Intelligent decision-making methods, devices, and electronic equipment based on multi-dimensional profile linkage

Abstract

This disclosure presents an intelligent decision-making method, apparatus, and electronic device based on multi-dimensional profile linkage. One specific implementation of the method includes: collecting multi-dimensional business data, wherein the multi-dimensional business data includes: user datasets, product datasets, sales datasets, organizational resource datasets, and cooperation datasets; utilizing the multi-dimensional business data to construct a five-dimensional profile model, wherein the five-dimensional profile model includes: user profiles, product profiles, sales profiles, organizational resource profiles, and cooperation profiles; constructing a dynamic decision matrix based on the five-dimensional profile model; and generating intelligent decision information based on the dynamic decision matrix. This implementation can improve the decision success rate.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the fields of computer technology, sales automation technology, and artificial intelligence technology, and specifically to intelligent decision-making methods, devices, and electronic devices based on multi-dimensional profile linkage. Background Technology

[0002] Multidimensional profiling is a feature model that systematically depicts a target (e.g., user / item / enterprise) from multiple core dimensions. Currently, traditional CRM (Customer Relationship Management System) systems can only achieve single-point data recording and lack the ability to deeply integrate multidimensional profiles; sales strategies rely on updates from human experience databases, resulting in a response speed that lags behind market changes; and rigid resource matching mechanisms lead to the loss of high-value business opportunities due to response delays or capability mismatches.

[0003] Specifically, the current field of enterprise intelligent decision-making suffers from systemic technological deficiencies, resulting in a digital paradox of "abundant data but lack of insight," manifested in three core contradictions: First, fragmented data dimensions lead to biased decision-making. Traditional systems rely on isolated data sources to build single-point profiles (such as basic customer files), failing to integrate multi-dimensional information such as product characteristics, sales strategies, and organizational capabilities. Gartner research confirms that 83% of enterprises suffer from distorted business opportunity assessments due to data silos, with user demand prediction errors exceeding 40%. For example, delays in technical resource scheduling result in a large number of annual lost orders, while the lack of partner capability data further increases the error in calculating the probability of joint bidding to ±25%, exposing a fundamental deficiency in the ecosystem collaboration mechanism.

[0004] Second, delayed strategy iteration leads to market mismatch. Static knowledge bases are updated over 30 days (Forrester data), failing to adapt to frequent market changes. Historical case studies lack adversarial validation, resulting in a strategy failure rate as high as 68% when the competitive environment changes. For example, standardized scripts have a conversion rate of only 12% in the healthcare industry, 41 percentage points lower than in the financial industry, highlighting a structural deficiency in contextual adaptability. In particular, the delayed strategy response (average 72 hours) directly causes companies to miss 19% of end-of-quarter sales opportunities.

[0005] Third, there is a double loss from resource mismatch and ecosystem failure. The manual scheduling model results in a 37% quantitative mismatch rate between customer needs and resource characteristics (e.g., the vague demand for "cloud computing experts" does not match specific skill sets). A Boston Consulting Group survey shows that the average time from resource application to deployment is 52 hours, with over 60% of business opportunities lost during this window. More seriously, partner management remains at the information ledger level. IDC (Internet Data Center) data indicates that cross-enterprise coordination requires 7.3 communications (4 times higher than internal collaboration), and the decision-making noise increases by 18% for each additional partner, revealing an out-of-control state of ecosystem risk transmission.

[0006] Therefore, the root cause of the defect can be attributed to four technological gaps: 1. Data layer: Lacks unstructured processing capabilities; only 22% of the data is usable. 2. Algorithm layer: The accuracy of a single-dimensional linear model in complex scenarios is <51%; 3. Application layer: Asynchronous operation of business flow and resource flow results in a resource idle rate of ≥34%; 4. Ecosystem layer: The lack of cross-entity encrypted sharing mechanism leads to an annual decline in collaborative efficiency of >15%.

[0007] This forces enterprises to suffer triple performance losses: 37% due to biased decision-making, 72-hour response delay, and 46% due to resource misallocation. Systemic restructuring through multi-dimensional dynamic coupling technology is urgently needed. Summary of the Invention

[0008] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0009] Some embodiments of this disclosure propose intelligent decision-making methods, devices, and electronic devices based on multi-dimensional profile linkage to solve the technical problems mentioned in the background section above.

[0010] Firstly, some embodiments of this disclosure provide an intelligent decision-making method based on multi-dimensional profile linkage. This method includes: collecting multi-dimensional business data, wherein the multi-dimensional business data includes: user datasets, product datasets, sales datasets, organizational resource datasets, and cooperation datasets; using the multi-dimensional business data to construct a five-dimensional profile model, wherein the five-dimensional profile model includes: user profiles, product profiles, sales profiles, organizational resource profiles, and cooperation profiles; constructing a dynamic decision matrix based on the five-dimensional profile model; and generating intelligent decision information based on the dynamic decision matrix.

[0011] Secondly, some embodiments of this disclosure provide an intelligent decision-making device based on multi-dimensional profile linkage. The device includes: an acquisition unit configured to collect multi-dimensional business data, wherein the multi-dimensional business data includes: user datasets, product datasets, sales datasets, organizational resource datasets, and cooperation datasets; a first construction unit configured to construct a five-dimensional profile model using the multi-dimensional business data, wherein the five-dimensional profile model includes: user profiles, product profiles, sales profiles, organizational resource profiles, and cooperation profiles; a second construction unit configured to construct a dynamic decision matrix based on the five-dimensional profile model; and a decision generation unit configured to generate intelligent decision information based on the dynamic decision matrix.

[0012] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0013] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0014] The above-described embodiments of this disclosure have the following beneficial effects: 1. Breakthrough in decision-making: Establishing the industry's first joint simulation sandbox covering users, products, methods, resources, and partners, improving the completeness of decision-making basis by 300%; 2. Response mechanism reconstruction: Compressing strategy generation and resource scheduling speed to the second level, overcoming the industry-wide problem of "decision-making not keeping up with changes"; 3. Improvement of resource entropy reduction: By using a capability vectorization matching model, the resource mismatch rate is reduced from 37% to 6%, saving enterprise operating costs; 4. Unleashing Ecological Value: Building a cross-enterprise intelligent collaborative network increases the value contributed by ecosystem partners by 2.8 times. Attached Figure Description

[0015] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0016] Figure 1 This is a flowchart of some embodiments of the intelligent decision-making method based on multi-dimensional profile linkage according to the present disclosure; Figure 2This is an architecture diagram of an intelligent decision-making system; Figure 3 This is a schematic diagram of the triple innovation results; Figure 4 This is a diagram of a strategy evolutionary neural network architecture; Figure 5 This is a schematic diagram of the structure of some embodiments of the intelligent decision-making device based on multi-dimensional profile linkage according to the present disclosure; Figure 6 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0017] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0018] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0019] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0020] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0021] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0022] Before performing any of the operations involving the collection, storage, and use of user personal information (such as user profiles and user historical behavior) disclosed in this disclosure, the relevant organizations or individuals shall fulfill their obligations, including conducting personal information security impact assessments, informing personal information subjects, and obtaining prior authorization and consent from personal information subjects.

[0023] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] Figure 1 A flowchart 100 is shown, illustrating some embodiments of the intelligent decision-making method based on multi-dimensional profile linkage according to this disclosure. This intelligent decision-making method based on multi-dimensional profile linkage, applied to an intelligent decision-making system, includes the following steps: Step 101: Collect multi-dimensional business data.

[0025] In some embodiments, the executor of the intelligent decision-making method based on multi-dimensional profile linkage (e.g., a computing device) can collect multi-dimensional business data corresponding to the target project via wired or wireless means. The aforementioned intelligent decision-making system includes a data acquisition layer, a multi-profile collaboration engine, and a strategy instruction distribution module. Here, the data acquisition layer employs multi-source heterogeneous fusion technology to construct a data lake supporting real-time and batch processing, providing comprehensive and high-quality data fuel for upper-layer intelligent applications. Its collection scope covers internal, external, structured, and unstructured data. Structured data may include product flow records, etc. Unstructured data may include personnel relationship data, unstructured documents, etc. Therefore, multi-dimensional business data can be extracted from the data lake.

[0026] As an example, internal data may include, but is not limited to, at least one of the following: core business data (e.g., customer profiles, contacts, business opportunity information, activity records, order information, contract documents, product catalogs, invoices, customer service system work orders, resolution records, customer feedback, etc.), behavioral data operation logs (e.g., number of times customers are viewed, frequency of field modifications), internal collaboration information (e.g., internal chat logs and email exchanges related to a business opportunity topic), etc. External data includes: publicly available network data (e.g., business registration information, public opinion information, recruitment information, macroeconomic indicators, industry reports, industrial policies, official website updates, product release information, market activity information, etc.).

[0027] Secondly, the multi-profile collaboration engine serves as the system's intelligent decision-making hub. Through a dynamic coupling algorithm, it integrates five-dimensional profile data—user, product, sales method, organizational resources, and partner—in real time. The engine first calculates the real-time weight of each profile using a context-aware module (e.g., opportunity stage, competition intensity), then invokes a correlation analysis model to uncover deep connections between profiles (e.g., the matching degree between customer technology preferences and product characteristics). Finally, a decision model trained on reinforcement learning generates optimal strategy solutions, including precise strategy recommendations, resource allocation instructions, and risk warnings, transforming multi-dimensional data into actionable sales actions and achieving a closed loop from "data insight" to "decision-driven" decision-making.

[0028] Finally, the strategy instruction distribution module generates intelligent decision information and distributes the generated intelligent decision information to the corresponding execution terminals.

[0029] As an example, see Figure 2 The diagram shows the architecture of the intelligent decision-making system. Here, data is collected through a data acquisition layer and input into a multi-profile collaboration engine to construct a five-dimensional profile model, including modules for generating user profiles, product profiles, sales profiles, organizational resource profiles, and collaboration profiles. Then, combining the user profiles, product profiles, sales profiles, organizational resource profiles, and collaboration profiles from the five-dimensional profile model, a dynamic decision matrix is ​​constructed to generate intelligent decision information, which is then sent to the corresponding execution terminals via a strategy instruction distribution module.

[0030] It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (Ultra Wide Band) connections, and other currently known or future wireless connection methods.

[0031] It should be noted that the aforementioned computing devices can be either hardware or software. When the computing device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. When the computing device is software, it can be installed on the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here.

[0032] Step 102: Utilize multi-dimensional business data to construct a five-dimensional profile model.

[0033] In some embodiments, the aforementioned implementing entity can utilize the multi-dimensional business data to construct a five-dimensional profile model. This five-dimensional profile model includes: user profile, product profile, sales profile, organizational resource profile, and cooperation profile. The user profile can be used to represent user characteristics. The product profile corresponds to product information. The sales profile corresponds to historical sales strategies. The organizational resource profile can be used to represent the organization's technical resources, such as the number of engineers and their technical skill levels. The cooperation profile describes the data profiles of cooperating units.

[0034] In some optional implementations of certain embodiments, the aforementioned executing entity utilizes the aforementioned multi-dimensional business data to construct a five-dimensional profile model, including: Step S1 involves employing a graph neural network algorithm to perform representation learning on the user dataset included in the aforementioned multidimensional business data, thereby generating an influence embedding vector set. The user data includes user enterprise attributes, historical interaction behavior sequences, and external public opinion data. Here, user attributes in the user dataset can be used as nodes, and the relationships between user attributes as edges, inputting them into the graph neural network algorithm (GraphSAGE) to perform representation learning on the decision chain relationship network within the user, outputting the influence embedding vector for each node.

[0035] Step S2 involves constructing the user profile in the aforementioned five-dimensional profile model using the extreme gradient boosting tree model, based on the user dataset and the aforementioned influence embedding vector group. Specifically, the XGBoost extreme gradient boosting tree model can be used to predict demand and provide churn risk warnings using user enterprise attributes, historical interaction behavior sequences, and external public opinion data as features, yielding demand information and churn risk information. Finally, each user, along with their corresponding demand information and churn risk information, can be defined as the user profile.

[0036] Optionally, the aforementioned executing entity may further construct a five-dimensional profile model using the aforementioned multi-dimensional business data, including: Step S1: Construct a product knowledge graph using the product dataset from the aforementioned multidimensional business data. Entities in the product knowledge graph include product identifiers, functional information, parameters, and application scenario information. Relationships between entities can be described using terms such as "superior to," "contains," and "applies to."

[0037] Step S2 involves using a semantic text embedding model to perform structural analysis on the aforementioned product knowledge graph, generating product profiles within the five-dimensional profile model. Specifically, the semantic text embedding model can be used to semantically identify unstructured data (e.g., product identifiers, functional information, parameters, or unstructured documents) corresponding to each product in the product knowledge graph, transforming them into semantic vectors. These semantic vectors are then used to calculate the semantic similarity between the product and customer needs. Finally, the product profiles are constructed by combining each product in the knowledge graph with its corresponding semantic vectors.

[0038] Here, the semantic text embedding model can be a sentence embedding model optimized based on BERT (Bidirectional Encoder Representations from Transformers).

[0039] Optionally, the aforementioned executing entity may further construct a five-dimensional profile model using the aforementioned multi-dimensional business data, including: Step S1: Select sales data that meet preset data conditions from the sales dataset of the aforementioned multidimensional business data, and use this as noise data to obtain the noise dataset. The preset data conditions can be sales data in the sales dataset that includes sales success indicators. Here, sales data can include: customer context information, customer strategy information, and sales transaction results.

[0040] Step S2 involves using a Transformer encoder to extract features from the historical strategy text to generate strategy features. The target strategy text includes historical sales strategy information.

[0041] Step S3: Using the aforementioned noisy dataset, the aforementioned policy features, and the pre-trained generative adversarial network (GAN), a sales profile in the aforementioned five-dimensional profile model is generated. The aforementioned GAN includes a generator and a discriminator. The aforementioned noisy dataset is the input to the generator, the aforementioned policy features are the input to the discriminator, and the aforementioned sales profile includes policy feature values ​​corresponding to each of the at least one sales strategy. Here, the generator is used to generate new sales strategies based on the current data. The discriminator is used to determine whether each new sales strategy is close to the noisy data, i.e., whether it is close to a successful case. Thus, through adversarial training and optimization, at least one sales strategy and its corresponding policy feature values ​​can be generated.

[0042] Optionally, the aforementioned executing entity may further construct a five-dimensional profile model using the aforementioned multi-dimensional business data, including: Step S1 involves performing multi-dimensional feature vectorization on each organizational resource data point in the aforementioned multi-dimensional business data organizational resource dataset to generate an organizational resource feature vector set. The organizational resource data may include information such as engineer skills, engineer experience, and historical project resources. Here, each organizational resource feature vector can correspond to a project or a collaborating unit.

[0043] Step S2: Based on the dense deep neural network and the aforementioned organizational resource feature vector set, generate the organizational resource profile in the five-dimensional profile model. This organizational resource profile includes a comprehensive capability vector. Here, the comprehensive capability vector can be obtained by learning the non-linear relationship between various organizational resource features and project success rates in the organizational resource feature vector set through the dense deep neural network. Furthermore, the organizational resource data corresponding to each unit or project can be combined with the comprehensive capability vector to construct the organizational resource profile.

[0044] Optionally, the aforementioned executing entity may further construct a five-dimensional profile model using the aforementioned multi-dimensional business data, including: Step S1 involves using a collaborative filtering algorithm to filter the collaborative datasets included in the aforementioned multidimensional business data to generate similar datasets. Specifically, collaborative filtering identifies collaborating units with similar user groups to the current organization, thus obtaining similar datasets. Here, similar data may include user information from collaborating units with similar user groups.

[0045] Step S2 involves constructing the collaboration profile within the five-dimensional profile model using a logistic regression model and the aforementioned similar dataset. The logistic regression model calculates the joint winning probability for each similar data point in the dataset. Here, the joint winning probability represents the probability of the unit and its collaborating units jointly winning a bid. Furthermore, the collaboration identifier and corresponding joint winning probability for each similar data point can be used to define the collaboration profile.

[0046] Step 103: Construct a dynamic decision matrix based on the five-dimensional profile model.

[0047] In some embodiments, the aforementioned executing entity may construct a dynamic decision matrix based on the aforementioned five-dimensional profiling model.

[0048] In some optional implementations of certain embodiments, the aforementioned execution entity constructs a dynamic decision matrix based on the aforementioned five-dimensional profiling model, including: Step S1 involves dynamically calculating the importance weights of the user profile, product profile, sales profile, organizational resource profile, and collaboration profile in the aforementioned five-dimensional profile model using current project stage information and an attention mechanism, resulting in an importance weight reorganization. The current project stage information can be information about a project at the current moment. For example, it could include project requirements, project plans, and project stage identifiers.

[0049] Specifically, the attention mechanism acts as a "contextual perceiver," dynamically calculating the importance weights of each of the five dimensions in the five-dimensional profile model based on the current business opportunity stage (e.g., demand confirmation, solution demonstration, cooperation negotiation, etc.) and the intensity of competition, resulting in an importance weight reorganization. For example, during the cooperation negotiation stage, the weights of the "product profile" and the "partner profile" will proactively increase.

[0050] As an example, taking sales profiles as an example, the code for calculating the importance weights is as follows: def weight_calculator(context): urgency = context['opp_stage']* 0.3 complexity = (1 - context['product_fit'])* 0.4 return min(1, urgency + complexity) Here, `weight_calculator` represents the dynamic calculation model for profile weights. `opp_stage` represents the opportunity stage. `context` represents the current project stage information. `product_fit` represents product fit. `urgency` represents urgency. `complexity` represents complexity. Furthermore, for user profiles, product profiles, organizational resource profiles, and collaboration profiles, the same code structure (with different parameters) can be used to generate corresponding importance weights.

[0051] Step S2: Based on the aforementioned importance weighting, the five-dimensional profile model, and the pre-trained deep reinforcement learning model, a dynamic decision matrix is ​​constructed. First, the importance weights in the importance weighting are weighted with their corresponding profiles to obtain weighted profile information. Then, the sales process is modeled as a Markov decision process. The state of the deep reinforcement learning model (Deep Q-Network, DQN) is a business opportunity state vector incorporating the weighted profile information; the actions are a set of strategies recommended by the intelligent decision-making system; and the reward is the final transaction result and efficiency based on the business opportunity state vector. By training the deep reinforcement learning model, the action with the highest expected reward in a specific state is generated. Finally, the expected rewards corresponding to each strategy are constructed into a dynamic decision matrix.

[0052] Step 104: Generate intelligent decision information based on the dynamic decision matrix.

[0053] In some embodiments, the aforementioned executing entity can generate intelligent decision information based on the aforementioned dynamic decision matrix.

[0054] In some optional implementations of certain embodiments, the execution entity generates intelligent decision information based on the dynamic decision matrix, including: Step S1 involves using the aforementioned dynamic decision matrix and importance weighting to determine the deep nonlinear relationships among the user profile, product profile, sales profile, organizational resource profile, and cooperation profile in the five-dimensional profile model, and performing real-time feature cross-calculation to obtain the associated feature vector. Specifically, a policy evolutionary neural network is used to determine the deep nonlinear relationships among these five dimensions and to perform real-time feature cross-calculation to obtain the associated feature vector. Here, the policy evolutionary neural network can be another GAN network with the same structure but different parameters.

[0055] Step S2: Generate intelligent decision features based on the aforementioned associated feature vectors.

[0056] Step S3 involves using the knowledge bases corresponding to the large language model and the five-dimensional profile model to determine the intelligent decision-making information corresponding to the aforementioned intelligent decision-making features. Specifically, the strategy instruction distribution module can be pre-configured with a large language model interface, allowing the intelligent decision-making features to be sent to this interface and the knowledge base corresponding to the five-dimensional profile model to generate dialogue-based intelligent decision-making information.

[0057] In practice, the dynamic decision matrix is ​​the core processing unit of the multi-profile collaborative engine. Its working principle can be likened to a high-speed, self-adjusting precision dashboard. It is not a simple weighted average, but rather performs collaborative calculations through a three-layer processing mechanism: First, the context awareness layer: uses attention mechanisms to capture the current business opportunity status (such as stage and competition intensity) in real time, and dynamically adjusts the weight coefficients of the input data of each profile (for example, significantly increasing the weight of terms and delivery capabilities in the "product profile" and "partner profile" during the negotiation stage).

[0058] Secondly, the association computation layer: This layer uses a policy evolution neural network to uncover deep non-linear relationships between user profiles (e.g., identifying a strong correlation between the academic background of a decision-maker in a "customer profile" and a specific technical parameter in a "product profile"), and performs feature cross-computation in real time. For example, it uses a logistic regression model to associate user features.

[0059] Finally, the decision generation layer integrates all weighted and correlated feature vectors and outputs atomic-level strategy instructions, including the optimal decision path, resource scheduling scheme, and risk warning. This transforms multidimensional data into precise and executable action strategies, enabling intelligent decision-making and generating the aforementioned intelligent decision features.

[0060] Here, the collaborative decision-making formula for profiles is as follows: decision_score = (user_profile × w1 + product_profile × w2 +sales_method × w3 + org_resource ×w4+partner x w5user_profile × w1 + product_profile × w2 + sales_method × w3 +org_resource × w4 + partner × w5 ×weight_calculator(ctx) Here, `decision_score` represents the decision score. `user_profile` represents the user profile. `product_profile` represents the product profile. `w1`, `w2`, `w3`, `w4`, and `w5` represent weight coefficients. `sales_method` represents the sales strategy. `org_resource` represents the organizational resource profile. `partner` represents the partnership profile. `weight_calculator(ctx)` represents the weight calculator.

[0061] Optionally, the aforementioned execution entity can also match the best resources for the user using resource fusion technology upon receiving a scheduling request. Here, resource fusion technology refers to the technology of integrating various user data. Cross-system resource fusion technology can be implemented in the following ways: 1. Transform organizational capabilities into a comprehensive capability vector using a resource encoder. For example, organizational capabilities include: [skill tags, project experience, response speed]. The resulting comprehensive capability vector would be: [0.78, -0.23, 1.2]. Here, the resource encoder can be used to convert the individual attribute values ​​of organizational capabilities into vector parameters to obtain the comprehensive capability vector.

[0062] 2. Map customer needs to the resource space. This involves mapping the comprehensive capability vector and the demand vector generated from business opportunity needs to the same high-dimensional vector space. Here, the demand vector can be generated based on scheduling requests. For example, for user demand attributes in a scheduling request, the corresponding attribute values ​​are combined into a demand vector according to a preset attribute order. In practice, the demand vector represents the demand description of a user or user group.

[0063] 3. Dynamically match the best resources based on cosine similarity. This involves using an approximate nearest neighbor search algorithm to select a number of comprehensive capability vectors with the highest cosine similarity to the aforementioned demand vectors from the mapped vector space, thus obtaining a resource vector group.

[0064] As an example, an approximate nearest neighbor search algorithm could be an HNSW (Hierarchical Navigable SmallWorld, an index structure for approximate nearest neighbor search) graph algorithm.

[0065] Finally, the strategy instruction distribution module can be connected to generate intelligent decision-making information corresponding to the resource vector groups using the large language model and the aforementioned resource vector groups. Specifically, the resource vector groups can be input into a pre-selected large language model, combined with the five-dimensional profile model and the corresponding knowledge base, to output intelligent decision-making information.

[0066] Technical effect See Figure 3and Figure 4 This application achieves three major technological breakthroughs through triple innovation: a multi-profile collaborative engine (92% decision accuracy), a strategy evolutionary neural network (0.8s response time), and resource fusion technology (95% utilization rate). The most crucial connection lies in Figure 4 The feedback loop within the system. The results of each sales action (historical sales cases, real-time session data, efficiency levels, resource utilization effectiveness, etc.) are fed back as feedback data to the feature extractor, and failed cases are input into the discriminator. 1. Feedback to the "Policy Evolutionary Neural Network": Success and failure cases are used for continuous training and optimization of the model.

[0067] 2. Feedback to "Resource Fusion Technology": Performance data of resources in performing tasks (such as customer satisfaction and task completion time) is used to update their resource encoders, making future matching more accurate.

[0068] The three innovations in this application constitute a tightly coupled, closed-loop, self-evolving intelligent decision-making system. The multi-profile collaborative engine is the system's brain, responsible for real-time fusion and interpretation of multi-data from customers, products, methods, resources, and partners, outputting deep insights and decision directions for the current business opportunity situation. The strategy evolution neural network receives decision instructions from the multi-profile collaborative engine and, based on historical success cases and real-time adversarial training (GAN architecture), generates specific, executable, and refined strategy solutions (i.e., generated strategies). This increases the business opportunity conversion rate by 25%. Consequently, it can reduce operating costs by approximately 30%, and further increase the ecosystem value by 2.8 times.

[0069] These strategic solutions are then passed to resource fusion technology for execution. This technology rapidly matches strategic requirements (such as "a pre-sales engineer proficient in artificial intelligence") with the vectorized capability tags of internal resources, enabling precise talent allocation and deployment.

[0070] Ultimately, the performance data of the solution (success or failure, resource utilization, etc.) will serve as feedback signals, flowing back in real time to the multi-profile collaborative engine and the strategy evolution neural network to optimize profile weight allocation and iterate the strategy model. This forms a continuously self-optimizing closed-loop intelligent agent from "perception-decision-execution-feedback". The three work together to ensure the accuracy of intelligent decision-making.

[0071] The above-described embodiments of this disclosure have the following beneficial effects: 1. Breakthrough in Decision-Making Dimensions: Establishing the industry's first joint simulation sandbox covering users, products, methods, resources, and partners, improving the completeness of decision-making basis by 300%. Panoramic Decision-Making: Constructing the industry's first joint simulation sandbox covering users, products, methods, resources, and partners, breaking through the limitations of traditional single-point profiling systems: the accuracy of business opportunity assessment increased from 63% to 92% (an increase of 46%), the error in demand prediction decreased from >40% to <12% (a decrease of 70%), and the success rate of ecosystem collaboration increased by 62% (55% → 89%).

[0072] 2. Response Mechanism Restructuring: Compressing strategy generation and resource scheduling speed to the second level, overcoming the industry-wide problem of "decision-making failing to keep up with changes." Instantaneous Response: Overcoming the industry-wide problem of strategy lag, achieving a second-level decision-making closed loop. Strategy generation speed: 5.2s → 0.8s (84% improvement), resource scheduling latency: 52 hours → 5 seconds (99.97% reduction), strategy iteration cycle: 30 days → real-time self-evolution.

[0073] 3. Improvement in Resource Entropy Reduction: By using a capability vectorized matching model, the resource mismatch rate is reduced from 37% to 6%, saving enterprise operating costs. Zero Resource Loss: By eliminating mismatch losses through capability vectorized matching, the ultimate goal is to establish an intelligent decision-making paradigm of "panoramic insight → instant response → zero-loss scheduling," creating a quantifiable generational advantage for enterprises in digital transformation. Pilot testing has verified that this reduces decision bias by 37%, reduces response latency by 99%, and improves resource efficiency by 51%.

[0074] 4. Unleashing Ecosystem Value: Building a cross-enterprise intelligent collaborative network increases the value contributed by ecosystem partners by 2.8 times. Evolution Mechanism: A new strategy version is automatically generated every 200 business opportunity interactions.

[0075] Further reference Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of an intelligent decision-making device based on multi-dimensional profile linkage. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, this intelligent decision-making device based on multi-dimensional profile linkage can be specifically applied to various electronic devices.

[0076] like Figure 5As shown, an intelligent decision-making device 500 based on multi-dimensional profile linkage in some embodiments includes: an acquisition unit 501, a first construction unit 502, a second construction unit 503, and a decision generation unit 504. The acquisition unit 501 is configured to collect multi-dimensional business data, including user datasets, product datasets, sales datasets, organizational resource datasets, and cooperation datasets. The first construction unit 502 is configured to use the multi-dimensional business data to construct a five-dimensional profile model, including user profiles, product profiles, sales profiles, organizational resource profiles, and cooperation profiles. The second construction unit 503 is configured to construct a dynamic decision matrix based on the five-dimensional profile model. The decision generation unit 504 is configured to generate intelligent decision information based on the dynamic decision matrix.

[0077] It is understandable that the various units and references recorded in the intelligent decision-making device 500 based on multi-dimensional profile linkage are... Figure 5 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method are also applicable to the intelligent decision-making device 500 based on multi-dimensional profile linkage and the units contained therein, and will not be repeated here.

[0078] The following is for reference. Figure 6 It illustrates a schematic diagram of the structure of an electronic device (such as a computing device) suitable for implementing some embodiments of the present disclosure. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of this disclosure. Figure 6 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory may include a non-volatile storage medium and internal memory. The non-volatile storage medium may store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any of the methods described above. The processor provides computational and control capabilities to support the operation of the entire computer device. The internal memory provides an environment for the execution of the computer program in the non-volatile storage medium; when executed by the processor, the computer program causes the processor to perform any of the methods described above. The network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the computer device to which the present disclosure is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0079] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0080] In one embodiment, the processor is configured to run a computer program stored in a memory to perform the following steps: collecting multi-dimensional business data, wherein the multi-dimensional business data includes: user datasets, product datasets, sales datasets, organizational resource datasets, and cooperation datasets; using the multi-dimensional business data, constructing a five-dimensional profile model, wherein the five-dimensional profile model includes: user profiles, product profiles, sales profiles, organizational resource profiles, and cooperation profiles; constructing a dynamic decision matrix based on the five-dimensional profile model; and generating intelligent decision information based on the dynamic decision matrix.

[0081] This disclosure also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, and the method implemented when the program instructions are executed can be referred to the various embodiments of the methods described above.

[0082] The aforementioned computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as a hard disk or memory of the computer device. Alternatively, the aforementioned computer-readable storage medium may be an external storage device of the computer device, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0083] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0084] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. An intelligent decision-making method based on multi-dimensional profile linkage, applied to an intelligent decision-making system, characterized in that, include: Collect multidimensional business data, which includes: user datasets, product datasets, sales datasets, organizational resource datasets, and cooperation datasets; Using the aforementioned multidimensional business data, a five-dimensional profile model is constructed, which includes: user profile, product profile, sales profile, organizational resource profile, and cooperation profile. Based on the aforementioned five-dimensional profiling model, a dynamic decision matrix is ​​constructed; Intelligent decision-making information is generated based on the dynamic decision matrix.

2. The method according to claim 1, characterized in that, The process of constructing a five-dimensional profile model using the multi-dimensional business data includes: A graph neural network algorithm is used to perform representation learning on the user dataset included in the multidimensional business data to generate an influence embedding vector set, wherein the user data includes user enterprise attributes, historical interaction behavior sequences and external public opinion data; Using an extreme gradient boosting tree model, a user profile is constructed in the five-dimensional profile model based on the user dataset and the influence embedding vector group.

3. The method according to claim 2, characterized in that, The method of constructing a five-dimensional profile model using the multi-dimensional business data also includes: Using the product dataset from the multidimensional business data, a product knowledge graph is constructed; Using a semantic text embedding model, the product knowledge graph is structurally analyzed to generate product profiles in the five-dimensional profile model.

4. The method according to claim 3, characterized in that, The method of constructing a five-dimensional profile model using the multi-dimensional business data also includes: Sales data that meet preset data conditions are selected from the sales dataset in the multidimensional business data and used as noise data to obtain a noise dataset; The target strategy text is extracted using a Transformer encoder to generate strategy features, wherein the target strategy text includes at least one sales strategy. Using the noise dataset, the policy features, and the pre-trained generative adversarial network, a sales profile in the five-dimensional profile model is generated. The generative adversarial network includes a generator and a discriminator. The noise dataset is the input to the generator, the policy features are the input to the discriminator, and the sales profile includes policy feature values ​​corresponding to each of at least one sales strategy.

5. The method according to claim 4, characterized in that, The method of constructing a five-dimensional profile model using the multi-dimensional business data also includes: Multidimensional feature vectorization is performed on each organizational resource data in the organizational resource dataset of the multidimensional business data to generate an organizational resource feature vector set; Based on the dense deep neural network and the organizational resource feature vector set, an organizational resource profile is generated in the five-dimensional profile model, wherein the organizational resource profile includes a comprehensive capability vector.

6. The method according to claim 5, characterized in that, The method of constructing a five-dimensional profile model using the multi-dimensional business data also includes: The collaborative filtering algorithm is used to filter the cooperative datasets included in the multidimensional business data to generate similar datasets. Using a logistic regression model and the similar dataset, a collaborative profile is constructed in the five-dimensional profile model.

7. The method according to claim 6, characterized in that, The construction of a dynamic decision matrix based on the five-dimensional profiling model includes: By utilizing information from the current project phase and attention mechanisms, the importance weights of the user profile, product profile, sales profile, organizational resource profile, and cooperation profile in the five-dimensional profile model are dynamically calculated, resulting in an importance weight reorganization. Based on the aforementioned importance weight reorganization, the aforementioned five-dimensional profile model, and the pre-trained deep reinforcement learning model, a dynamic decision matrix is ​​constructed.

8. The method according to claim 7, characterized in that, The step of generating intelligent decision information based on the dynamic decision matrix includes: Using the dynamic decision matrix and the importance weight reorganization, the deep nonlinear relationship between user profile, product profile, sales profile, organizational resource profile and cooperation profile in the five-dimensional profile model is determined, and feature cross calculation is performed in real time to obtain the associated feature vector; Based on the associated feature vector, intelligent decision features are generated; By utilizing the knowledge bases corresponding to the large language model and the five-dimensional profile model, the intelligent decision-making information corresponding to the intelligent decision-making features is determined.

9. An intelligent decision-making device based on multi-dimensional profile linkage, applied to an intelligent decision-making system, characterized in that, include: The acquisition unit is configured to collect multidimensional business data, which includes: user datasets, product datasets, sales datasets, organizational resource datasets, and cooperation datasets. The first construction unit is configured to use the multi-dimensional business data to construct a five-dimensional profile model, wherein the five-dimensional profile model includes: user profile, product profile, sales profile, organizational resource profile, and cooperation profile. The second construction unit is configured to construct a dynamic decision matrix based on the five-dimensional profile model; The decision generation unit is configured to generate intelligent decision information based on the dynamic decision matrix.

10. An electronic device, characterized in that, include: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-8.

Images