CN-121998742-A - Cross-channel consumer intention fusion analysis method and system

Abstract

The application relates to the technical field of electric digital data processing, in particular to a cross-channel consumer intention fusion analysis method and system; the method comprises the steps of constructing a consumer dynamic consumption trip map, wherein the dynamic consumption trip map at least comprises consumer nodes, contact nodes, intention nodes and distribution of target objects, calculating intention uncertainty entropy of consumers corresponding to the consumer nodes, responding to the fact that the intention uncertainty entropy is smaller than a preset threshold, adopting a graph attention network to aggregate feature vectors of all contact nodes connected with the same consumer node to generate full-channel feature vectors, determining a current consumption stage of the consumers based on the full-channel feature vectors, and pushing corresponding information to the consumers according to the consumption stage. The application has the effects of reducing repeated pushing of marketing content and saving marketing resources.

Inventors

• DING ZIJIAN
• DING JUNWEI
• CHEN DEPIN

Assignees

• 钛动科技股份有限公司

Dates

Publication Date

20260508

Application Date

20260207

Claims (10)

1. 1. A cross-channel consumer intention fusion analysis method is characterized in that, Constructing a consumer dynamic consumption trip map, wherein the dynamic consumption trip map at least comprises consumer nodes, contact nodes, intention nodes and target objects, wherein the contact points are channels for interaction between consumers corresponding to the consumer nodes and the target objects; For any consumer, determining all contact nodes corresponding to the consumer, and calculating the intention uncertainty entropy of the consumer corresponding to the consumer node according to the distribution of the intention nodes connected with the contact nodes; In response to the intention uncertainty entropy being smaller than a preset threshold, aggregating feature vectors of all contact nodes connected with the same consumer node by adopting a graph attention network to generate a full channel feature vector; and determining the current consumption stage of the consumer based on the full channel feature vector, and pushing corresponding information to the consumer according to the consumption stage.
2. 2. The cross-channel consumer intent fusion analysis method of claim 1, wherein heterogeneous identities on different contact nodes of a consumer are associated by using a mask self-encoder, and connection edges between the consumer nodes and the contacts are determined according to the association between different heterogeneous identities.
3. 3. The method of claim 1, wherein the types of intent nodes in the dynamic consumption trip map include at least quality exploration intent nodes, price comparison intent nodes, and urgent purchase intent nodes.
4. 4. The method for cross-channel consumer intent fusion analysis as claimed in claim 1, wherein the contact nodes refer to specific media of interaction between the consumer and the target object, and the feature vectors of the contact nodes comprise individual stay time, individual emotion tendency scores and Boolean values of whether the consumer shares.
5. 5. The method of claim 1, wherein the contact nodes comprise a platform interface for interaction between the consumer and the target object and code scanning points of off-line stores.
6. 6. The cross-channel consumer intention fusion analysis method according to claim 1, wherein calculating the intention uncertainty entropy of the consumer comprises obtaining implicit intention weights of all contact nodes, accumulating the implicit intention weights of all contact nodes associated with any intention node to obtain intention accumulated intensity, taking the ratio of the intention accumulated intensity corresponding to each intention to the sum of all intention accumulated intensities as the intention probability, and calculating the entropy of all intention probabilities to obtain the intention uncertainty entropy of the consumer.
7. 7. The cross-channel consumer intent fusion analysis method of claim 6, wherein obtaining the implicit intent weights of the contact nodes comprises extracting a plurality of implicit characteristic values corresponding to the contacts, and taking the result of weighted summation of the plurality of implicit characteristic values as the implicit intent weights of the contact nodes.
8. 8. The method of claim 7, wherein the implicit characteristic values include a focus score calculated based on a length of stay of the consumer at a contact node, and a goodness score for the commodity characterizing the consumer's expression at the contact.
9. 9. The method of claim 8, wherein the implicit characteristic value further comprises a sharing score determined based on the sharing behavior of the consumer at the contact node.
10. 10. A cross-channel consumer intent fusion analysis system comprising a processor and a memory storing computer program instructions that when executed by the processor implement a cross-channel consumer intent fusion analysis method as claimed in any one of claims 1 to 9.

Description

Cross-channel consumer intention fusion analysis method and system Technical Field The application relates to the technical field of electric digital data processing, in particular to a cross-channel consumer intention fusion analysis method and system. Background With the development of digital marketing and intelligent recommendation technologies, enterprises analyze consumer behaviors in a data-driven manner to support business decisions such as advertisement delivery, commodity recommendation, customer relationship management and the like. In the related art, consumer portrait and intention recognition generally depends on user behavior data collected by an internet platform, such as explicit behaviors of browsing, clicking, collecting, purchasing and the like, and a user tag system or an interest model is constructed based on the data, so that commodity or content recommendation is realized through algorithms such as collaborative filtering and the like. Existing consumer intent analysis and recommendation systems mostly use data of a single platform or channel as the analysis basis. For example, a part of advertisement data management Platform (DATA MANAGEMENT Platform, DMP) mainly analyzes users based on behavior data in a single system such as WeChat, tremble, etc., and completes advertisement targeting and recommendation decision inside the Platform. In an actual consumption scenario, the process of purchasing goods by consumers often spans multiple online platforms, presenting distinct multi-stage, multi-touch features. For example, a consumer may first obtain product information and develop a preliminary interest in a content community platform, then go to store online for experience or comparison, and then complete a repurchase or long-term retention via an instant messaging or e-commerce platform. The above-described "cognition-consideration-decision-loyalty" consumption process has cross-platform, cross-medium continuity. However, in the related art, data collected by different channels are generally stored in separate systems in a scattered manner, so that behaviors of consumers on different platforms are split into mutually independent events, and the same consuming main body and continuous intention evolution process corresponding to the consumer cannot be identified. On one hand, the data island phenomenon makes it difficult for the system to accurately judge the real consumption stage of the consumer, and the consumer who enters the decision making or repurchase stage is easy to be misjudged as the initial interest stage, on the other hand, on the marketing execution level, the system can repeatedly push similar marketing contents in different channels for the same consumer based on incomplete or fragmented intention judgment, so that the low-efficiency use of marketing resources is caused. Disclosure of Invention In order to solve the problem of repeated pushing of marketing content caused by misjudgment of consumer intention, the application provides a cross-channel consumer intention fusion analysis method and system. In a first aspect, the present application provides a cross-channel consumer intention fusion analysis method, which adopts the following technical scheme: a cross-channel consumer intent fusion analysis method, comprising: constructing a consumer dynamic consumption trip map, wherein the dynamic consumption trip map at least comprises consumer nodes, contact nodes, intention nodes and target objects, the contact nodes are channels for interaction between consumers corresponding to the consumer nodes and the target objects, and the intention nodes are used for representing psychological states of the consumers corresponding to the consumer nodes; For any consumer, determining all contact nodes corresponding to the consumer, and calculating the intention uncertainty entropy of the consumer corresponding to the consumer node according to the distribution of the intention nodes connected with the contact nodes; In response to the intention uncertainty entropy being smaller than a preset threshold, aggregating feature vectors of all contact nodes connected with the same consumer node by adopting a graph attention network to generate a full channel feature vector; and determining the current consumption stage of the consumer based on the full channel feature vector, and pushing corresponding information to the consumer according to the consumption stage. A dynamic consumption trip map is constructed that includes consumer nodes, contact nodes, and intent nodes such that the behavior of the consumer at the different contact nodes is no longer in the form of isolated events, but is organized as structured graph data with chronological and semantic associations. Through connection between the contact nodes and the consumer nodes, specific interaction behaviors of consumers on different platforms and different media are explicitly modeled so as to distinguish action differences of