Search

CN-122022865-A - Digital economic label updating system and method based on big data analysis

CN122022865ACN 122022865 ACN122022865 ACN 122022865ACN-122022865-A

Abstract

The invention provides a digital economic label updating system and method based on big data analysis, comprising a label determining module, a label updating module and a label updating module, wherein the label determining module can acquire the return frequency of a user in real time The return category for each return Time of return Text data of return goods Promotional data And logistics data Forming a return database T, extracting according to the return database T and based on the return motivation characteristics of each return The semantic identification module comprises a semantic processing unit and a label identification unit, and is used for solving the problems that the existing label updating system is difficult to effectively identify semantic mutex labels due to the lack of multidimensional semantic analysis capability on return behaviors, so that user portrait distortion and business decision deviation are caused, the reliability of the existing system is reduced, and even the business decision is invalid and the credibility is collapsed.

Inventors

  • PENG DONG
  • DONG PEI
  • CHEN JINGWEI

Assignees

  • 平顶山工业职业技术学院(平顶山煤矿技工学校)

Dates

Publication Date
20260512
Application Date
20260130

Claims (10)

  1. 1. A digital economic label updating system based on big data analysis, comprising: The label determining module can acquire the return frequency of the user in real time The return category for each return Time of return Text data of return goods Promotional data And logistics data Forming a return database T, extracting according to the return database T and based on the return motivation characteristics of each return Determining a user tag set Y; The semantic recognition module comprises a semantic processing unit and a label recognition unit, wherein the semantic processing unit can be used for carrying out the process of identifying each label in the user label set Y Carry out loss value Calculating and generating a mutual exclusion matrix based on predefined semantic mutual exclusion rules The tag identification unit is capable of identifying a tag based on the loss value Mutual exclusion matrix Calculate each tag Mutual exclusion strength with other tags in user tag set Y And outputting the identified mutually exclusive tag pairs; The label updating module is used for updating the label according to the target business association rule Computing each tag in each mutually exclusive tag pair Semantic importance coefficients of (a) To be based on each tag Loss value of (2) Mutual exclusion strength And semantic importance coefficients Calculate its semantic score And updating the user tag set Y to include only semantic scores The highest tag.
  2. 2. The big data analysis based digital economic label updating system of claim 1, wherein the label determination module further comprises a user label set generation unit, the user label set generation unit based on return machine characteristics Obtaining each label Weights of (2) : Wherein: And Represented as a hyper-parameter; expressed as an indication function, where, if The secondary goods returning machine is Then Otherwise , And screening the weights Each tag above a preset threshold A user tag set Y is generated.
  3. 3. The big data analysis based digital economic label updating system according to claim 2, wherein the semantic processing unit performs a Word2Vec model on each label And return motor features Processing to obtain label vector Feature vector of return machine To be based on tag vectors Feature vector of return machine Calculate each tag And return motor features Semantic similarity of (2) The method comprises the following steps: , And according to the return frequency Calculating return incentive weights : Wherein: represented by the attenuation coefficient(s), By semantic similarity Weight of return machine Calculating a loss value : 。
  4. 4. The big data analysis based digital economic label updating system according to claim 3, wherein the mutex matrix Is generated based on a predefined semantic mutex rule formed by structured knowledge data acquired by a data synchronization mechanism from an external industry database and by Traversing all mutually exclusive label pairs in a predefined semantic mutually exclusive rule, filling the intensity value of each mutually exclusive label pair in the corresponding position of the initial matrix And is realized.
  5. 5. The big data analysis based digital economic label updating system of claim 4, wherein the mutual exclusion strength The method comprises the following steps: wherein: And Respectively represent the labels in any mutually exclusive label pair And labels Is a loss value of (2); representing all loss values in user tag set Y Is the maximum value of (a).
  6. 6. The big data analysis based digital economic label updating system according to claim 5, wherein the label identifying unit is based on a mutual exclusion strength Traversing mutex matrix Each tag of (3) Mutually exclusive labels If the mutual exclusion strength Is greater than a preset mutual exclusion threshold The output is a mutually exclusive tag pair.
  7. 7. The big data analysis based digital economic label updating system of claim 6, wherein the semantic importance coefficient Is calculated from each tag extracted from the historical service data of the target service according to the Apriori algorithm Association weights with target traffic And formed target business association rule To obtain: wherein: Expressed as target business association rules For each tag Wherein, upon positive influence, In the case of a negative effect, the negative effect, 。
  8. 8. The big data analysis based digital economic label updating system of claim 7, wherein the semantic score The method comprises the following steps: wherein: represented as each tag Mutually exclusive tag of (a) Is a set of (3).
  9. 9. The big data analysis based digital economic label updating system of claim 8, wherein the label updating module is based on each label Semantic score of (2) Traversing the user tag set Y and extracting semantic scores Maximum specific label To update the user tag set Y to include only specific tags 。
  10. 10. A digital economic label updating method based on big data analysis, for implementing the digital economic label updating system based on big data analysis according to any one of claims 1 to 9, the digital economic label updating method comprising: A label determining step of obtaining the return frequency of the user in real time The return category for each return Time of return Text data of return goods Promotional data And logistics data Forming a return database T, extracting according to the return database T and based on the return motivation characteristics of each return Determining a user tag set Y; semantic recognition procedure, namely, for each tag in the user tag set Y Carry out loss value Calculating and generating a mutual exclusion matrix based on predefined semantic mutual exclusion rules According to the loss value Mutual exclusion matrix Calculate each tag Mutual exclusion strength with other tags in user tag set Y And outputting the identified mutually exclusive tag pairs; Label updating step, according to target business association rule Computing each tag in each mutually exclusive tag pair Semantic importance coefficients of (a) To be based on each tag Loss value of (2) Mutual exclusion strength And semantic importance coefficients Calculate its semantic score And updating the user tag set Y to include only semantic scores The highest tag.

Description

Digital economic label updating system and method based on big data analysis Technical Field The invention relates to the technical field of digital economic label updating, in particular to a digital economic label updating system and method based on big data analysis. Background In the context of massive growth of digital economic data, massive data-driven tag update systems have become the core hub for achieving data value conversion. However, when the conventional label updating system is used for dealing with frequent return scenes of users, the multi-dimensional semantic analysis capability of return behaviors is lacking, and the semantic mutual exclusion labels are difficult to effectively identify, so that the forward labels and the negative labels are simply overlapped to form logically contradictory user portraits. Meanwhile, the transition track from high-frequency goods return to behavior improvement of the user cannot be captured, so that the label with high goods return risk is solidified for a long time, and the image distortion of the user and the business decision deviation are caused. This not only reduces the reliability of existing systems, but also results in failure of their business decisions and collapse of reliability. In view of the above, the present application provides a digital economic label updating system and method based on big data analysis. Disclosure of Invention The invention aims to solve the problems that the prior label updating system is difficult to effectively identify the semantic mutex labels due to the lack of multidimensional semantic analysis capability on return behaviors, so that user portrait distortion and business decision deviation are caused, the reliability of the prior system is reduced, and even the business decision is invalid and the reliability is collapsed. In order to achieve the above object, the present invention provides a digital economic label updating system and method based on big data analysis. According to a first aspect of the present invention, there is provided a digital economic label updating system based on big data analysis, comprising: The label determining module can acquire the return frequency of the user in real time The return category for each returnTime of returnText data of return goodsPromotional dataAnd logistics dataForming a return database T, extracting according to the return database T and based on the return motivation characteristics of each returnDetermining a user tag set Y; The semantic recognition module comprises a semantic processing unit and a label recognition unit, wherein the semantic processing unit can be used for carrying out the process of identifying each label in the user label set Y Carry out loss valueCalculating and generating a mutual exclusion matrix based on predefined semantic mutual exclusion rulesThe tag identification unit is capable of identifying a tag based on the loss valueMutual exclusion matrixCalculate each tagMutual exclusion strength with other tags in user tag set YAnd outputting the identified mutually exclusive tag pairs; The label updating module is used for updating the label according to the target business association rule Computing each tag in each mutually exclusive tag pairSemantic importance coefficients of (a)To be based on each tagLoss value of (2)Mutual exclusion strengthAnd semantic importance coefficientsCalculate its semantic scoreAnd updating the user tag set Y to include only semantic scoresThe highest tag. Optionally, the tag determination module further includes a user tag set generation unit, the user tag set generation unit based on the return machine featureObtaining each labelWeights of (2): Wherein: And Represented as a hyper-parameter; expressed as an indication function, where, if The secondary goods returning machine isThenOtherwise, And screening the weightsEach tag above a preset thresholdA user tag set Y is generated. Alternatively, the semantic processing unit processes each tag through a Word2Vec modelAnd return motor featuresProcessing to obtain label vectorFeature vector of return machineTo be based on tag vectorsFeature vector of return machineCalculate each tagAnd return motor featuresSemantic similarity of (2)The method comprises the following steps: , And according to the return frequency Calculating return incentive weights: Wherein: represented by the attenuation coefficient(s), By semantic similarityWeight of return machineCalculating a loss value: 。 Alternatively, the mutual exclusion matrixIs generated based on a predefined semantic mutex rule formed by structured knowledge data acquired by a data synchronization mechanism from an external industry database and byTraversing all mutually exclusive label pairs in a predefined semantic mutually exclusive rule, filling the intensity value of each mutually exclusive label pair in the corresponding position of the initial matrixAnd is realized. Alternatively, the mutual exclusion strength