Intelligent evaluation system for user service needs based on portrait construction

Abstract

The present invention discloses a user service demand intelligent evaluation system based on portrait construction, which relates to the technical field of user service demand evaluation and solves the technical problem in the prior art that a portrait label system cannot be constructed according to portrait label settings, so that evaluation standards cannot be formulated in real time to perform targeted service demand evaluation. Specifically, a data dimension detection unit performs dimensional analysis on data collected in the process of user service demand evaluation; a demand forecast lag analysis unit performs demand forecast lag analysis on user dimensional data collection; a data screening module screens the collected dimensional data and divides the data of each dimensionality; a portrait update module sets portrait labels for the collected user dimensional data, sets weights for the portrait labels, and constructs a portrait label system according to portrait labels with different weights; a service demand evaluation unit performs intelligent evaluation of user service needs according to the portrait label system.

Description

Technical Field

[0001] The present invention relates to the technical field of user service demand assessment, and in particular to a user service demand intelligent assessment system based on portrait construction. Background Art

[0002] Intelligent assessment of user service needs based on portrait construction refers to the process of building dynamic user portraits by integrating multi-dimensional information such as user basic data, behavioral data, historical service records, and using machine learning, natural language processing, data analysis and other technologies to accurately identify, prioritize and provide personalized responses to users' potential service needs.

[0003] However, in the existing technology, the data dimension is single when evaluating user service needs, making it difficult to capture the dynamic behavioral characteristics of users; the demand forecast is delayed, and it is impossible to respond to scene changes in real time; at the same time, it is impossible to collect and screen dimensional data, and the impact of misoperation cannot be eliminated; in addition, it is impossible to build a portrait label system based on the portrait label settings, so that it is impossible to formulate evaluation standards in real time for targeted service needs evaluation.

[0004] In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0005] The purpose of the present invention is to solve the above-mentioned problems and to propose an intelligent user service demand evaluation system based on portrait construction.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] The user service demand intelligent assessment system built based on portraits includes a service demand assessment platform, which includes:

[0008] Data dimension detection unit, which performs dimensional analysis on data collected during the user service demand assessment process;

[0009] Demand forecast lag analysis unit, which collects user dimension data and performs demand forecast lag analysis;

[0010] After the data dimension detection unit determines that the dimension detection meets the actual needs, the data screening module screens the collected dimension data and divides the individual dimension data;

[0011] The portrait update module sets portrait tags for the collected user dimension data, sets weights for the portrait tags, and builds a portrait tag system based on portrait tags with different weights;

[0012] The service demand assessment unit conducts intelligent assessment of users' service needs based on the portrait label system.

[0013] As a preferred embodiment of the present invention, the operation process of the data dimension detection unit is as follows:

[0014] Collect user service demand data, where service demand refers to product purchase demand or consulting service demand; perform dimensionality testing on the currently collected data, where data dimensions are divided into user dimension, behavior dimension, and emotion dimension; when facing different service demands, if the user data in any dimension has different numerical values, it is determined that the currently collected data dimension meets the actual evaluation requirements;

[0015] If there are no different numerical values ​​for any dimension of the user's data, a single dimension evaluation will be performed;

[0016] When the user generates dimensional data, the static and dynamic collection of user data is determined, and the ratio of the number of static collection and dynamic collection during the current dimensional data collection is counted. If the number ratio exceeds the set ratio threshold, or the number ratio continues to increase, it is inferred that as the user's dimensional data increases, the current data dimension is single, and the satisfaction of the dimensional data detection needs is on a downward trend; if the number ratio does not exceed the set ratio threshold, and the number ratio does not continue to increase, it is inferred that as the user's dimensional data increases, the current data dimension meets the user's demand assessment.

[0017] As a preferred embodiment of the present invention, the operation process of the demand forecast hysteresis analysis unit is as follows:

[0018] The user dimension data collection phase is divided into several sub-phases. The floating trend of dimension data in each sub-phase and the floating trend of dimension data in the entire collection phase are collected. If the corresponding floating trends are inconsistent, the sub-phase is marked as a floating phase; if the corresponding floating trends are consistent, the sub-phase is marked as a stable phase.

[0019] During the floating phase, the interval between the time when user dimension data trend analysis is collected and the time when the floating phase ends. During the stable phase, the value span of the dimension data corresponding to multiple stable phases is increased when analyzing user dimension data trend.

[0020] When the time interval duration in the floating stage exceeds the interval duration threshold, or the value increase span in the stable stage exceeds the span threshold, it is inferred that the hysteresis analysis in the dimension data collection process is abnormal. The service demand assessment platform shortens the dimension data collection prediction cycle, that is, floating real-time prediction is performed in the floating stage, and the value floating span critical value is set in the stable stage. If it exceeds the critical value, demand prediction is performed in time; when the time interval duration in the floating stage does not exceed the interval duration threshold, and the value increase span in the stable stage does not exceed the span threshold, it is inferred that the hysteresis analysis in the dimension data collection process is normal.

[0021] As a preferred embodiment of the present invention, the operation process of the data screening module is as follows:

[0022] The generation mode is divided into active generation and passive generation according to the analysis of the generation time of dimension data;

[0023] Collect the number of times users actively generate and passively generate data. If the type of dimension data corresponding to the number of times actively generated is continuously generated during the data collection phase, the corresponding dimension data will be marked as continuously watched data. If the type of dimension data corresponding to the number of times actively generated is generated in a concentrated period during the data collection phase and is not generated continuously, the corresponding dimension data will be marked as short-term watched data.

[0024] If the ratio of the interval duration between the continuous generation moments of the type of dimension data corresponding to the number of passive generation times to the duration of the continuous generation of the dimension data is lower than the duration ratio threshold, the corresponding dimension data will be marked as pushed attention-grabbing data; if the ratio of the interval duration between the continuous generation moments of the type of dimension data corresponding to the number of passive generation times to the duration of the continuous generation of the dimension data is not lower than the duration ratio threshold, the corresponding dimension data will be marked as pushed but no attention data.

[0025] As a preferred embodiment of the present invention, the operation process of the portrait update module is as follows:

[0026] Set portrait labels for each type of dimension data, and the portrait labels are divided into risk preference labels, category preference labels, and demand preference labels; each type of label is converted into a feature vector, and the feature vector conversion is performed by selecting one type of feature vector conversion within the range of each type of label;

[0027] The eigenvector of the risk preference label is the excess of the service demand set period below the floating frequency threshold and the user's demand period; the category preference label is represented by the overlapping browsing frequency of the service demand type in the dimensional data; the demand preference label is represented by the frequency of users' continuous browsing within the set period of obtaining the service demand.

[0028] As a preferred embodiment of the present invention, after completing the feature vector conversion, the corresponding data of each type of portrait label is used to construct a portrait label system, and the ratio of the feature vector value corresponding to the same type of label to the corresponding feature vector setting threshold is obtained according to the service needs of the user's historical selection, and marked as the portrait evaluation ratio; the real-time portrait evaluation ratio of each type of label is compared with the historical portrait evaluation ratio of the same type of label in real time, wherein whether the real-time portrait evaluation ratio of each type of label exceeds the portrait evaluation ratio corresponding to the historical selection service demand is used as the weight setting standard for the current portrait label, that is, the three portrait labels are set as high weight, medium weight and low weight respectively, and the portrait label system is constructed according to the weight.

[0029] As a preferred embodiment of the present invention, the operation process of the service demand evaluation unit is as follows:

[0030] According to the portrait label system, data analysis is performed on each weighted portrait label in the system, and the span of the peak value increase of the eigenvector corresponding to the high-weight portrait label and the span of the valley value decrease of the eigenvector corresponding to the low-weight portrait label in the portrait label system are collected. The span value ratio is obtained based on the ratio calculation. At the same time, in the floating stage of the eigenvector corresponding to the medium-weight portrait label in the portrait label system, the frequency value ratio of increasing to the high-weight eigenvector value range and decreasing to the weighted eigenvector value range is obtained.

[0031] As a preferred embodiment of the present invention, if the span value ratio exceeds the span value ratio threshold, it is inferred that the bipolar weight division of the current portrait label system is consistent, and the bipolar weights of the current portrait label system are used as the user service demand evaluation standard; if the span value ratio does not exceed the span value ratio threshold, it is inferred that the bipolar weight division of the current portrait label system is not consistent, and the floating trend of the corresponding bipolar weights of the current portrait label system is used as the user service demand evaluation standard;

[0032] If the frequency value ratio exceeds the maximum value of the frequency value ratio range, it is inferred that there is an increase in data for the high-weight portrait tags of the current portrait tag system, and data analysis is performed based on the increased portrait tags to conduct service demand assessment; if the frequency value ratio does not exceed the minimum value of the frequency value ratio range, it is inferred that there is an increase in data for the low-weight portrait tags of the current portrait tag system, and data analysis is performed based on the increased portrait tags to conduct service demand assessment;

[0033] If the frequency value ratio is within the frequency value ratio range, it is inferred that the floating impact of the weight data of the current portrait label system is small, and the current portrait label system is used as the user service demand evaluation standard.

[0034] As a preferred embodiment of the present invention, user evaluation is performed after determining the user service demand evaluation standard, and the current user's service demand is inferred based on the weight of each portrait tag reflected in the portrait tag and the influence of the corresponding dimensional data.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. In the present invention, detection and collection are carried out during user service data detection to ensure that the user's collected data is highly available when the portrait is constructed, to avoid lags in collected data or incomplete data, which may cause deviations in user service demand evaluation; demand forecast lag analysis is performed on user dimension data collection, and the dimension data collection and analysis process is used to infer whether there is a lag in collected data when the current dimension data is used as a user demand evaluation parameter, resulting in untimely real-time demand evaluation, so that the accuracy of user service demand evaluation is reduced, and it is impossible to make accurate predictions in line with actual needs.

[0037] 2. In the present invention, the collected dimensional data is screened to avoid user misoperation when generating each dimensional data. Personal habits or advertising push locations affect the amount of collected dimensional data, resulting in errors in user demand assessment and affecting the accuracy of the assessment results. Portrait labels are set for the collected user dimensional data, and feature vectors are converted according to the portrait labels. Through the portrait label setting and corresponding feature vector analysis, the portrait labels are updated in real time to facilitate user demand assessment based on the real-time latest portrait labels, thereby improving the accuracy of user demand assessment and avoiding excessive dimensional data corresponding to multiple portrait labels, resulting in a large amount of data processing workload for service demand assessment and prone to errors. The latest portrait labels are obtained through portrait updates to maximize the availability of existing collected data and the accuracy of analysis.

[0038] 3. In the present invention, an intelligent evaluation of user service needs is performed based on the portrait tag system, and service needs are evaluated based on the real-time updated portrait tag system, which effectively improves the accuracy of user evaluation and ensures the availability of real-time collected data. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] To facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0040] Figure 1 This is a principle block diagram of the first embodiment of the present invention;

[0041] Figure 2 This is a principle block diagram of the second embodiment of the present invention;

[0042] Figure 3 This is a principle block diagram of embodiment 3 of the present invention. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0044] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0045] Example 1

[0046] See also Figure 1 As shown, the user service demand intelligent evaluation system based on the portrait construction includes a service demand evaluation platform, wherein the service demand evaluation platform includes a data dimension detection unit and a demand prediction lag analysis unit; this embodiment is used to perform detection and collection during user service data detection to ensure high availability of user collected data when the portrait is constructed, and avoid the occurrence of lags in collected data or incomplete data, which may cause deviations in user service demand evaluation;

[0047] The service demand assessment platform generates a data dimension detection signal and sends it to the data dimension detection unit;

[0048] After receiving the data dimension detection signal, the data dimension detection unit performs dimension analysis on the data collected during the user service demand assessment process to ensure that the real-time data collection dimensions are comprehensive, overcome the impact of single collected data, and improve the accuracy of user service demand assessment;

[0049] Collect user service demand data, where service demand refers to product purchase demand or consulting service demand; perform dimension detection on the currently collected data, where the data dimensions are divided into user dimension, behavior dimension, and emotion dimension. The user dimension is represented by the user's age, location, and consumption level; the behavior dimension is represented by the user's web browsing history, interaction duration, click heat map, etc.; the emotion dimension is represented by customer service conversation text and evaluation semantic analysis;

[0050] When facing different service requirements, if the user's data in any dimension has different numerical values, it is determined that the current collected data dimension meets the actual evaluation requirements;

[0051] For example, the age in the user dimension data does not overlap with the current service demand target group, the number of browsing records corresponding to service demand in the behavior dimension does not increase, and the continuous frequency of conversation texts in the emotion dimension gradually decreases;

[0052] If there are no different numerical values ​​for any dimension of the user's data, a single dimension evaluation will be performed;

[0053] When a user generates dimension data, the static and dynamic collection of user data is determined. Static collection means that the user records the current dimension data in a timely manner after it is generated, such as recording the user's web browsing history after it is added; dynamic collection means that the user continues to analyze and record the current dimension data after it is generated, such as dynamically recording the browsing tracks related to the user's current web browsing history after the user's web browsing history is added and recorded. The related browsing tracks correspond to the same type of goods or services;

[0054] The ratio of static collection to dynamic collection times for each dimension data is counted. If the ratio exceeds the set threshold or continues to increase, it is inferred that as the user's dimension data increases, the current data dimension is single, and the satisfaction of dimension data detection requirements is declining. This makes it impossible to evaluate user service needs with the current dimension data.

[0055] If the frequency ratio does not exceed the set ratio threshold and the frequency ratio does not continue to increase, it is inferred that as the user's dimension data increases, the current data dimension meets the user's demand assessment;

[0056] The dimension data in this section is limited to behavioral and emotional dimensions, which can reflect the initiative of users' service needs. The service demand assessment platform adjusts the dimension data collection type based on the dimension detection results.

[0057] After completing the data dimension detection, the service demand assessment platform generates a demand forecast lag analysis signal and sends it to the demand forecast lag analysis unit;

[0058] After receiving the demand forecast lag analysis signal, the demand forecast lag analysis unit performs demand forecast lag analysis on the user dimension data collection. Through the dimension data collection and analysis process, it is inferred whether there is a data lag in the current dimension data when it is used as a user demand assessment parameter, resulting in untimely real-time demand assessment, which reduces the accuracy of user service demand assessment and makes it impossible to make an accurate forecast that meets actual demand.

[0059] The user dimension data collection phase is divided into several sub-phases. The floating trend of dimension data in each sub-phase and the floating trend of dimension data in the entire collection phase are collected. If the corresponding floating trends are inconsistent, the sub-phase is marked as a floating phase; if the corresponding floating trends are consistent, the sub-phase is marked as a stable phase.

[0060] During the floating phase, the interval between the time when user dimension data trend analysis is collected and the time when the floating phase ends. During the stable phase, the value span of the dimension data corresponding to multiple stable phases is increased when analyzing user dimension data trend.

[0061] When the time interval duration in the floating phase exceeds the interval duration threshold, or the value increase span in the stable phase exceeds the span threshold, it is inferred that the hysteresis analysis in the dimension data collection process is abnormal. The service demand assessment platform shortens the dimension data collection and prediction cycle. That is, it performs floating real-time prediction in the floating phase and sets a critical value for the value floating span in the stable phase. If it exceeds the critical value, it will conduct demand forecast in a timely manner.

[0062] If the time interval duration in the floating phase does not exceed the interval duration threshold, and the value increase span in the stable phase does not exceed the span threshold, it is inferred that the hysteresis analysis during the dimension data collection process is normal;

[0063] The threshold values ​​used in this embodiment are parameters set manually by those skilled in the art during actual operation and are used to detect various types of collected data;

[0064] Example 2

[0065] This embodiment improves the service demand assessment platform based on the previous embodiment, and performs further processing after dimension detection and hysteresis analysis. Figure 2 As shown, the data dimension detection unit is connected to the data screening module in communication; the demand forecast lag analysis unit is connected to the portrait updating module in communication;

[0066] After determining that the dimension detection meets the actual requirements, the data dimension detection unit generates a data screening signal and sends it to the data screening module;

[0067] After receiving the data, the data screening module screens the collected dimension data to avoid user misoperation when generating dimension data. Personal habits or ad push locations affect the amount of dimension data collected, causing errors in user demand assessment and affecting the accuracy of the assessment results.

[0068] Based on the analysis of the time when dimension data is generated, the generation method is divided into active generation and passive generation. Active generation and passive generation respectively refer to users actively searching or passively clicking on ads. Passive ad clicks refer to users clicking too quickly and clicking on the wrong location.

[0069] Collect the number of times users actively generate and passively generate data. If the type of dimension data corresponding to the number of times actively generated is continuously generated during the data collection phase, the corresponding dimension data will be marked as continuously watched data. If the type of dimension data corresponding to the number of times actively generated is generated in a concentrated period during the data collection phase and is not generated continuously, the corresponding dimension data will be marked as short-term watched data.

[0070] If the ratio of the duration between the consecutive generation moments of the dimension data of the type corresponding to the number of passive generation times to the duration of the continuous generation of the dimension data is lower than the duration ratio threshold, the corresponding dimension data will be marked as data that attracts attention when pushed; if the ratio of the duration between the consecutive generation moments of the dimension data of the type corresponding to the number of passive generation times to the duration of the continuous generation of the dimension data is not lower than the duration ratio threshold, the corresponding dimension data will be marked as data that is pushed but not attracts attention;

[0071] Upload each type of dimension data and corresponding users to the service demand assessment platform for storage;

[0072] The demand forecast lag analysis unit generates a portrait update signal and sends it to the portrait update module;

[0073] After receiving the portrait update signal, the portrait update module sets portrait tags for the collected user's dimensional data and converts the eigenvectors according to the portrait tags. Through the portrait tag setting and corresponding eigenvector analysis, the portrait tags are updated in real time to facilitate user demand assessment based on the latest real-time portrait tags, thereby improving the accuracy of user demand assessment and avoiding the excessive amount of dimensional data corresponding to multiple portrait tags, which will cause a large amount of data processing workload for service demand assessment and easily generate errors. The latest portrait tags are obtained through portrait updates to maximize the availability of existing collected data and the accuracy of analysis.

[0074] Each type of dimension data is labeled with a portrait, and the portrait labels are divided into risk preference labels, category preference labels, and demand preference labels. The risk preference label represents risk parameters such as the floating frequency of each service demand period and the user's demand period; the category preference label represents the product type of the service demand; and the demand preference label represents the user's real-time demand for the service demand.

[0075] Each type of label is converted into a feature vector, and the feature vector conversion is performed by selecting one type of label within the range of each type of label. For example, the risk deviation label contains multiple data such as usage period and usage cost. Any one of them can be analyzed and all data are applicable to this system;

[0076] The eigenvector of the risk preference label is the excess of the service demand set usage period below the floating frequency threshold and the user's required usage period; the category preference label is represented by the overlapping browsing frequency of the service demand type in the dimension data; the demand preference label is represented by the frequency of continuous browsing by the user within the set usage period of the service demand;

[0077] After completing the feature vector conversion, the corresponding data of each type of portrait label is used to construct a portrait label system, and the ratio of the feature vector value of the corresponding label of the same type to the threshold set for the corresponding feature vector is obtained according to the service needs of the user's historical selection, and marked as the portrait evaluation ratio; the real-time portrait evaluation ratio of each type of label is compared with the historical portrait evaluation ratio of the same type of label in real time, among which whether the real-time portrait evaluation ratio of each type of label exceeds the portrait evaluation ratio corresponding to the historical selection service needs is used as the weight setting standard for the current portrait label, that is, the three portrait labels are set as high weight, medium weight and low weight respectively, and the portrait label system is constructed according to the weight;

[0078] The portrait label system with completed weight setting will be sent to the service demand assessment platform.

[0079] Example 3

[0080] in accordance with Figure 3 As shown, the service demand assessment platform is communicatively connected to the service demand assessment unit;

[0081] After receiving the portrait label system, the service demand assessment platform generates a service demand assessment signal and sends it to the service demand assessment unit;

[0082] After receiving the service demand assessment signal, the service demand assessment unit conducts an intelligent assessment of the user's service demand based on the portrait tag system. This assessment is performed based on the real-time updated portrait tag system, effectively improving the accuracy of user assessment and ensuring the availability of real-time collected data;

[0083] According to the portrait label system, data analysis is performed on each weighted portrait label in the system, and the span of the peak value increase of the eigenvector corresponding to the high-weighted portrait label and the span of the valley value decrease of the eigenvector corresponding to the low-weighted portrait label in the portrait label system are collected. The span value ratio is calculated based on the ratio. At the same time, in the floating stage of the eigenvector corresponding to the medium-weighted portrait label in the portrait label system, the frequency value ratio of the increase to the high-weighted eigenvector value range and the decrease to the weighted eigenvector value range is obtained;

[0084] If the span value ratio exceeds the span value ratio threshold, it is inferred that the current portrait label system's two-pole weight division is consistent, and the two-pole weight of the current portrait label system is used as the user service demand evaluation standard, where the two-pole weight is expressed as high weight and low weight;

[0085] If the span value ratio does not exceed the span value ratio threshold, it is inferred that the current portrait label system's two-pole weight division is not appropriate, and the floating trend of the current portrait label system's corresponding two-pole weights is used as the user service demand evaluation standard;

[0086] If the frequency value ratio exceeds the maximum value of the frequency value ratio range, it is inferred that there is an increase in data of the high-weight portrait tags in the current portrait tag system, and data analysis is performed based on the increased portrait tags to conduct service demand assessment;

[0087] If the frequency value ratio does not exceed the minimum value of the frequency value ratio range, it is inferred that there is data increase in the low-weight portrait tags of the current portrait tag system, and data analysis is performed based on the increased portrait tags to conduct service demand assessment;

[0088] If the frequency value ratio is within the frequency value ratio range, it is inferred that the weight data fluctuation impact of the current portrait label system is small, and the current portrait label system is used as the user service demand evaluation standard;

[0089] Among them, user evaluation is carried out after determining the user service demand evaluation standards, and the current user's service needs are inferred based on the weight of each portrait label in the portrait label system and the influence of the corresponding dimension data.

[0090] When the present invention is in use, the data dimension detection unit performs dimensional analysis on the data collected during the user service demand evaluation process; the demand forecast lag analysis unit performs demand forecast lag analysis on the user dimensional data collection; after the data dimension detection unit determines that the dimensional detection meets the actual demand, the data screening module screens the collected dimensional data and divides the data of each dimension; the portrait update module sets portrait labels for the collected user dimensional data, sets weights for the portrait labels, and constructs a portrait label system based on portrait labels with different weights; the service demand evaluation unit performs intelligent service demand evaluation on the user based on the portrait label system.

[0091] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to specific embodiments. Obviously, many modifications and variations are possible based on the contents of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. The user service demand intelligent evaluation system based on portrait construction is characterized by: Including service demand assessment platform, the service demand platform includes: The data dimension detection unit performs dimensional analysis on the data collected during the user service demand assessment process; the data dimensions are divided into user dimension, behavior dimension, and emotion dimension; The demand forecast lag analysis unit collects dimension data and performs demand forecast lag analysis. The operation process of the demand forecast lag analysis unit is as follows: Divide the dimension data collection phase into several sub-phases, collect the dimension data floating trends of each sub-phase and the dimension data floating trends of the entire collection phase. If the corresponding floating trends are inconsistent, mark the sub-phase as a floating phase; if the corresponding floating trends are consistent, mark the sub-phase as a stable phase. In the floating phase, the interval between the time when the dimension data trend analysis is collected and the time when the floating phase ends is determined. In the stable phase, the value span of the dimension data corresponding to multiple stable phases is increased when the dimension data trend analysis is performed. If the time interval duration in the floating phase exceeds the interval duration threshold, or the value increase span in the stable phase exceeds the span threshold, it is inferred that the hysteresis analysis during the dimension data collection process is abnormal. The service demand assessment platform shortens the dimension data collection and prediction cycle, that is, it performs floating real-time prediction in the floating phase and sets a critical value for the value floating span in the stable phase. If it exceeds the critical value, it will promptly perform demand prediction. If the time interval duration in the floating phase does not exceed the interval duration threshold, and the value increase span in the stable phase does not exceed the span threshold, it is inferred that the hysteresis analysis during the dimension data collection process is normal. After the data dimension detection unit determines that the dimension detection meets the actual needs, the data screening module screens the dimension data and divides the data into various dimensions; The portrait update module sets portrait tags for dimension data, sets weights for portrait tags, and builds a portrait tag system based on portrait tags with different weights; The service demand assessment unit conducts intelligent assessment of users' service needs based on the portrait label system.

2. The user service demand intelligent evaluation system based on portrait construction according to claim 1 is characterized in that: The operation process of the data dimension detection unit is as follows: Collect user service demand data, where service demand refers to product purchase demand or consulting service demand; perform dimensionality testing on the currently collected data, where data dimensions are divided into user dimension, behavior dimension, and emotion dimension; when facing different service demands, if the user data in any dimension has different numerical values, it is determined that the currently collected data dimension meets the actual evaluation requirements; If there are no different numerical values ​​for any dimension of the user's data, a single dimension evaluation will be performed; When the user generates dimensional data, the static and dynamic collection of user data is determined, and the ratio of the number of static collection and dynamic collection during the current dimensional data collection is counted. If the number ratio exceeds the set ratio threshold, or the number ratio continues to increase, it is inferred that as the dimensional data increases, the current data dimension is single, and the satisfaction of the dimensional data detection needs is on a downward trend; if the number ratio does not exceed the set ratio threshold, and the number ratio does not continue to increase, it is inferred that as the dimensional data increases, the current data dimension meets the user demand assessment.

3. The user service demand intelligent evaluation system based on portrait construction according to claim 1 is characterized in that: The operation process of the data screening module is as follows: The generation mode is divided into active generation and passive generation according to the analysis of the generation time of dimension data; Collect the number of times users actively generate and passively generate data. If the type of dimension data corresponding to the number of times actively generated is continuously generated during the data collection phase, the corresponding dimension data will be marked as continuously watched data. If the type of dimension data corresponding to the number of times actively generated is generated in a concentrated period during the data collection phase and is not generated continuously, the corresponding dimension data will be marked as short-term watched data. If the ratio of the interval duration between the continuous generation moments of the type of dimension data corresponding to the number of passive generation times to the duration of the continuous generation of the dimension data is lower than the duration ratio threshold, the corresponding dimension data will be marked as pushed attention-grabbing data; if the ratio of the interval duration between the continuous generation moments of the type of dimension data corresponding to the number of passive generation times to the duration of the continuous generation of the dimension data is not lower than the duration ratio threshold, the corresponding dimension data will be marked as pushed but no attention data.

4. The user service demand intelligent evaluation system based on portrait construction according to claim 1 is characterized in that: The operation process of the portrait update module is as follows: Set portrait labels for each type of dimension data, and the portrait labels are divided into risk preference labels, category preference labels, and demand preference labels; each type of label is converted into a feature vector, and the feature vector conversion is performed by selecting one type of feature vector conversion within the range of each type of label; The eigenvector of the risk preference label is the excess of the service demand set period below the floating frequency threshold and the user's demand period; the category preference label is represented by the overlapping browsing frequency of the service demand type in the dimensional data; the demand preference label is represented by the frequency of users' continuous browsing within the set period of obtaining the service demand.

5. The user service demand intelligent evaluation system based on portrait construction according to claim 4 is characterized in that: After completing the feature vector conversion, the corresponding data of each type of portrait label will be used to construct a portrait label system, and the ratio of the feature vector value of the corresponding label of the same type to the threshold set for the corresponding feature vector will be obtained according to the service needs of the user's historical selection, and marked as the portrait evaluation ratio; the real-time portrait evaluation ratio of each type of label will be compared with the historical portrait evaluation ratio of the same type of label in real time, among which whether the real-time portrait evaluation ratio of each type of label exceeds the portrait evaluation ratio corresponding to the historical selection service needs will be used as the weight setting standard for the current portrait label, that is, the three portrait labels will be set as high weight, medium weight and low weight respectively, and the portrait label system will be constructed according to the weight.

6. The user service demand intelligent evaluation system based on portrait construction according to claim 1 is characterized in that: The operation process of the service demand assessment unit is as follows: According to the portrait label system, data analysis is performed on each weighted portrait label in the system, and the span of the peak value increase of the eigenvector corresponding to the high-weight portrait label and the span of the valley value decrease of the eigenvector corresponding to the low-weight portrait label in the portrait label system are collected. The span value ratio is obtained based on the ratio calculation. At the same time, in the floating stage of the eigenvector corresponding to the medium-weight portrait label in the portrait label system, the frequency value ratio of increasing to the high-weight eigenvector value range and decreasing to the weighted eigenvector value range is obtained.

7. The user service demand intelligent evaluation system based on portrait construction according to claim 6 is characterized in that: If the span value ratio exceeds the span value ratio threshold, it is inferred that the current portrait label system's two-pole weight division is consistent, and the two-pole weights of the current portrait label system are used as the user service demand evaluation standard; if the span value ratio does not exceed the span value ratio threshold, it is inferred that the current portrait label system's two-pole weight division is inconsistent, and the floating trend of the corresponding two-pole weights of the current portrait label system is used as the user service demand evaluation standard; If the frequency value ratio exceeds the maximum value of the frequency value ratio range, it is inferred that there is an increase in data for the high-weight portrait tags of the current portrait tag system, and data analysis is performed based on the increased portrait tags to conduct service demand assessment; if the frequency value ratio does not exceed the minimum value of the frequency value ratio range, it is inferred that there is an increase in data for the low-weight portrait tags of the current portrait tag system, and data analysis is performed based on the increased portrait tags to conduct service demand assessment; If the frequency value ratio is within the frequency value ratio range, it is inferred that the floating impact of the weight data of the current portrait label system is small, and the current portrait label system is used as the user service demand evaluation standard.

8. The user service demand intelligent evaluation system based on portrait construction according to claim 7 is characterized in that: After determining the user service demand evaluation criteria, conduct user evaluation. Based on the weight of each portrait tag in the portrait tag system and the influence of the corresponding dimension data, infer the current user's service needs.

Images