Multi-asset class real-time streaming intelligent investment advisory system

Abstract

This novel multi-asset class real-time streaming intelligent investment advisory system includes a processor, memory, a communication interface, and a front-end presentation module. The memory stores multiple agent modules, including at least a traffic distribution agent module, an analysis agent module, and a dialogue agent module. In different embodiments, the system can further coordinate and optimize agent modules, comparison agent modules, planning agent modules, and pluggable data access modules to perform hybrid intent classification, real-time streaming response, in-depth parallel analysis, cross-asset data normalization, order execution, and asset allocation monitoring and rebalancing alerts, thereby improving investment analysis efficiency, real-time interactivity, and multi-asset scalability.

Claims

1. A multi-asset class real-time streaming intelligent investment advisory system, comprising: One processor; A memory coupled to the processor and storing an instruction set and multiple modules; a communication interface coupled to the processor and configured to maintain a connection with a client device for bidirectional transmission of messages. A front-end presentation module, coupled to the processor, is configured to present an interactive interface to a user; wherein the plurality of modules stored in the memory include at least: a routing agent module configured to receive user input, perform pre-classification of the user input, and generate guided interactive data under specific conditions; an analysis agent module configured to classify an intent of the user input using an intent classification mechanism and output an intent category and related parameters; and a dialogue agent module configured to generate a response in a streaming manner based on the result of the intent classification and based on financial product information of different asset classes and the conversation context.

2. The multi-asset class real-time streaming intelligent investment advisor system as described in claim 1, wherein the memory further stores an optimized agent module configured to perform rule-based format optimization on the response generated by the dialogue agent module.

3. The multi-asset class real-time streaming smart investment advisor system as described in claim 1, wherein the memory further stores a comparison agent module configured to perform parallel comparison analysis on multiple financial products across different asset classes.

4. The multi-asset class real-time streaming intelligent investment advisory system as described in claim 1, wherein the memory further stores a planning agent module configured to break down a complex query into multiple subtasks for parallel execution and to merge the results of such subtasks.

5. The multi-asset class real-time streaming smart investment advisory system as described in claim 1, wherein the memory further stores a pluggable data adapter module configured to normalize heterogeneous data related to different asset classes into a unified data format through a unified data interface and multiple asset-specific adapters.

6. The multi-asset class real-time streaming intelligent investment advisory system as described in claim 1, wherein the intent classification mechanism used by the analysis agent module is a hybrid intent classification mechanism, comprising: The first stage of semantic analysis is performed by a large language model to produce a preliminary classification result; The second stage of correction involves a rule-based classification and correction engine performing a logical verification and parameter completion procedure to correct the consistency and completeness of the initial classification results. It also includes a fallback mechanism configured to switch to a rule-only classification mode when the large language model is unavailable.

7. The multi-asset class real-time streaming smart investment advisor system as described in claim 1 further includes a dialogue history compression module, which is configured to extract features and reduce the dimensionality of historical dialogue data that exceeds preset conditions based on a resource control mechanism, generate a simplified data structure to retain the user's core parameters, and integrate the simplified data structure into the context of subsequent processing procedures.

8. The multi-asset class real-time streaming intelligent investment advisory system as described in claim 7, wherein, This streamlined data structure retains at least one of the following: financial entity identification code, user investment profile, and system inference trajectory.

9. The multi-asset class real-time streaming intelligent investment advisory system as described in claim 4, wherein the planning agent module is configured to: execute a task planning phase, wherein the complex query is broken down into multiple sub-tasks, and each sub-task is assigned an agent type and an analysis direction; execute a parallel execution phase, wherein the multiple sub-tasks are executed in parallel, and task status information is pushed to the front-end presentation module; and execute a streaming integration phase, wherein the results of the multiple sub-tasks are cross-analyzed and streamed as a deep analysis report to the front-end presentation module.

10. The multi-asset class real-time streaming intelligent investment advisory system as described in claim 5, wherein the pluggable data connection module includes: The unified data interface defines a standardized data structure; multiple asset-specific adapters are configured to connect to different external data sources and normalize heterogeneous data into the standardized data structure; and an adapter manager is configured to register a new asset-specific adapter to extend the system to a new asset class.

11. The multi-asset class real-time streaming smart investment advisor system as described in claim 1 further includes an intent-driven data delayed loading module, which is configured to load asset-related data only for a selected intent class, dynamically determine a loading range based on an analysis requirement, and cooperate with a caching mechanism to reduce response latency.

12. The multi-asset class real-time streaming smart investment advisory system as described in claim 1 further includes an order execution module configured to convert an analysis suggestion into a structured order preview and, after user confirmation and security verification, transmit an order instruction to a brokerage interface for execution, wherein the order execution module supports multiple order types.

13. The multi-asset class real-time streaming smart investment advisory system as described in claim 1 further includes an asset allocation monitoring and rebalancing module, configured to: store a user-defined target asset allocation ratio; continuously monitor an actual holding allocation ratio; determine whether the actual holding allocation ratio deviates from the target asset allocation ratio by more than a preset threshold; and when the preset threshold is exceeded, generate a rebalancing reminder and an adjustment suggestion, the adjustment suggestion including at least one of a target asset, an adjustment quantity, an estimated cost, and an adjusted risk-reward characteristic.

14. The multi-asset class real-time streaming intelligent investment advisory system as described in claim 1, further comprising: A visualization processing submodule is configured to display a dynamic processing animation based on progress information received from the communication interface, and to display independent progress animations corresponding to each subtask during parallel execution; and an authentication and auditing module is configured to perform scepter-based authentication, role-based access control, password hashing, and auditing log recording of dialogue exchanges and order placement operations.

Images