Data visualization method and device based on fused large model, equipment and medium

Abstract

The invention discloses a data visualization method and device based on a fused large model, equipment and a medium. The method comprises the following steps: receiving a query statement input by a user on an interaction page, and performing vectorization processing by utilizing an embedded function to obtain a target feature vector; inputting the target feature vector into a pre-deployed knowledge vector database, and obtaining a target knowledge vector most similar to the target feature vector by calculating cosine similarity retrieval; jointly inputting the target feature vector and the target knowledge vector into a pre-established fusion large model platform to obtain target result data; and outputting a visual chart of a specified type for the target result data based on user requirements for a user to carry out data analysis. According to the embodiment of the invention, the data analysis platform with fusion of various models is established, so that the user directly dialogues the data in a question and answer mode of character input, and the result data is dynamically displayed in a chart form, so that the user is helped to intuitively understand a relationship and a mode among the data.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a data visualization method, device, equipment and medium based on a fusion large model. Background Art

[0002] The methods and technologies of traditional anti-money laundering data analysis platforms rely on the expert model experience accumulated over many years of work experience, and adopt the method of task distribution to conduct human-wave tactical identification. With the explosive growth of financial business, the ability to process massive data is limited. Data from different upstream systems and in different formats are difficult to effectively integrate, and it is difficult to identify the hidden nature of transaction links. The data is poorly displayed intuitively, resulting in limited breadth and depth of data analysis.

[0003] At present, the existing anti-money laundering visual data analysis platform mainly processes data manually according to business scenario requirements, and then displays it statically in specific visual charts, or exports data for manual analysis of charts. Visual charts cannot be dynamically adjusted according to user dynamic needs, and updates are not timely. When users are analyzing, there is a lack of real-time dynamic adjustment of chart visualization styles, and chart visualization cannot be dynamically displayed according to multiple dimensions. Summary of the invention

[0004] Based on this, the present invention provides a data visualization method, device, equipment and medium based on a fusion large model to solve the problems of poor processing effect and lack of flexibility of the existing visualization data analysis platform.

[0005] In a first aspect, an embodiment of the present invention provides a data visualization method based on a fusion large model, the method comprising:

[0006] Receive the query statement entered by the user on the interactive page, perform vectorization processing using the embedding function, and obtain the target feature vector;

[0007] Input the target feature vector into a pre-deployed knowledge vector database, and retrieve the target knowledge vector that is most similar to the target feature vector by calculating the cosine similarity;

[0008] Input the target feature vector and the target knowledge vector into the pre-built fusion model platform to obtain the target result data;

[0009] Based on user needs, the target result data is output with a specified type of visualization chart for users to perform data analysis.

[0010] In a second aspect, an embodiment of the present invention provides a data visualization device based on a fusion large model, the device comprising:

[0011] The feature vector extraction module is used to receive the query statement input by the user on the interactive page, and perform vectorization processing using the embedded function to obtain the target feature vector;

[0012] A knowledge vector retrieval module is used to input the target feature vector into a pre-deployed knowledge vector database, and retrieve the target knowledge vector that is most similar to the target feature vector by calculating the cosine similarity;

[0013] A result data generation module is used to input the target feature vector and the target knowledge vector into the pre-built fusion model platform to obtain the target result data;

[0014] The visualization chart display module is used to output a specified type of visualization chart for the target result data based on user needs for users to perform data analysis.

[0015] In a third aspect, an embodiment of the present invention provides an electronic device, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a data visualization method based on a fused large model as described in any embodiment of the present invention.

[0019] In a fourth aspect, a computer-readable storage medium is also provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement a data visualization method based on a fused large model as described in any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present invention uses a large language model as the foundation to build a data analysis platform that integrates multiple models, allowing users to directly communicate with data through text-based question-and-answer methods, and using visual analysis technology to dynamically display the result data of user question-and-answer content in the form of charts and generate data reports, which helps users intuitively understand the relationship and pattern between data. Compared with traditional platforms that have the risk of data leakage when analyzing large amounts of sensitive data, the embodiment of the present invention improves data security and the accuracy of data analysis to a certain extent through unified data processing and model services.

[0021] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 is a flow chart of a data visualization method based on a fusion large model provided according to the first embodiment of the present invention;

[0024] Figure 2 is a flow chart of another data visualization method based on a fusion large model provided according to the second embodiment of the present invention;

[0025] Figure 3 is a structural schematic diagram of a data visualization device based on a fusion large model provided according to Embodiment 3 of the present invention;

[0026] Figure 4 It is a structural schematic diagram of an electronic device according to a data visualization method based on a fused large model provided in accordance with a fourth embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] Figure 1 This is a flowchart of a data visualization method based on a fusion big model provided in the first embodiment of the present invention. This embodiment can be applied to the case of displaying a visualization chart for query statement data input by a user. The method can be executed by a data visualization device based on a fusion big model. The device can be implemented in the form of hardware and / or software, and the device can be configured in a data analysis platform. Figure 1 As shown, the method includes:

[0031] S110, receiving a query statement input by a user on an interactive page, and performing vectorization processing using an embedded function to obtain a target feature vector.

[0032] Users describe their questions on the interactive page through text-based question-and-answer mode. The text data entered by the user is preprocessed through natural language interaction and passed to the Embedding function to convert the text into a vector. The target feature vector is the vector after the query statement is vectorized.

[0033] S120, inputting the target feature vector into a pre-deployed knowledge vector database, and retrieving a target knowledge vector that is most similar to the target feature vector by calculating cosine similarity.

[0034] The pre-deployed knowledge vector database may be a prompt template library, and the target knowledge vector is the prompt word template in the prompt template library that has the highest similarity to the target feature vector.

[0035] Optionally, before inputting the target feature vector into a pre-deployed knowledge vector database, the following steps may also be included:

[0036] Acquire multi-source heterogeneous data from different upstream systems, and classify the multi-source heterogeneous data into structured and unstructured data after data cleaning;

[0037] Perform feature extraction on the classified structured data and unstructured data to obtain structured data vectors and unstructured data vectors;

[0038] The structured data vectors and unstructured data vectors are indexed and enhanced using an artificial intelligence framework to obtain multiple knowledge vectors, which are then stored in a knowledge vector database.

[0039] In an embodiment of the present invention, when constructing a knowledge vector library, the user's multi-source heterogeneous files or multi-source heterogeneous data are obtained, and the data includes structured data and unstructured data, such as text or multi-type charts input by the user in history, and the structured data and unstructured data are preprocessed to ensure data quality, perform feature extraction according to specific dimensions, and vectorize the data. Among them, the artificial intelligence framework can be a RAG framework, which is used to establish a fast index link with an external database, use the RAG framework retrieval enhancement technology to improve the retrieval capability, and return the index knowledge vector.

[0040] S130, input the target feature vector and the target knowledge vector into the pre-built fusion large model platform to obtain the target result data.

[0041] The pre-built fusion large model platform is a multi-model fusion platform based on a large language model creatively proposed by the present invention. In the multi-model fusion platform, data of different topics are classified and graded, and the characteristics of different data topics are extracted to construct a labeling platform.

[0042] Furthermore, before the target feature vector and the knowledge vector are input into the pre-built fusion large model platform, the following steps may also be included:

[0043] Through the language model application construction framework, the natural language processing model, large-scale language model and multimodal model are integrated using a chain call strategy to obtain an initialized fusion large model;

[0044] All knowledge vectors are input into the initialization fusion big model for training, and the trained initialization fusion big model is registered into the intelligent routing distributor used to call the model service to construct a fusion big model platform.

[0045] Specifically, in an embodiment of the present invention, a natural language processing model, a large-scale language model, and a multimodal model are integrated, and a question-answering interaction method is provided for users by using distributed technology, artificial intelligence technology, data service bus technology, interface real-time scanning technology, etc. At the same time, the intelligent routing distributor can flexibly select the above models to optimize resource utilization. Among them, the chain call strategy is to decompose the task into several subtasks, and establish upstream and downstream relationships for several subtasks, and use the prompt words of the upstream subtask as input, and provide the obtained output result and the input of the upstream subtask together as the input of the downstream subtask to the model again, so as to enhance the reasoning ability of the fusion large model platform.

[0046] Optionally, the natural language processing model, the large-scale language model, and the multimodal model are integrated by using a chain call strategy through a language model application construction framework to obtain an initialized integrated large model, including:

[0047] Get APIs for natural language processing models, large-scale language models, and multimodal models respectively;

[0048] Using different interfaces, natural language processing models, large-scale language models, and multimodal models are chained together as components in the language model application construction framework to obtain multiple sequential chains.

[0049] All sequential chains are fused to obtain an initialized fused large model that provides different input and output interfaces.

[0050] In an embodiment of the present invention, the language model application construction framework may be a LangChain open source framework, in which the natural language processing model, the large-scale language model, and the multimodal model are respectively used as components with different functions to create chains. Among them, the intelligent distribution routing can dynamically select the upstream and downstream nodes in the chain combination through the interfaces of different components to obtain a dynamic sequence chain containing different upstream and downstream relationships. In the face of different input requirements, the intelligent distribution routing selects different sequence chains to perform different interaction processes, thereby generating the data results that are closest to the user input.

[0051] S140. Output a specified type of visualization chart for the target result data based on user needs, so that the user can perform data analysis.

[0052] The user's needs are determined based on the query statement entered by the user. If the user specifies the output chart type in the query statement, the data is visualized according to the chart type specified by the user. The chart type can be a histogram, a line chart, etc.

[0053] The embodiment of the present invention uses a large language model as the foundation to build a data analysis platform that integrates multiple models, allowing users to directly communicate with data through text-based question-and-answer methods, and using visual analysis technology to dynamically display the result data of user question-and-answer content in the form of charts and generate data reports, which helps users intuitively understand the relationship and pattern between data. Compared with traditional platforms that have the risk of data leakage when analyzing large amounts of sensitive data, the embodiment of the present invention improves data security and the accuracy of data analysis to a certain extent through unified data processing and model services.

[0054] Embodiment 2

[0055] Figure 2 This is a flowchart of another data visualization method based on a fusion large model provided in the second embodiment of the present invention. This embodiment is refined based on the above embodiment. Figure 2 As shown, the method specifically includes:

[0056] S210, receiving a query statement input by a user on an interactive page, and performing vectorization processing using an embedded function to obtain a target feature vector.

[0057] S220, inputting the target feature vector into a pre-deployed knowledge vector database, and retrieving a target knowledge vector that is most similar to the target feature vector by calculating cosine similarity.

[0058] S230, input the target feature vector and the target knowledge vector into the pre-built fusion large model platform to obtain the target result data.

[0059] Optionally, the target feature vector and the target knowledge vector are input into a pre-built fusion model platform to obtain target result data, which may include:

[0060] According to the target feature vector and target knowledge vector, the large model Agent is called through the intelligent routing distributor in the pre-built fusion large model platform;

[0061] Using the Agent intelligent body to identify keywords of the target knowledge vector, and split the target knowledge vector into multiple sub-vectors according to the keywords;

[0062] Inputting the multiple sub-vectors into the fusion large model platform to obtain multiple sub-task results corresponding to each sub-vector;

[0063] Send multiple subtask results to the intelligent routing distributor for result aggregation to obtain the target result data.

[0064] It should be noted that the intelligent routing distributor makes decisions on the next action through the Agent intelligent body and provides a certain model service by calling the corresponding data interface. Each sub-vector corresponds to a minimum unit of keyword, generates search results for each keyword, and aggregates the results through the intelligent distribution router. The target result data is the final search result of the target knowledge vector.

[0065] S240, convert the target output result into a format to obtain target message data in Json format, and identify chart type characteristic parameters in the target message data.

[0066] S250: Using a visual chart tool, determine a target chart type for the target message data based on chart type characteristic parameters.

[0067] S260: After assembling the target message data, a visual chart is obtained based on the determined target chart type, and the visual chart is rendered to a user interaction page.

[0068] Specifically, the visualization chart tool may be Plotly, Echars, BI, etc., and the embodiment of the present invention has no limitation to this. At the same time, the visualization chart may be displayed according to the chart type specified by the user.

[0069] Further, after assembling the target message data, a visualization chart is obtained based on the determined target chart type, which may specifically include:

[0070] If the time type keyword included in the query statement entered by the user on the interactive page is empty, the target message data is assembled and a static visualization chart is obtained based on the determined target chart type;

[0071] If the time type keyword included in the query statement input by the user on the interactive page is not empty, the target message data is assembled and a dynamic visualization chart is obtained based on the determined target chart type;

[0072] Among them, the dynamic visualization chart is dynamically updated according to the specified period.

[0073] The time type keyword may be "real-time", "same day", etc. When such keywords are involved in the user's query statement, the time type keyword is not empty. At this time, the data is collected and processed according to the specified period, and the front-end visualization chart is updated by dynamic plotting until the user exits the task, which is considered to be the end of this round of query task. On the contrary, if such keywords are not involved in the query statement, it means that the data required by the user is historical static data. At this time, the most matching static visualization chart is generated for the user according to the scale and theme of the data queried by the user, which means that this round of query task is considered to be completed.

[0074] The technical solution of the embodiment of the present invention, through the refinement of the overall solution, focuses on the generation of target result data and the display of visual charts. Specifically, by processing massive multi-source heterogeneous data in advance, a knowledge vector database is constructed, which can quickly integrate multi-source heterogeneous data channels, promote data sharing, and improve the timeliness of data processing and analysis. Furthermore, the embodiment of the present invention does not require users to master complex data analysis technology, and can generate visual charts that meet the requirements for users through question-and-answer operations, thereby improving the efficiency and accuracy of decision-making.

[0075] Embodiment 3

[0076] Figure 3 This is a schematic diagram of the structure of a data visualization device based on a fusion large model provided in the second embodiment of the present invention. Figure 3 As shown, the device comprises:

[0077] The feature vector extraction module 310 is used to receive the query statement input by the user on the interactive page, and perform vectorization processing using the embedded function to obtain the target feature vector;

[0078] A knowledge vector retrieval module 320 is used to input the target feature vector into a pre-deployed knowledge vector database, and retrieve a target knowledge vector that is most similar to the target feature vector by calculating cosine similarity;

[0079] The result data generation module 330 is used to input the target feature vector and the target knowledge vector into the pre-built fusion large model platform to obtain the target result data;

[0080] The visualization chart display module 340 is used to output a specified type of visualization chart for the target result data based on user needs for the user to perform data analysis.

[0081] The embodiment of the present invention uses a large language model as the foundation to build a data analysis platform that integrates multiple models, allowing users to directly communicate with data through text-based question-and-answer methods, and using visual analysis technology to dynamically display the result data of user question-and-answer content in the form of charts and generate data reports, which helps users intuitively understand the relationship and pattern between data. Compared with traditional platforms that have the risk of data leakage when analyzing large amounts of sensitive data, the embodiment of the present invention improves data security and the accuracy of data analysis to a certain extent through unified data processing and model services.

[0082] Optionally, on the basis of the above embodiments, it further includes a knowledge vector database construction unit, which is used to obtain multi-source heterogeneous data in different upstream systems before inputting the target feature vector into the pre-deployed knowledge vector database, and classify the multi-source heterogeneous data into structured and unstructured data after data cleaning;

[0083] Perform feature extraction on the classified structured data and unstructured data to obtain structured data vectors and unstructured data vectors;

[0084] The structured data vectors and unstructured data vectors are indexed and enhanced using an artificial intelligence framework to obtain multiple knowledge vectors, which are then stored in a knowledge vector database.

[0085] Optionally, on the basis of the above embodiments, it further includes a fusion large model platform building unit, which is used to fuse the natural language processing model, the large-scale language model and the multimodal model using a chain call strategy through a language model application construction framework before inputting the target feature vector and the target knowledge vector into the pre-built fusion large model platform to obtain an initialized fusion large model;

[0086] All knowledge vectors are input into the initialization fusion big model for training, and the trained initialization fusion big model is registered into the intelligent routing distributor used to call the model service to build a fusion big model platform.

[0087] Optionally, based on the above embodiments, the result data generation module 330 is further used to call the large model Agent intelligent body through the intelligent routing distributor in the pre-built fusion large model platform according to the target feature vector and the target knowledge vector;

[0088] Using the Agent intelligent body to identify keywords of the target knowledge vector, and split the target knowledge vector into multiple sub-vectors according to the keywords;

[0089] Inputting the multiple sub-vectors into the fusion large model platform to obtain multiple sub-task results corresponding to each sub-vector;

[0090] Send multiple subtask results to the intelligent routing distributor for result aggregation to obtain the target result data.

[0091] Optionally, based on the above embodiments, the visualization chart display module 340 further includes:

[0092] A chart type identification unit is used to convert the format of the target output result, obtain the target message data in Json format, and identify the chart type characteristic parameters in the target message data;

[0093] A chart type determination unit, configured to determine a target chart type for the target message data based on a chart type characteristic parameter using a visual chart tool;

[0094] The visualization chart rendering unit is used to assemble the target message data, obtain a visualization chart based on the determined target chart type, and render the visualization chart to a user interaction page.

[0095] Optionally, based on the above embodiments, the integrated large model platform building unit is further used to: obtain application programming interfaces of the natural language processing model, the large-scale language model and the multimodal model respectively;

[0096] Using different interfaces, natural language processing models, large-scale language models, and multimodal models are chained together as components in the language model application construction framework to obtain multiple sequential chains.

[0097] All sequential chains are fused to obtain an initialized fused large model that provides different input and output interfaces.

[0098] Optionally, based on the above embodiments, the visualization chart rendering unit is further used to: if the time type keyword contained in the query statement input by the user on the interactive page is empty, assembling the target message data, and obtaining a static visualization chart based on the determined target chart type;

[0099] If the time type keyword included in the query statement input by the user on the interactive page is not empty, the target message data is assembled and a dynamic visualization chart is obtained based on the determined target chart type;

[0100] Among them, the dynamic visualization chart is dynamically updated according to the specified period.

[0101] A data visualization device based on a fused large model provided in an embodiment of the present invention can execute a data visualization method based on a fused large model provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0102] Embodiment 4

[0103] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0104] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0105] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0106] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as a data visualization method based on a fusion large model.

[0107] That is: receiving the query statement entered by the user on the interactive page, performing vectorization processing using the embedding function, and obtaining the target feature vector;

[0108] Input the target feature vector into a pre-deployed knowledge vector database, and retrieve the target knowledge vector that is most similar to the target feature vector by calculating the cosine similarity;

[0109] Input the target feature vector and the target knowledge vector into the pre-built fusion model platform to obtain the target result data;

[0110] Output the specified type of visualization chart based on the target result data for users to perform data analysis.

[0111] In some embodiments, a data visualization method based on a fused large model may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the data visualization method based on a fused large model described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute a data visualization method based on a fused large model in any other appropriate manner (e.g., by means of firmware).

[0112] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0113] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0114] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0115] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0116] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0117] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0118] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0119] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A data visualization method based on a fusion large model, characterized in that: include: Receive the query statement entered by the user on the interactive page, perform vectorization processing using the embedding function, and obtain the target feature vector; Input the target feature vector into a pre-deployed knowledge vector database, and retrieve the target knowledge vector that is most similar to the target feature vector by calculating the cosine similarity; Input the target feature vector and the target knowledge vector into the pre-built fusion model platform to obtain the target result data; Based on user needs, the target result data is output with a specified type of visualization chart for users to perform data analysis.

2. The method according to claim 1, characterized in that Before inputting the target feature vector into a pre-deployed knowledge vector database, the method further includes: Acquire multi-source heterogeneous data from different upstream systems, and classify the multi-source heterogeneous data into structured and unstructured data after data cleaning; Perform feature extraction on the classified structured data and unstructured data to obtain structured data vectors and unstructured data vectors; The structured data vectors and unstructured data vectors are indexed and enhanced using an artificial intelligence framework to obtain multiple knowledge vectors, which are then stored in a knowledge vector database.

3. The method according to claim 1, characterized in that Before the target feature vector and the knowledge vector are input into the pre-built fusion model platform, the following steps are also included: Through the language model application construction framework, the natural language processing model, large-scale language model and multimodal model are integrated using a chain call strategy to obtain an initialized fusion large model; All knowledge vectors are input into the initialization fusion big model for training, and the trained initialization fusion big model is registered into the intelligent routing distributor used to call the model service to build a fusion big model platform.

4. The method according to claim 1, characterized in that: The target feature vector and target knowledge vector are input into the pre-built fusion model platform to obtain the target result data, including: According to the target feature vector and target knowledge vector, the large model Agent is called through the intelligent routing distributor in the pre-built fusion large model platform; Using the Agent intelligent body to identify keywords of the target knowledge vector, and split the target knowledge vector into multiple sub-vectors according to the keywords; Inputting the multiple sub-vectors into the fusion large model platform to obtain multiple sub-task results corresponding to each sub-vector; Send multiple subtask results to the intelligent routing distributor for result aggregation to obtain the target result data.

5. The method according to claim 1, characterized in that Output the specified type of dynamic visualization charts based on the target result data for users to analyze data, including: Convert the target output result to obtain the target message data in Json format, and identify the chart type characteristic parameters in the target message data; Determine a target chart type for the target message data based on a chart type characteristic parameter using a visual chart tool; After the target message data is assembled, a visual chart is obtained based on the determined target chart type, and the visual chart is rendered to a user interaction page.

6. The method according to claim 3, characterized in that The language model application construction framework integrates the natural language processing model, large-scale language model and multimodal model using a chain call strategy to obtain an initialized fusion model, including: Get APIs for natural language processing models, large-scale language models, and multimodal models respectively; Using different interfaces, natural language processing models, large-scale language models, and multimodal models are chained together as components in the language model application construction framework to obtain multiple sequential chains. All sequential chains are fused to obtain an initialized fused large model that provides different input and output interfaces.

7. According to claim 5, it is characterized in that: After assembling the target message data, a visualization chart is obtained based on the determined target chart type, including: If the time type keyword included in the query statement entered by the user on the interactive page is empty, the target message data is assembled and a static visualization chart is obtained based on the determined target chart type; If the time type keyword included in the query statement input by the user on the interactive page is not empty, the target message data is assembled and a dynamic visualization chart is obtained based on the determined target chart type; Among them, the dynamic visualization chart is dynamically updated according to the specified period.

8. A data visualization device based on a fusion large model, characterized in that: include: The feature vector extraction module is used to receive the query statement input by the user on the interactive page, and perform vectorization processing using the embedded function to obtain the target feature vector; A knowledge vector retrieval module is used to input the target feature vector into a pre-deployed knowledge vector database, and retrieve the target knowledge vector that is most similar to the target feature vector by calculating the cosine similarity; A result data generation module is used to input the target feature vector and the target knowledge vector into the pre-built fusion model platform to obtain the target result data; The visualization chart display module is used to output a specified type of visualization chart for the target result data based on user needs for users to perform data analysis.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a data visualization method based on a fused large model as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement a data visualization method based on a fused large model as described in any one of claims 1 to 7 when executed.

Images