User intelligent analysis method and device based on multi-source data and electronic equipment

Abstract

The invention discloses a user intelligent analysis method and device based on multi-source data and electronic equipment, and relates to the field of user analysis, and the user intelligent analysis method comprises the following steps: 1, deploying an image collection part and an information release part, and registering authorized interaction point data; 2, obtaining the attention state of a certain user in the image acquisition area for the released information, performing quantitative processing, and obtaining the attention coefficient of the user for the released information; 3, calling historical interaction data of the user from the database, analyzing past consumption behaviors of the user, preferences of watching and reading the released information and interaction behaviors of browsing interaction points, and obtaining historical behavior coefficients; the real-time data collected by the fusion image, the user historical behavior data and the group preference data are obtained for analysis and recommendation, the recommendation content better meeting the instant demand of the user is generated, the comprehensiveness and accuracy of recommendation are improved in a multi-source data fusion mode, and the recommendation result is more personalized and intelligent.

Description

Technical Field

[0001] The present invention relates to the technical field of user analysis, and in particular to a user intelligent analysis method, device and electronic equipment based on multi-source data. Background Art

[0002] In the era of big data, using big data technology for user intelligent analysis has become the choice of all walks of life. User intelligent analysis is also one of the most important application scenarios in the current field of artificial intelligence. The analysis result is a visual display of user data, which briefly and vividly describes the characteristics of a user with a series of user information. Therefore, only by fully and accurately exploring user needs and correctly grasping the iteration direction can the vitality and competitiveness of merchants be fully demonstrated;

[0003] However, the existing user intelligent analysis solutions can still be improved in some scenarios, such as:

[0004] 1. Traditional user-based intelligent analysis mainly relies on historical user behavior data and lacks measures to capture users' current interests in real time, resulting in the recommendation content not matching users' immediate needs. In addition, data from a single source is difficult to cope with the dynamic changes in user interests and cannot integrate multiple data sources, resulting in the recommendation results being a simple continuation of past behaviors and lacking innovation.

[0005] 2. The feedback mechanism of intelligent analysis is not flexible enough, and it is difficult to respond to users' actual operations and preference changes in a timely manner. It is easy to repeatedly push information that users have already interacted with or are not interested in, resulting in a decline in user experience and obvious information redundancy. Summary of the invention

[0006] 1. Technical issues to be solved

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides a user intelligent analysis method, device and electronic device based on multi-source data, which can effectively solve the problems of the prior art.

[0008] (II) Technical solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0010] The present invention discloses a user intelligent analysis method based on multi-source data, comprising the following steps:

[0011] Step 1: Deploy image acquisition components and information delivery components, and register authorized interaction point data;

[0012] Step 2: Quantify the attention state of a user in the image acquisition area to the information being delivered, and obtain the user's attention coefficient to the information being delivered;

[0013] Step 3: retrieve the user's historical interaction data from the database, analyze the user's past consumption behavior, preference for viewing delivery information, and interaction behavior at browsing interaction points, and obtain the historical behavior coefficient;

[0014] Step 4: Obtain the historical interaction data of all user groups in the current image acquisition area, and calculate the group interaction preference trend coefficient through big data analysis;

[0015] Step 5: Using the group interaction trend coefficient as a reference for weight distribution of several interaction points, comprehensively analyze the user's historical behavior coefficient, the coefficient of attention to the current delivery information, and the group interaction preference trend coefficient, output the recommended label, and push information bidirectionally between the interaction point corresponding to the recommended label and the user;

[0016] Step 6: Analyze the user's interaction operations and results in the current cycle in real time, and eliminate similar interaction points that have been completed.

[0017] Furthermore, in the quantitative processing of step 2, eye tracking and facial expression technology are combined to analyze whether the user pays attention to a certain delivery information, and after confirming that the user pays attention to the certain delivery information, it is determined whether the stay time reaches or exceeds a preset threshold. If it is determined to be yes, the result is used as a quantitative reference.

[0018] Furthermore, if there is no newly added interaction point information in the historical interaction data of the user group in step 4, historical information of similar interaction points is collected through the Internet, and several samples are extracted for prediction through a deep learning algorithm to obtain the predicted data as the group interaction preference trend coefficient of the newly added interaction point.

[0019] Furthermore, in the process of two-way push of information between the interaction point corresponding to the recommended tag and the user in step 5, if either party refuses the push, the push will be terminated, and the push between the user and the interaction point with the same attribute or the push between the interaction point and the user with the same attribute will be eliminated in the current cycle and reset in the next cycle.

[0020] Furthermore, in the process of real-time analysis of the user's interactive operations and results in the current cycle, step 6 adjusts the weight ratio of the user's behavior coefficient and the attention coefficient of the delivery information through data feedback from the interaction points and image acquisition devices, and re-outputs the recommended tags in a preset cycle.

[0021] Furthermore, in the process of removing similar interaction points of completed interaction points in step 6, if it is detected that the user has completed the target operation of a certain interaction point, the interaction point and the interaction points whose similarity meets the preset threshold will be removed from the recommendation pool of the current cycle.

[0022] The user intelligent analysis device based on multi-source data includes:

[0023] The management module is used to control the start and stop of each functional module and issue control permissions;

[0024] Data storage end, used for cloud storage of collected and analyzed data, and supports reading of local storage media;

[0025] An image acquisition module is used to acquire monitoring images and deliver information in a designated area, and convert the acquired images into character data;

[0026] The attention analysis module is used to analyze the image character data and obtain the attention coefficient of the target user to one or more interaction points;

[0027] The historical behavior analysis module is used to analyze the user's past purchase records, content preferences, and usage records of interaction points, calculate the user's individual preferences, and output the historical behavior coefficient;

[0028] The group trend analysis module is used to analyze all user historical data in the current area to obtain the actual group preference trend. For newly added interaction points, the module collects Internet data and uses deep learning algorithms to obtain the predicted group preference trend. The actual group preference trend and the predicted group preference trend are comprehensively analyzed, and each interaction point is assigned a corresponding weight to output the final group preference trend.

[0029] The tag generation module is used to comprehensively analyze the user's historical behavior coefficient, current attention coefficient, and group interaction preference trend coefficient to generate recommended tags;

[0030] The push module is used to push content in both directions based on the tag content. During the push process, if the user or interaction point refuses to push, the combination will be excluded in the current cycle and reset in the next cycle.

[0031] The dynamic adjustment module is used to analyze the user's current operations and results, and to adjust the dynamic coefficients. At the end of each cycle, the recommendation tags are regenerated to remove the interaction points of completed interaction operations and their similar interaction points.

[0032] Furthermore, the management module is interactively connected to the data storage end, the image acquisition module and the dynamic adjustment module through a wireless network, the image acquisition module is interactively connected to the attention analysis module through a wireless network, the dynamic adjustment module is interactively connected to the attention analysis module and the historical behavior analysis module through a wireless network, and the label generation module is interactively connected to the attention analysis module, the historical behavior analysis module, the group trend analysis module and the push module through a wireless network.

[0033] An electronic device comprises: one or more processors; a memory for storing one or more computer-executable instructions; when the one or more computer-executable instructions are executed by the one or more processors, the one or more processors implement a user intelligent analysis method based on multi-source data.

[0034] A computer-readable storage medium, wherein the computer-executable instructions, when executed by a processor, implement a user intelligent analysis method based on multi-source data.

[0035] (III) Beneficial effects

[0036] Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects:

[0037] 1. By acquiring real-time data collected by fusion images, user historical behavior data and group preference data for analysis and recommendation, the user's current interest status can be quantified in real time, and the user's real-time attention to the delivered information can be obtained, so as to generate recommended content that better meets the user's immediate needs. By combining the user's historical behavior, current attention and group trends, the user's interest changes can be analyzed in multiple dimensions, and the comprehensiveness and accuracy of the recommendation can be improved. The comprehensiveness and accuracy of the recommendation are improved by fusing multi-source data, making the recommendation results more personalized and intelligent.

[0038] 2. By analyzing the user's interaction results in real time, we can eliminate completed interaction points and their similar items that are irrelevant to the user's current needs to avoid repeated pushes. When one party refuses to push, the push within this cycle will be automatically terminated, thereby optimizing the user's push experience.

[0039] 3. Through real-time feedback from image acquisition devices and interaction points, the recommendation strategy can be adjusted dynamically. Based on the user's immediate behavior and interaction results, the recommendation tags can be adjusted in a timely manner so that the user's latest interests can be quickly reflected in the recommended content.

[0040] 4. Obtain the interaction preferences of user groups through big data analysis, and predict the group's preference trend coefficient based on newly added interaction points through deep learning algorithms, so that the analysis solution is also highly adaptable to new data points. While being able to consider individual historical behaviors, it can also make comprehensive recommendations based on group trends, which is particularly suitable for personalized recommendation needs in public scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 It is a flowchart of the user intelligent analysis method based on multi-source data in the present invention;

[0043] Figure 2 It is a schematic diagram of the framework of the user intelligent analysis device based on multi-source data in the present invention;

[0044] Figure 3 Schematic diagram of the architecture of the user intelligent analysis method based on multi-source data in the present invention.

[0045] The numbers in the figure represent: 1. Management module; 2. Data storage terminal; 3. Image acquisition module; 4. Attention analysis module; 5. Historical behavior analysis module; 6. Group trend analysis module; 7. Label generation module; 8. Push module; 9. Dynamic adjustment module. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0047] The present invention will be further described below in conjunction with the embodiments.

[0048] Example

[0049] The user intelligent analysis method based on multi-source data in this embodiment is as follows: Figure 1 and Figure 3 As shown, the following steps are included:

[0050] Step 1: Deploy image acquisition components (such as cameras) and information delivery components (such as display screens and smart billboards) to register authorized interaction point data;

[0051] Step 2: Quantify the attention state of a user in the image acquisition area to the information being delivered, and obtain the user's attention coefficient to the information being delivered;

[0052] During the quantitative processing, eye tracking and facial expression technology are combined to analyze whether the user pays attention to a certain delivery information. After confirming that the user pays attention to a certain delivery information, it is determined whether the stay time reaches or exceeds the preset threshold. If it is determined to be yes, the result is used as a quantitative reference; it can obtain interactive behavior data in real time and provide basic data support for subsequent analysis, accurately identify the user's attention to specific information, and provide a basis for personalized recommendations;

[0053] Step 3: Retrieve the user's historical interaction data from the database, analyze the user's past consumption behavior, preferences for viewing delivery information, and interaction behavior at browsing interaction points, and obtain the historical behavior coefficient; based on the in-depth analysis of the user's historical behavior, improve the accuracy and effectiveness of the recommendation system;

[0054] Step 4: Obtain the historical interaction data of all user groups in the current image acquisition area, and calculate the group interaction preference trend coefficient through big data analysis;

[0055] If there is no newly added interaction point information in the historical interaction data of the user group, the historical information of similar interaction points is collected through the Internet, and several samples are extracted for prediction through deep learning algorithms. The predicted data is obtained as the group interaction preference trend coefficient of the newly added interaction point; then the group preference is determined and the collective recommendation strategy is optimized. Even in the face of new interaction points, preliminary prediction support can be provided to avoid data sparsity problems;

[0056] Step 5: Using the group interaction trend coefficient as a reference for weight allocation of several interaction points, the user's historical behavior coefficient, the attention coefficient to the current delivery information, and the group interaction preference trend coefficient are comprehensively analyzed to output the recommendation tag. The interaction point corresponding to the recommendation tag and the user are pushed information in a two-way manner; this improves the accuracy and personalization of the recommendation and enhances the user experience;

[0057] During the two-way push of information between the interaction point corresponding to the recommended tag and the user, if either party refuses the push, the push will be terminated, and the push between the user and the interaction point with the same attribute or the interaction point and the user with the same attribute will be removed in the current cycle, and reset in the next cycle; flexibly adapt to changes in user needs, prevent user disgust, and improve system friendliness;

[0058] Step 6: Real-time analysis of the user's current interactive operations and results. In the process of real-time analysis of the user's current interactive operations and results, the weight ratio of the user's behavior coefficient and the attention coefficient of the delivery information is adjusted through the data feedback of the interaction point and the image acquisition device, and the recommendation label is re-output in the preset period; the dynamics and adaptability of the recommendation system are improved, so that the recommended content is always consistent with the user's current interests;

[0059] Eliminate similar interaction points of completed interaction points. In the process of eliminating similar interaction points of completed interaction points, if it is detected that the user has completed the target operation of a certain interaction point, then the interaction point and the interaction points whose similarity meets the preset threshold will be removed from the recommendation pool of the current cycle to avoid redundant recommendations and improve the push efficiency of effective information and user satisfaction.

[0060] Compared with existing technologies, it can improve user experience in multiple dimensions, including accurately capturing user interests, dynamically adjusting recommendations, integrating group preferences and individual behaviors, reducing redundant information push, etc. It effectively solves the problems of inaccurate interest capture, single data and delayed feedback in traditional user intelligent analysis and recommendation solutions.

[0061] In other aspects, this embodiment also provides a user intelligent analysis device based on multi-source data, such as Figure 2 As shown, including:

[0062] Management module 1, used to control the start and stop of each functional module and issue control permissions;

[0063] Data storage terminal 2, used for cloud storage of collected data and analysis data, and supports reading of local storage media;

[0064] The image acquisition module 3 is used to acquire monitoring images and deliver information in a designated area, and convert the acquired images into character data;

[0065] The attention analysis module 4 is used to analyze the image character data to obtain the attention coefficient of the target user to one or more interaction points;

[0066] The historical behavior analysis module 5 is used to analyze the user's past purchase records, content preferences and usage records of interaction points, calculate the user's individual preferences, and output the historical behavior coefficient;

[0067] Group trend analysis module 6 is used to analyze all user historical data in the current area to obtain the actual group preference trend. For newly added interaction points, the predicted group preference trend is obtained by collecting Internet data and using a deep learning algorithm. The actual group preference trend and the predicted group preference trend are comprehensively analyzed, and each interaction point is assigned a corresponding weight to output the final group preference trend.

[0068] The tag generation module 7 is used to comprehensively analyze the user's historical behavior coefficient, current attention coefficient and group interaction preference trend coefficient to generate a recommended tag;

[0069] Push module 8, used for bidirectional push according to tag content. During the push process, if the user or interaction point refuses to push, the combination will be excluded in the current cycle and reset in the next cycle;

[0070] The dynamic adjustment module 9 is used to analyze the user's current operation and results, and to adjust the dynamic coefficient. At the end of each cycle, the recommendation tags are regenerated to remove the interaction points of the completed interaction operations and their similar interaction points.

[0071] As a preferred implementation in this embodiment, Figure 2 As shown, the management module 1 is interactively connected with the data storage terminal 2, the image acquisition module 3 and the dynamic adjustment module 9 through a wireless network, the image acquisition module 3 is interactively connected with the attention analysis module 4 through a wireless network, the dynamic adjustment module 9 is interactively connected with the attention analysis module 4 and the historical behavior analysis module 5 through a wireless network, and the label generation module 7 is interactively connected with the attention analysis module 4, the historical behavior analysis module 5, the group trend analysis module 6 and the push module 8 through a wireless network.

[0072] After the user intelligent analysis device based on multi-source data in this embodiment is installed, the management personnel register the authorized interaction points through the management module 1 and send the control instructions of each module. The data storage terminal 2 records the current analysis and collection data in real time and stores it in the cloud. The image acquisition module 3 obtains the user's viewing status of the delivery information, which is quantified into a specific attention coefficient by the attention analysis module 4. The historical behavior analysis module 5 identifies the user's historical behavior data, and the group preference trend coefficient of the current area is obtained by the group trend analysis module 6. After comprehensive analysis, the label generation module 7 produces recommended labels suitable for the user, and the push module 8 performs two-way information push between the interaction point and the user. The dynamic adjustment module 9 removes the interaction points and similar interaction points that have completed the interaction operation in real time within a preset period.

[0073] As a preferred implementation manner of this embodiment, an electronic device is provided, including: one or more processors; a memory for storing one or more computer execution instructions; when the one or more computer execution instructions are executed by the one or more processors, the one or more processors implement a user intelligent analysis method based on multi-source data.

[0074] This embodiment also provides a computer-readable storage medium, and when the computer execution instructions are executed by the processor, a user intelligent analysis method based on multi-source data is implemented.

[0075] In summary, the present invention quantifies the user's current interest status in real time, obtains the user's real-time attention to the delivery information, and thus generates recommended content that better meets the user's immediate needs. Combining the user's historical behavior, current attention, and group trends, the present invention can analyze the user's interest changes in multiple dimensions and improve the comprehensiveness and accuracy of recommendations.

[0076] Based on the user's immediate behavior and interaction results, the recommendation tags are adjusted in a timely manner to ensure that the user's latest interests are quickly reflected in the recommended content. The interaction preferences of the user group are obtained through big data analysis, especially for newly added interaction points. The preference trend coefficient of the group is predicted through deep learning algorithms. It can not only consider personal historical behavior, but also make comprehensive recommendations based on group trends, which is suitable for personalized recommendation needs in public scenarios.

[0077] By analyzing the user's interaction results in real time, the present invention can eliminate completed interaction points and similar items that are irrelevant to the user's current needs to avoid repeated pushes. At the same time, when a party refuses to push, the present invention can automatically terminate the push within the current cycle to optimize the user's push experience;

[0078] Finally, the present invention improves the comprehensiveness and accuracy of recommendations by fusing multi-source data, making the recommendation results more personalized and intelligent.

[0079] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A user intelligent analysis method based on multi-source data, characterized in that: The following steps are involved: Step 1: Deploy image acquisition components and information delivery components, and register authorized interaction point data; Step 2: Quantify the attention state of a user in the image acquisition area to the information being delivered, and obtain the user's attention coefficient to the information being delivered; Step 3: retrieve the user's historical interaction data from the database, analyze the user's past consumption behavior, preference for viewing delivery information, and interaction behavior at browsing interaction points, and obtain the historical behavior coefficient; Step 4: Obtain the historical interaction data of all user groups in the current image acquisition area, and calculate the group interaction preference trend coefficient through big data analysis; Step 5: Using the group interaction trend coefficient as a reference for weight distribution of several interaction points, comprehensively analyze the user's historical behavior coefficient, the coefficient of attention to the current delivery information, and the group interaction preference trend coefficient, output the recommended label, and push information bidirectionally between the interaction point corresponding to the recommended label and the user; Step 6: Analyze the user's interaction operations and results in the current cycle in real time, and eliminate similar interaction points that have been completed.

2. The user intelligent analysis method based on multi-source data according to claim 1 is characterized in that: During the quantification process of step 2, eye tracking and facial expression technology are combined to analyze whether the user pays attention to a certain delivery information, and after confirming that the user pays attention to the certain delivery information, it is determined whether the stay time reaches or exceeds the preset threshold. If it is determined to be yes, the result is used as a quantitative reference.

3. The user intelligent analysis method based on multi-source data according to claim 1 is characterized in that: If there is no newly added interaction point information in the user group historical interaction data in step 4, historical information of similar interaction points is collected through the Internet, and several samples are extracted for prediction through a deep learning algorithm to obtain the predicted data as the group interaction preference trend coefficient of the newly added interaction point.

4. The user intelligent analysis method based on multi-source data according to claim 1 is characterized in that: During the two-way push of information between the interaction point corresponding to the recommended tag and the user in step 5, if either party refuses the push, the push will be terminated, and the push between the user and the interaction points with similar attributes or the push between the interaction point and the users with similar attributes will be eliminated in the current cycle and reset in the next cycle.

5. The user intelligent analysis method based on multi-source data according to claim 1 is characterized in that: In the process of analyzing the user's interactive operations and results in the current cycle in real time, step 6 adjusts the weight ratio of the user's behavior coefficient and the attention coefficient of the delivery information through data feedback from the interaction points and image acquisition devices, and re-outputs the recommended tags in a preset cycle.

6. The user intelligent analysis method based on multi-source data according to claim 1 is characterized in that: In the process of removing similar interaction points of completed interaction points in step 6, if it is detected that the user has completed the target operation of a certain interaction point, the interaction point and the interaction points whose similarity meets the preset threshold will be removed from the recommendation pool of the current cycle.

7. A user intelligent analysis device based on multi-source data, wherein the method is a device equipped with a user intelligent analysis method based on multi-source data according to any one of claims 1 to 6, characterized in that: include: Management module (1), used to control the start and stop of each functional module and issue control authority; A data storage terminal (2), used for cloud storage of collected data and analyzed data, and supporting reading of local storage media; An image acquisition module (3) is used to acquire monitoring images and deliver information in a designated area, and convert the acquired images into character data; An attention analysis module (4) is used to analyze the image character data to obtain the attention coefficient of the target user to one or more interaction points; A historical behavior analysis module (5) is used to analyze the user's past purchase records, content preferences, and usage records of interaction points, calculate the user's individual preferences, and output historical behavior coefficients; The group trend analysis module (6) is used to analyze the historical data of all users in the current area to obtain the actual group preference trend. For newly added interaction points, the module collects Internet data and uses a deep learning algorithm to obtain the predicted group preference trend. The actual group preference trend and the predicted group preference trend are comprehensively analyzed, and a corresponding weight is assigned to each interaction point to output the final group preference trend. A tag generation module (7) is used to comprehensively analyze the user's historical behavior coefficient, current attention coefficient and group interaction preference trend coefficient to generate a recommended tag; A push module (8) is used to perform bidirectional push based on the tag content. During the push process, if the user or the interaction point refuses to push, the combination will be excluded in the current cycle and reset in the next cycle; The dynamic adjustment module (9) is used to analyze the user's current operation and results, and to adjust the dynamic coefficient. At the end of each cycle, the recommendation tags are regenerated to remove the interaction points of the completed interaction operations and their similar interaction points.

8. The user intelligent analysis device based on multi-source data according to claim 7, characterized in that: The management module (1) is interactively connected with the data storage terminal (2), the image acquisition module (3) and the dynamic adjustment module (9) via a wireless network; the image acquisition module (3) is interactively connected with the attention analysis module (4) via a wireless network; the dynamic adjustment module (9) is interactively connected with the attention analysis module (4) and the historical behavior analysis module (5) via a wireless network; and the tag generation module (7) is interactively connected with the attention analysis module (4), the historical behavior analysis module (5), the group trend analysis module (6) and the push module (8) via a wireless network.

9. An electronic device, characterized in that: include: one or more processors; a memory for storing one or more computer executable instructions; When the one or more computer-executable instructions are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

10. A computer-readable storage medium storing computer-executable instructions, characterized in that: When the computer-executable instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images