CN117808008A - LTV (Low temperature Co-fired ceramic) estimated inspection method - Google Patents
LTV (Low temperature Co-fired ceramic) estimated inspection method Download PDFInfo
- Publication number
- CN117808008A CN117808008A CN202410226742.XA CN202410226742A CN117808008A CN 117808008 A CN117808008 A CN 117808008A CN 202410226742 A CN202410226742 A CN 202410226742A CN 117808008 A CN117808008 A CN 117808008A
- Authority
- CN
- China
- Prior art keywords
- semantic
- ltv
- marketing
- model
- creative content
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000007689 inspection Methods 0.000 title claims abstract description 18
- 239000000919 ceramic Substances 0.000 title description 3
- 238000013135 deep learning Methods 0.000 claims abstract description 7
- 230000003993 interaction Effects 0.000 claims description 19
- 238000006243 chemical reaction Methods 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 15
- 238000012512 characterization method Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 238000000844 transformation Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 4
- 239000002699 waste material Substances 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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
- G06Q30/0242—Determining effectiveness of advertisements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/18—Extraction of features or characteristics of the image
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Accounting & Taxation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Multimedia (AREA)
- Entrepreneurship & Innovation (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an LTV pre-estimated inspection method, which comprises the following steps: constructing a semantic question-answering model based on deep learning; constructing an LTV estimation model; calculating an LTV predicted value based on the semantic features and the scene data, and generating a patrol result; and judging the effectiveness of the marketing content based on the inspection result and generating a proposal scheme. According to the invention, the marketing effect inspection is performed based on the LTV estimation, the LTV estimated value is calculated according to the current marketing creative content, so that whether the marketing creative content matches with the marketing scene is judged, scientific and reasonable reference is provided for marketing operation, the marketing creative content delivery strategy can be adjusted in time, and the waste of time cost and popularization cost is avoided.
Description
Technical Field
The invention relates to the field of artificial intelligence technology and data analysis, in particular to an LTV (Low temperature Co-fired ceramic) predictive inspection method.
Background
At present, the marketing process of enterprise products mainly depends on manual design of marketing schemes by operators, and marketing effects depend on experiences of the operators more, but lack of finer data analysis support and scientific prejudgment. For enterprises, the manual configuration cost in the marketing process is high, and the operation effect is excessively dependent on the experience judgment of operators. Marketing creative content that imparts marketing or advertising functionality plays an important role in the marketing process. LTV is understood to be the sum of all economic benefits an enterprise receives from all interactions of users, i.e., all profits an entire user life cycle brings to the enterprise. LTV is commonly used in the marketing field to measure the value of marketing actions or marketing creative content to an enterprise, and is also defined as an important reference indicator of whether an enterprise can achieve high profits. How to generate matching marketing creative content for use by businesses or merchants is a problem to be solved.
Disclosure of Invention
In order to solve the above problems, the present invention provides an LTV estimation inspection method.
An LTV estimated inspection method comprises the following steps:
s1, constructing a semantic question-answering model based on deep learning, and performing learning training by using the semantic question-answering model to obtain semantic features corresponding to image areas and characters, wherein the semantic question-answering model comprises an autocorrelation attention module and a cross-mode interaction attention module, and the cross-mode interaction attention module comprises an image-guided character self-attention model and a character-guided image self-attention model;
s2, constructing an LTV pre-estimation model, acquiring released marketing creative content data and scene data, inputting the marketing creative content data into the semantic question-answer model to obtain a semantic feature data set, and training and learning to obtain the LTV pre-estimation model by combining the scene data aiming at a single semantic feature or a combination of a plurality of semantic features;
s3, calculating an LTV predicted value based on semantic features and scene data, wherein the method specifically comprises the following steps:
s31, inputting the marketing creative content which is put in at present, and acquiring scene data corresponding to the marketing creative content which is put in at present;
s32, extracting a group of semantic features of the currently released marketing creative content by using a semantic question-answer model;
s33, calculating an individual characteristic LTV predicted value by using an LTV prediction model aiming at a single semantic characteristic or a combination of a plurality of semantic characteristics;
s34, calculating an LTV predicted value corresponding to the currently released marketing creative content, and generating a patrol result;
s4, judging the effectiveness of marketing creative content and generating a proposal scheme based on the inspection result, wherein the proposal scheme specifically comprises the following steps:
s41, judging whether the LTV predicted value is lower than a preset value threshold, if so, determining that the currently released marketing creative content meets the marketing requirement, and if so, executing a step S42;
s42, eliminating semantic features with individual feature LTV predicted values lower than a preset threshold value according to the group of semantic features extracted in the step S32, and generating a group of new marketing creative content by using a content generation method based on the reserved semantic features.
Preferably, the step S1 specifically includes the following substeps:
s11, acquiring marketing creative content sample data corresponding to the industry field, wherein the marketing creative content sample data at least comprises characters and images;
s12, constructing a semantic question-answering model based on deep learning, and performing learning training by using the semantic question-answering model to obtain semantic features corresponding to the image areas and the characters.
Preferably, the step S12 specifically includes the following substeps:
s121, respectively carrying out feature extraction on text information and image information by using an autocorrelation attention module, carrying out feature extraction on the text information by using an embedded coding method to obtain character features, obtaining image region features by using a feature extraction network through feature extraction of the image information, capturing semantic autocorrelation by using features between image regions and features between characters through autocorrelation learning, and carrying out feature update;
s122, learning semantic association between the image and the text by using a cross-modal interaction attention module and updating the characteristics, wherein the cross-modal interaction attention module is an interaction guiding model and is used for realizing the semantic association of the image and the text;
s123, overlapping the attention layers by adopting a cascading method, and carrying out multistage updating on the image region characteristics and the character characteristics to obtain the semantic characteristics with clear characterization corresponding to the image region and the character.
Preferably, the step S4 specifically further includes:
s43, judging the accuracy of the marketing creative content generated in the step S42 by utilizing a result voting method.
Preferably, the content of the marketing creative put in the current time is a picture, a text or a combination of pictures and texts, and the scene data at least comprises one or more of a put period, a put channel, a put cost, an exposure quantity, a click quantity, a conversion quantity and a conversion cost.
Preferably, the LTV estimation model is specifically:
，/>the method comprises the steps of carrying out a first treatment on the surface of the Wherein LTV is a predicted value, i is the i th day in the delivery period, M is the total number of days in the delivery period, k is the kth semantic feature, N is the total number of semantic features, A k B is the click number corresponding to the kth semantic feature k For the number of transformations corresponding to the kth semantic feature, C k Exposure number corresponding to kth semantic feature, < +.>For the decay function +.>For the conversion cost on day i, p is a preset correction constant, 0 < p < 0.5M, q is a preset attenuation parameter, 0 < q < 1, < ->For click value weight, ++>To transform the value weights.
Preferably, the question-answering model comprises an autocorrelation attention module and a cross-modal interaction attention module, wherein the cross-modal interaction attention module comprises an image-guided text self-attention model and a text-guided image self-attention model.
After the technical scheme is adopted, compared with the background technology, the invention has the following advantages:
according to the invention, the marketing effect inspection is performed based on the LTV estimation, the LTV estimated value is calculated according to the current marketing creative content, so that whether the marketing creative content matches with the marketing scene is judged, scientific and reasonable reference is provided for marketing operation, the marketing creative content delivery strategy can be adjusted in time, and the waste of time cost and popularization cost is avoided.
Drawings
FIG. 1 is a schematic flow chart of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. 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 invention.
Examples
Referring to fig. 1, the invention discloses an LTV estimation inspection method, which comprises the following steps:
s1, constructing a semantic question-answering model based on deep learning.
S11, acquiring marketing creative content sample data corresponding to the industry field, wherein the marketing creative content sample data at least comprises characters and images;
s12, constructing a semantic question-answering model based on deep learning, and performing learning training by using the semantic question-answering model to obtain semantic features corresponding to the image areas and the characters. The question-answering model comprises an autocorrelation attention module and a cross-modal interaction attention module, wherein the cross-modal interaction attention module comprises an image-guided text self-attention model and a text-guided image self-attention model. The method comprises the following steps:
the method comprises the steps of respectively carrying out feature extraction on text information and image information by using an autocorrelation attention module, carrying out characterization on the feature extraction of the text information by using an embedded coding method to obtain character features, obtaining image region features by using a feature extraction network through feature extraction of the image information, capturing semantic autocorrelation by using features between image regions and features between characters through autocorrelation learning, and carrying out feature updating. And (3) giving text information Q, performing characterization by using a word embedding coding method, and learning and extracting word vector characteristics of words. For each given text message Q, a text feature Y is obtained, expressed as:wherein t is j For the j-th word feature vector, m is the total number of region features.
Given image information T, training is carried out by utilizing a feature extraction network, and image region features X are obtained and expressed as:wherein r is i For the ith region feature, n is the total number of region features.
In this embodiment, the feature extraction of the text information adopts a gating loop unit mode, and the feature extraction of the image information uses a fast R-CNN network.
And learning semantic association between the image and the text and updating the characteristics by using a cross-modal interaction attention module, wherein the cross-modal interaction attention module is an interaction guiding model and is used for realizing the semantic association of the image and the text. Based on the image area characteristic X and the character characteristic Y, an attention weight matrix of the image and the character is obtained, and an autocorrelation relation is established between the two modes of the image and the character. And taking the image area characteristic X as the input of the image-guided text self-attention model, and calculating the inner product of the image area characteristic and the text characteristic. And taking the character feature Y as the input of a character guiding image self-attention model, and calculating the inner product of the character feature and the image area feature. Normalizing the obtained inner product result and obtaining two bi-directional cross-modal weight matrixes between the image and the text. And weighting and updating the character features and the image area features by using the obtained cross-modal weight matrix.
And overlapping the attention layers by adopting a cascading method, and carrying out multistage updating on the image region characteristics and the character characteristics to obtain the semantic characteristics with clear characterization corresponding to the image region and the character.
S2, constructing an LTV estimation model. LTV estimation refers to the results of marketing creative content delivery actions over a period of time in the future.
S21, acquiring the released marketing creative content data and scene data. The released marketing creative content data is pictures, characters or a combination of pictures and texts, and the scene data at least comprises one or more of release period, release channel, release cost, exposure quantity, click quantity, conversion quantity and conversion cost. The scene data related to the invention refer to data related to marketing creative content delivery of a business owner or a merchant, wherein a delivery period refers to total delivery duration, a delivery channel refers to advertisement channels corresponding to different accurate user groups, a delivery cost refers to consumed advertisement cost, and a conversion cost refers to average advertisement cost corresponding to single successful conversion. Therefore, the relationship between the marketing creative content and the click number, the conversion number and the conversion cost can be estimated through the LTV estimation in the follow-up process, and the future release value of the marketing creative content can be predicted.
S22, inputting marketing creative content data into a semantic question-answer model to obtain a semantic feature data set;
s23, training and learning to obtain an LTV estimation model by combining scene data aiming at single semantic features or combination of multiple semantic features. The scene data related to the LTV pre-estimated model is required to be normalized and then input into the model for training, and the LTV pre-estimated model can be a summation model based on weight assignment or a multidimensional matrix model. In this embodiment, the LTV estimation model is specifically:
，/>the method comprises the steps of carrying out a first treatment on the surface of the Wherein LTV is a predicted value, i is the i th day in the delivery period, M is the total number of days in the delivery period, k is the kth semantic feature, N is the total number of semantic features, A k B is the click number corresponding to the kth semantic feature k For the number of transformations corresponding to the kth semantic feature, C k Exposure number corresponding to kth semantic feature, < +.>For the decay function +.>For the conversion cost on day i, p is a preset correction constant, 0 < p < 0.5M, q is a preset attenuation parameter, 0 < q < 1, < ->For click value weight, ++>To transform the value weights.
And S3, calculating an LTV predicted value based on the semantic features and the scene data and generating a patrol result.
S31, inputting the marketing creative content which is put down currently, and acquiring scene data corresponding to the marketing creative content which is put down currently. When the marketing creative content is a picture, a text or a picture-text combination, the scene data at least comprises one or more of a delivery period, a delivery channel, a delivery cost, an exposure quantity, a click quantity, a conversion quantity and a conversion cost.
S32, extracting a group of semantic features of the marketing creative content which is put in by using a semantic question-answer model.
S33, calculating an individual characteristic LTV predicted value by using an LTV prediction model aiming at a single semantic characteristic or a combination of a plurality of semantic characteristics.
S34, calculating an LTV predicted value corresponding to the currently released marketing creative content, and generating a patrol result.
And S4, judging the effectiveness of the marketing creative content based on the inspection result and generating a proposal scheme.
S41, judging whether the LTV predicted value is lower than a preset value threshold, if so, determining that the currently released marketing creative content meets the marketing requirement, and if so, executing the step S42.
S42, eliminating semantic features with individual feature LTV predicted values lower than a preset threshold value according to the group of semantic features extracted in the step S32, and generating a group of new marketing creative content by using a content generation method based on the reserved semantic features.
S43, judging the accuracy of the marketing creative content generated in the step S42 by utilizing a result voting method.
The present invention is not limited to the above-mentioned embodiments, and any changes or substitutions that can be easily understood by those skilled in the art within the technical scope of the present invention are intended to be included in the scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.
Claims (7)
1. The LTV estimated inspection method is characterized by comprising the following steps of:
s1, constructing a semantic question-answering model based on deep learning, and performing learning training by using the semantic question-answering model to obtain semantic features corresponding to image areas and characters, wherein the semantic question-answering model comprises an autocorrelation attention module and a cross-mode interaction attention module, and the cross-mode interaction attention module comprises an image-guided character self-attention model and a character-guided image self-attention model;
s2, constructing an LTV pre-estimation model, acquiring released marketing creative content data and scene data, inputting the marketing creative content data into the semantic question-answer model to obtain a semantic feature data set, and training and learning to obtain the LTV pre-estimation model by combining the scene data aiming at a single semantic feature or a combination of a plurality of semantic features;
s3, calculating an LTV predicted value based on semantic features and scene data, wherein the method specifically comprises the following steps:
s31, inputting the marketing creative content which is put in at present, and acquiring scene data corresponding to the marketing creative content which is put in at present;
s32, extracting a group of semantic features of the currently released marketing creative content by using a semantic question-answer model;
s33, calculating an individual characteristic LTV predicted value by using an LTV prediction model aiming at a single semantic characteristic or a combination of a plurality of semantic characteristics;
s34, calculating an LTV predicted value corresponding to the currently released marketing creative content, and generating a patrol result;
s4, judging the effectiveness of marketing creative content and generating a proposal scheme based on the inspection result, wherein the proposal scheme specifically comprises the following steps:
s41, judging whether the LTV predicted value is lower than a preset value threshold, if so, determining that the currently released marketing creative content meets the marketing requirement, and if so, executing a step S42;
s42, eliminating semantic features with individual feature LTV predicted values lower than a preset threshold value according to the group of semantic features extracted in the step S32, and generating a group of new marketing creative content by using a content generation method based on the reserved semantic features.
2. The LTV estimation routing inspection method as set forth in claim 1, wherein the step S1 specifically includes the following sub-steps:
s11, acquiring marketing creative content sample data corresponding to the industry field, wherein the marketing creative content sample data at least comprises characters and images;
s12, constructing a semantic question-answering model based on deep learning, and performing learning training by using the semantic question-answering model to obtain semantic features corresponding to the image areas and the characters.
3. The LTV estimation routing inspection method as set forth in claim 2, wherein the step S12 specifically includes the following sub-steps:
s121, respectively carrying out feature extraction on text information and image information by using an autocorrelation attention module, carrying out feature extraction on the text information by using an embedded coding method to obtain character features, obtaining image region features by using a feature extraction network through feature extraction of the image information, capturing semantic autocorrelation by using features between image regions and features between characters through autocorrelation learning, and carrying out feature update;
s122, learning semantic association between the image and the text by using a cross-modal interaction attention module and updating the characteristics, wherein the cross-modal interaction attention module is an interaction guiding model and is used for realizing the semantic association of the image and the text;
s123, overlapping the attention layers by adopting a cascading method, and carrying out multistage updating on the image region characteristics and the character characteristics to obtain the semantic characteristics with clear characterization corresponding to the image region and the character.
4. The LTV estimation routing inspection method as set forth in claim 1, wherein the step S4 further comprises:
s43, judging the accuracy of the marketing creative content generated in the step S42 by utilizing a result voting method.
5. The LTV predictive routing method as set forth in claim 1, wherein: the content of the marketing creative put in the current time is pictures, characters or a combination of pictures and texts, and the scene data at least comprises one or more of a put period, a put channel, a put cost, an exposure quantity, a click quantity, a conversion quantity and a conversion cost.
6. The LTV estimation routing inspection method of claim 5, wherein: the LTV estimation model specifically comprises:
，/>the method comprises the steps of carrying out a first treatment on the surface of the Wherein LTV is a predicted value, i is the i th day in the delivery period, M is the total number of days in the delivery period, k is the kth semantic feature, N is the total number of semantic features, A k B is the click number corresponding to the kth semantic feature k For the number of transformations corresponding to the kth semantic feature, C k Exposure number corresponding to kth semantic feature, < +.>For the decay function +.>For the conversion cost on day i, p is a preset correction constant, 0 < p < 0.5M, q is a preset attenuation parameter, 0 < q < 1, < ->For click value weight, ++>To transform the value weights.
7. The LTV predictive routing method as set forth in claim 1, wherein: the question-answering model includes an autocorrelation attention module and a cross-modal interaction attention module including an image-guided text self-attention model and a text-guided image self-attention model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410226742.XA CN117808008B (en) | 2024-02-29 | 2024-02-29 | LTV (Low temperature Co-fired ceramic) estimated inspection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410226742.XA CN117808008B (en) | 2024-02-29 | 2024-02-29 | LTV (Low temperature Co-fired ceramic) estimated inspection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117808008A true CN117808008A (en) | 2024-04-02 |
| CN117808008B CN117808008B (en) | 2024-07-26 |
Family
ID=90425871
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410226742.XA Active CN117808008B (en) | 2024-02-29 | 2024-02-29 | LTV (Low temperature Co-fired ceramic) estimated inspection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117808008B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140075004A1 (en) * | 2012-08-29 | 2014-03-13 | Dennis A. Van Dusen | System And Method For Fuzzy Concept Mapping, Voting Ontology Crowd Sourcing, And Technology Prediction |
| CN113792177A (en) * | 2021-08-05 | 2021-12-14 | 杭州电子科技大学 | Scene character visual question-answering method based on knowledge-guided deep attention network |
| CN116663664A (en) * | 2023-05-25 | 2023-08-29 | 沈阳林科信息技术有限公司 | Customer marketing scene data analysis system and method based on NLP algorithm |
| CN116843376A (en) * | 2023-07-07 | 2023-10-03 | 广州云智达创科技有限公司 | Marketing effect prejudging method, marketing effect prejudging device, storage medium and marketing effect prejudging equipment |
| CN117132328A (en) * | 2023-09-12 | 2023-11-28 | 广州商研网络科技有限公司 | Advertisement putting control method and device, equipment and medium thereof |
-
2024
- 2024-02-29 CN CN202410226742.XA patent/CN117808008B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140075004A1 (en) * | 2012-08-29 | 2014-03-13 | Dennis A. Van Dusen | System And Method For Fuzzy Concept Mapping, Voting Ontology Crowd Sourcing, And Technology Prediction |
| CN113792177A (en) * | 2021-08-05 | 2021-12-14 | 杭州电子科技大学 | Scene character visual question-answering method based on knowledge-guided deep attention network |
| CN116663664A (en) * | 2023-05-25 | 2023-08-29 | 沈阳林科信息技术有限公司 | Customer marketing scene data analysis system and method based on NLP algorithm |
| CN116843376A (en) * | 2023-07-07 | 2023-10-03 | 广州云智达创科技有限公司 | Marketing effect prejudging method, marketing effect prejudging device, storage medium and marketing effect prejudging equipment |
| CN117132328A (en) * | 2023-09-12 | 2023-11-28 | 广州商研网络科技有限公司 | Advertisement putting control method and device, equipment and medium thereof |
Non-Patent Citations (1)
| Title |
|---|
| 曾金;陆伟;丁恒;陈海华;: "基于图像语义的用户兴趣建模", 数据分析与知识发现, no. 04, 25 April 2017 (2017-04-25), pages 80 - 87 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117808008B (en) | 2024-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10943497B2 (en) | Personalized e-learning using a deep-learning-based knowledge tracing and hint-taking propensity model | |
| CN110598016B (en) | Method, device, equipment and medium for recommending multimedia information | |
| CN110751286B (en) | Training method and training system for neural network model | |
| US11080376B2 (en) | Anonymous cross-device, cross-channel, and cross-venue user identification using adaptive deep learning | |
| CN113015010B (en) | Push parameter determination method, device, equipment and computer readable storage medium | |
| CN110727864B (en) | User portrait method based on mobile phone App installation list | |
| CN112925911B (en) | Complaint classification method based on multi-modal data and related equipment thereof | |
| CN112053188A (en) | Internet advertisement recommendation method based on hybrid deep neural network model | |
| US20230139824A1 (en) | Media content enhancement based on user feedback of multiple variations | |
| CN117132330B (en) | Intelligent advertisement method, device, equipment and storage medium based on big data | |
| CN112633690A (en) | Service personnel information distribution method, service personnel information distribution device, computer equipment and storage medium | |
| CN112995690A (en) | Live content item identification method and device, electronic equipment and readable storage medium | |
| CN116468460A (en) | Consumer finance customer image recognition system and method based on artificial intelligence | |
| CN115018190A (en) | Overdue behavior prediction method and device, storage medium and electronic device | |
| CN114547466A (en) | Information diversity recommendation method based on deep reinforcement learning | |
| US20240046922A1 (en) | Systems and methods for dynamically updating machine learning models that provide conversational responses | |
| CN113836390B (en) | Resource recommendation method, device, computer equipment and storage medium | |
| CN115730125A (en) | Object identification method and device, computer equipment and storage medium | |
| CN117808008B (en) | LTV (Low temperature Co-fired ceramic) estimated inspection method | |
| CN113449103A (en) | Bank transaction flow classification method and system integrating label and text interaction mechanism | |
| CN116628332A (en) | Personalized recommendation method based on prompt learning | |
| CN110991711A (en) | Multi-factor perception terminal switching prediction method based on deep neural network | |
| CN112435079B (en) | Advertisement recommendation method oriented to pure social platform | |
| CN113806622A (en) | Recommendation method, device and equipment | |
| CN115883260B (en) | Digital collection traceability system based on steganography technology |
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 |