On the other hand, most of the video programs in the mainstream today have sponsors to put advertisements, and pushing products that are likely to be irrelevant to the put advertisements by capturing video frames on a video website is not unlike a snake-out. From a technical point of view, this method is almost impossible in practical use.
In view of the foregoing, there is a need for developing a new social media advertisement recommendation method based on short-term interests.
Disclosure of Invention
In order to achieve the above purpose, the application adopts the following technical scheme: the utility model provides a social media advertisement recommending method based on short-term interests, which is characterized in that: comprises the steps of,
step S102: extracting text characteristics;
step S103, similarity calculation is carried out on text features which can be identified by named entities in the natural language processing model and advertisement features, and whether the text features are matched with the advertisement features or not is judged according to a preset similarity threshold;
step S104: and judging whether text semantics are added by adopting images or not according to text features which cannot be identified by named entities in the natural language processing model, so as to confirm whether picture information is favorable for screening advertisements or not.
Optionally, in step S103, whether the text feature matches the advertisement feature is determined according to a preset similarity threshold; if yes, go to step S107; if not, executing step S104;
in step S104, judging whether the text semantics are added by adopting images; if yes, go to step S105; if not, step S109 is executed.
Optionally, in step S104, the unrecognizable text features include features of short text content, features of no text content, features of unrecognizable long text content, and features of picture content.
Optionally, in step S104, the natural language processing model includes a Bert model, a Chatgpt model.
Optionally, step S105: adding text semantics by adopting images to text features which cannot be identified by named entities in a natural language processing model, extracting image features by using a pre-trained image processing model, and executing a step S106;
and if the text features unrecognizable by the named entities in the natural language processing model are not added by adopting images to the text semantics, executing step S109.
Optionally, in step S105, the feature vector is obtained after the image feature processing, and the feature vector is output to step S106.
Optionally, step S106: performing similarity calculation on the image features and the advertisement features, and judging whether the image features and the advertisement features are matched or not according to a preset similarity threshold;
if yes, go to step S107; if not, step S109 is executed.
Optionally, step S109: when the text semantics are not added by adopting images in the step S104 or the image features and the advertisement features are not matched in the step S106, personalized recommendation is carried out for the user by using a big data technology depending on the past information of the user; or directly performing non-personalized recommendation, promoting the hot spot information in the platform, and executing step S107;
step S107: and outputting the corresponding advertisement.
Optionally, when the candidate advertisements are more than two, selecting corresponding advertisements as advertisements to be pushed currently according to the priority; and when the candidate advertisement is one, the candidate advertisement is used as the advertisement which is currently required to be pushed.
Optionally, in step S102, the server side receives the text content uploaded to the social media by the user, uses the pre-trained natural language processing model to identify a named entity in the text content uploaded by the user, and outputs the named entity in the form of text features.
The method for extracting text features by using named entity recognition extracts the named entities in the text as features, and can effectively match advertisements related to topic discussion objects. If the related advertisement cannot be matched or the condition that no text exists and the text is too short to cause no named entity exists, whether the image part corresponding to the text is added to the semantics of the text or not needs to be considered. If no semantic is added, the recommendation is carried out or the non-personalized recommendation is carried out according to the past information of the user. If the semantic addition exists, the image-text matching is carried out on the corresponding picture and the advertisement label through an image-text matching pre-training model (such as CLIP, LOUPE and the like). The object identification is aimed at, and the object commodity advertisement can be effectively matched. The application realizes accurate popularization aiming at short-term interests of users, and the practical range comprises various comprehensive social network platforms mainly containing image-text contents and corresponding mobile PC end applications thereof, which can extract various contents with characteristics for advertisement screening; has wide application prospect.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a short-term interest-based social media advertisement recommendation method of the present application.
Detailed Description
In order to make the technical problems, technical schemes and beneficial effects to be solved more clear, the application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The social media advertisement recommendation method based on short-term interests provided by the embodiment of the application is described. As shown in FIG. 1, in the social media advertisement recommendation method based on short-term interests, key characteristics of current interests of a user are obtained by analyzing text and picture information of a current browsed webpage of the user, and then advertisement contents matched with the key characteristics are screened, so that accurate pushing of advertisements is realized.
Specifically, the method comprises the following steps:
step S101: starting.
Step S102: text features are extracted. The server side receives text content uploaded to social media by a user, identifies named entities in the text content uploaded by the user by using a pre-trained natural language processing model, and outputs the named entities in the form of text features.
Named entities in the text content represent text features of short-term interests of users publishing and browsing the media content, which have strong directionality to the product and the promotional information of the product. The server side uses a pre-trained natural language processing model, processes and extracts characteristics of text contents issued by a user when the text contents are checked by artificial intelligence, and is convenient for screening out product information which is to be pushed by a current page according to the extracted text characteristics and matching related product advertisement characteristics. For example, when a user issues a corresponding discussion topic of a star, named entities in the natural language processing model can identify the name of the star in question by the user, thereby recommending corresponding pronouncing products or star surrounding information. When a user issues topics related to event discussion, the named entity identification can identify corresponding event names, so that corresponding video media viewing channels or event peripheral products are promoted.
Named entity recognition is largely divided into four categories: name of person, place name, organization name, miscellaneous items.
Name, including natural name, pen name, and art name;
an organization name including each organization name, such as company name, event name, etc.;
miscellaneous items include various proper nouns, such as the variety of the pet, etc.
The pretrained natural language processing model can be selected from a Bert model, a Chatgpt model and other natural language processing models with excellent performances, and can also adopt other natural language processing models in the prior art, so long as a section of natural language text is input, different results can be output according to the function type of the model. For example, the text content can be identified by using a natural language processing model, so that the named entity and the type thereof in the text content can be identified, and the named entity and the type thereof can be output.
Step S103, similarity calculation is carried out on text features which can be identified by named entities in the natural language processing model and advertisement features, and whether the text features are matched with the advertisement features or not is judged according to a preset similarity threshold; if yes, go to step S107; if not, go to step S104.
The similarity calculation can be realized by adopting the prior art. The similarity calculation can be performed by directly performing character-level similarity calculation on the text, or by using a model coding method as a feature vector and then calculating the similarity between the vectors.
The similarity threshold may be determined according to the actual needs of the respective operator and the person skilled in the art and the prevailing subscriber population of the platform.
Step S104: judging whether text semantics are added by adopting images or not according to text features which cannot be identified by named entities in a natural language processing model, thereby confirming whether picture information is favorable for screening advertisements or not; if yes, go to step S105; if not, step S109 is executed.
The unrecognizable text features include features of short text content, features of no text content, features of long text content that cannot be recognized, and features of picture content. Currently, in the use of mainstream social media, there is a large amount of short text, non-text content in the content published by individual users. For short text or text-free content, it is likely that the named entity cannot be identified in step S102 for screening the corresponding advertisement content; or even long text content, there may be instances where named entity recognition is incorrect or where no valid named entity matches the advertisement feature.
Not all of the information in the image is useful information and some images may have a negative effect on the screening of advertisements. For example, expression packages.
Text semantics are added with images that can help the text to refer to elimination. For example, the text "Hehaoshu" plus the content of the star image, the image can counteract the reference of "He" in the text, and turn to "XXX haoshu", thereby realizing the semantic addition of the text.
In the step S104, a corresponding image text matching pre-training model may be used, or the relevant image text matching model may be trained by itself using the data of the platform, and relevant technicians may select according to actual needs.
Step S105: and adding text semantic by adopting images to text features which cannot be identified by named entities in the natural language processing model, extracting image features by using the pre-trained image processing model, and executing step S106. In this step S105, the image feature is processed as a feature vector, and is output to step S106.
And if the text features unrecognizable by the named entities in the natural language processing model are not added by adopting images to the text semantics, executing step S109.
Step S105 obtains image features using the image processing model and text features of the advertisement using the natural language processing model, respectively. The image processing model and the natural language processing model must be learned together in a double-tower mode through an image text matching task in the training process, otherwise, similar features in images and texts like those adopted in the prior art cannot be encoded through the model, so that similar feature vectors are obtained, and serious deviation occurs in a final result. The step can be completed in the existing downloadable image text matching pre-training model, and the selection can be carried out according to actual needs.
Step S106: performing similarity calculation on the image features and the advertisement features, and judging whether the image features and the advertisement features are matched or not according to a preset similarity threshold; if yes, go to step S107; if not, go to step S109;
in this step S106, the similarity threshold may be determined by each operator and those skilled in the art according to the actual needs and the main user group of the platform.
Step S109: when the text semantics are not added by adopting images in the step S104 or the image features and the advertisement features are not matched in the step S106, personalized recommendation is carried out for the user by using a big data technology depending on the past information of the user; or directly performing non-personalized recommendation, promoting the hot spot information in the platform, and executing step S107;
the non-personalized recommendation comprises APP open screen advertisements, pure platform popularization information and the like.
Step S107: and outputting the corresponding advertisement. When the candidate advertisements are more than two, selecting the advertisement with the highest priority as the advertisement needing pushing currently; when the candidate advertisement is one, the candidate advertisement is used as the advertisement needing pushing currently, and the step S108 is executed;
in step S107, the priority of the advertisement may be set in advance by a related technician of the operation company according to actual needs.
Particularly contemplated priority setting factors include: matching degree/feature similarity with content, advertisement popularity/click-through rate/pay rate, advertisement spot amount, etc. The priority setting factors can be considered by each operation company and related technicians according to actual needs and the main stream user group of the platform to calculate the priority of each candidate advertisement.
All steps of the method can be completed in the process of issuing content audit by the user, the method is completely carried out at the server side, and each user participating in the topic can see the same advertisement after that, and repeated calculation is not needed. For the server side, although the auditing time of the content issued by the user is slightly improved, the calculation and time for personalized recommendation of different users by using a big data technology are greatly reduced, and the method is beneficial to saving considerable calculation cost for an operation company. For the user side, the time waiting for server side advertisement recommendation calculation when participating in the content is shortened, and the product information of the current interest topic can be obtained in real time.
Step S108: and (5) ending.
The method is mainly applied to a social media platform, and the product information related to the content is pushed for the user participating in the interaction of the content by extracting the characteristics of the content released by the user. From the perspective of the user, the user can intuitively consider that the user is interested in the corresponding content for a short time while participating in the discussion of the related content. And the content discussion on the social media platform is different from video playing, and a quite free space exists for inserting advertisements without worrying about influencing the use experience of users. From a technical perspective, with the development of the age, not only single-mode text feature extraction and image feature extraction, but also multi-mode image text matching has become possible, for example, a CLIP model developed by the OPENAI company. The image text matching model uses a large number of prompt texts to realize the matching task of images and texts during training, and achieves good effects. This technique is being adapted to match images with a large number of product promotional keywords, i.e. for realistic scenarios.
The method for extracting text features by using named entity recognition extracts the named entities in the text as features, and can effectively match advertisements related to topic discussion objects. If the related advertisement cannot be matched or the condition that no text exists and the text is too short to cause no named entity exists, whether the image part corresponding to the text is added to the semantics of the text or not needs to be considered. If no semantic is added, the recommendation is carried out or the non-personalized recommendation is carried out according to the past information of the user. If the semantic addition exists, the image-text matching is carried out on the corresponding picture and the advertisement label through an image-text matching pre-training model (such as CLIP, LOUPE and the like). The object identification is aimed at, and the object commodity advertisement can be effectively matched. The application realizes accurate popularization aiming at short-term interests of users, and the practical range comprises various comprehensive social network platforms mainly containing image-text contents and corresponding mobile PC end applications thereof, which can extract various contents with characteristics for advertisement screening; has wide application prospect.
The foregoing description of the preferred embodiments of the application is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the application.

Claims (10)

1. A social media advertisement recommendation method based on short-term interests is characterized by comprising the following steps: comprises the steps of,
step S102: extracting text characteristics;
step S103, similarity calculation is carried out on text features which can be identified by named entities in the natural language processing model and advertisement features, and whether the text features are matched with the advertisement features or not is judged according to a preset similarity threshold;
step S104: and judging whether text semantics are added by adopting images or not according to text features which cannot be identified by named entities in the natural language processing model, so as to confirm whether picture information is favorable for screening advertisements or not.
2. The short term interest based social media advertisement recommendation method as set forth in claim 1, wherein: in step S103, judging whether the text features are matched with the advertisement features according to a preset similarity threshold; if yes, go to step S107; if not, executing step S104;
in step S104, judging whether the text semantics are added by adopting images; if yes, go to step S105; if not, step S109 is executed.
3. The short term interest based social media advertisement recommendation method as set forth in claim 1, wherein: in step S104, the unrecognizable text features include features of short text content, features of no text content, features of long text content, and features of picture content.
4. The short term interest based social media advertisement recommendation method as set forth in claim 1, wherein: in step S104, the natural language processing model includes a Bert model and a Chatgpt model.
5. The short term interest based social media advertisement recommendation method as set forth in claim 2, wherein: step S105: adding text semantics by adopting images to text features which cannot be identified by named entities in a natural language processing model, extracting image features by using a pre-trained image processing model, and executing a step S106;
and if the text features unrecognizable by the named entities in the natural language processing model are not added by adopting images to the text semantics, executing step S109.
6. The short term interest based social media advertisement recommendation method as set forth in claim 5, wherein: in step S105, the image feature processing obtains a feature vector, and the feature vector is output to step S106.
7. The short term interest based social media advertisement recommendation method as set forth in claim 5, wherein: step S106: performing similarity calculation on the image features and the advertisement features, and judging whether the image features and the advertisement features are matched or not according to a preset similarity threshold;
if yes, go to step S107; if not, step S109 is executed.
8. The short term interest based social media advertisement recommendation method as set forth in claim 7, wherein: step S109: when the text semantics are not added by adopting images in the step S104 or the image features and the advertisement features are not matched in the step S106, personalized recommendation is carried out for the user by using a big data technology depending on the past information of the user; or directly performing non-personalized recommendation, promoting the hot spot information in the platform, and executing step S107;
step S107: and outputting the corresponding advertisement.
9. The short term interest based social media advertisement recommendation method as set forth in claim 8, wherein: when the candidate advertisements are more than two, selecting corresponding advertisements as advertisements to be pushed currently according to the priority; and when the candidate advertisement is one, the candidate advertisement is used as the advertisement which is currently required to be pushed.
10. The short term interest based social media advertisement recommendation method as set forth in claim 1, wherein: in step S102, the server receives the text content uploaded to the social media by the user, uses the pre-trained natural language processing model to identify the named entity in the text content uploaded by the user, and outputs the named entity in the form of text features.
CN202310716492.3A 2023-06-16 2023-06-16 Social media advertisement recommendation method based on short-term interests Pending CN116720894A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310716492.3A CN116720894A (en) 2023-06-16 2023-06-16 Social media advertisement recommendation method based on short-term interests

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310716492.3A CN116720894A (en) 2023-06-16 2023-06-16 Social media advertisement recommendation method based on short-term interests

Publications (1)

Publication Number Publication Date
CN116720894A true CN116720894A (en) 2023-09-08

Family

ID=87874840

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310716492.3A Pending CN116720894A (en) 2023-06-16 2023-06-16 Social media advertisement recommendation method based on short-term interests

Country Status (1)

Country Link
CN (1) CN116720894A (en)

Similar Documents

Publication Publication Date Title
CN110020437B (en) Emotion analysis and visualization method combining video and barrage
CN110557659B (en) Video recommendation method and device, server and storage medium
US20220237222A1 (en) Information determining method and apparatus, computer device, and storage medium
CN109697239B (en) Method for generating teletext information
KR20160055930A (en) Systems and methods for actively composing content for use in continuous social communication
CN112733654B (en) Method and device for splitting video
CN113010701A (en) Video-centered fused media content recommendation method and device
CN106503907B (en) Service evaluation information determination method and server
CN110569502A (en) Method and device for identifying forbidden slogans, computer equipment and storage medium
CN109460503B (en) Answer input method, answer input device, storage medium and electronic equipment
CN110929007A (en) Electric power marketing knowledge system platform and application method
CN110598095A (en) Method, device and storage medium for identifying article containing designated information
CN111782793A (en) Intelligent customer service processing method, system and equipment
CN110399473B (en) Method and device for determining answers to user questions
CN113111264B (en) Interface content display method and device, electronic equipment and storage medium
CN114186074A (en) Video search word recommendation method and device, electronic equipment and storage medium
CN114881685A (en) Advertisement delivery method, device, electronic device and storage medium
CN110516086B (en) Method for automatically acquiring movie label based on deep neural network
CN110827063A (en) Multi-strategy fused commodity recommendation method, device, terminal and storage medium
Madhok et al. Semantic understanding for contextual in-video advertising
CN116720894A (en) Social media advertisement recommendation method based on short-term interests
CN113077295B (en) Advertisement graded delivery method based on user terminal, user terminal and storage medium
CN112804580B (en) Video dotting method and device
Martina et al. A Virtual Assistant for the Movie Domain Exploiting Natural Language Preference Elicitation Strategies
CN117009577A (en) Video data processing method, device, equipment and readable 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