Method and device for supporting questions and answers through computing power of intelligent computing center and related equipment

Abstract

The invention provides a method and device for supporting questions and answers through computing power of an intelligent computing center and related equipment, and particularly relates to the technical field of computing power infrastructure. The method comprises the steps that S1, retrieval is conducted in a retrieval database according to a question text by means of computing power distributed by the intelligent computing center, and the question text is obtained; the plurality of recalled text blocks and text block connection information are obtained, and the text block connection information represents logic association among the plurality of recalled text blocks; s2, inputting the recalled text blocks and the text block connection information into a question and answer model to obtain an answer text; wherein the retrieval database comprises a plurality of pieces of vector data formed based on a plurality of preset text blocks and at least one knowledge graph formed based on the plurality of preset text blocks, each knowledge graph comprises a plurality of graph nodes, and the vector data formed based on the same preset text blocks and the graph nodes are mutually linked. The answer text output by the question and answer model can be more accurate.

Description

Technical Field

[0001] The present application relates to the technical field of computing power infrastructure, and specifically to methods, devices and related equipment for supporting question and answer through the computing power of an intelligent computing center. Background Art

[0002] With the development of artificial intelligence technology and computing power technology, the concept of intelligent computing center has emerged. "Intelligent computing center" refers to the use of large-scale heterogeneous computing power resources, including general computing power and intelligent computing power, mainly for artificial intelligence applications (such as artificial intelligence deep learning model development, model training and model reasoning and other scenarios) to provide the required computing power, data and algorithms. Intelligent computing center covers facilities, hardware, software, and can provide full-stack capabilities from bottom-level computing power to top-level application enablement.

[0003] Efficient question-and-answer interaction can be achieved based on the intelligent computing center. In related technologies, after receiving the question text raised by the user, the question-and-answer model will use the computing power of the intelligent computing center to perform vector retrieval on the question text, and first return multiple retrieved text blocks as context, so as to form the corresponding answer text by summarizing the context.

[0004] It is found in the application that the multiple recalled text blocks obtained based on the vector retrieval method usually cannot cover all the content required for the answer, and the multiple recalled text blocks are independent of each other, which will increase the complexity of the question-answering model's processing of text blocks, and ultimately make the answer text output by the question-answering model inaccurate. Summary of the invention

[0005] The purpose of this application is to provide a method, device and related equipment for supporting question answering through the computing power of an intelligent computing center, so as to solve the technical problem of low accuracy of answer text output based on vector retrieval in related technologies.

[0006] In a first aspect, an embodiment of the present application provides a method for supporting question-answering by computing power of an intelligent computing center, the method comprising:

[0007] Step S1: using the computing power allocated by the intelligent computing center, searching in a pre-set search database according to the question text, and obtaining a plurality of recalled text blocks and text block connection information, wherein the text block connection information is used to represent the logical association between the plurality of recalled text blocks;

[0008] Step S2: inputting the multiple recalled text blocks and the text block connection information into the question-answering model to generate text, thereby obtaining an answer text;

[0009] Among them, the retrieval database includes multiple vector data formed based on multiple preset text blocks and at least one knowledge graph formed based on the multiple preset text blocks, each of the knowledge graphs includes multiple graph nodes, and in the retrieval database, the vector data and the graph nodes formed based on the same preset text block are linked to each other.

[0010] In a second aspect, an embodiment of the present application further provides a device for supporting question and answer through computing power of an intelligent computing center, the device comprising:

[0011] A retrieval module, used to use the computing power allocated by the intelligent computing center to search in a pre-set retrieval database according to the question text, and obtain multiple recalled text blocks and text block connection information, wherein the text block connection information is used to represent the logical association between the multiple recalled text blocks;

[0012] A text generation module, used for inputting the multiple recalled text blocks and the text block connection information into the question-answering model to generate text and obtain an answer text;

[0013] Among them, the retrieval database includes multiple vector data formed based on multiple preset text blocks and at least one knowledge graph formed based on the multiple preset text blocks, each of the knowledge graphs includes multiple graph nodes, and in the retrieval database, the vector data and the graph nodes formed based on the same preset text block are linked to each other.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method described in the first aspect.

[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0016] In a fifth aspect, an embodiment of the present application provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the steps of the method described in the first aspect.

[0017] In the present application, the computing power allocated by the intelligent computing center is utilized to provide retrieval support based on a retrieval database composed of a mixture of multiple vector data and at least one knowledge graph, so that while multiple recalled text blocks are retrieved through the question text, the existence of the aforementioned knowledge graph is utilized to correspondingly obtain the logical associations between the multiple recalled text blocks, and all of them are input into the question-answering model, so that the question-answering model can output a more accurate answer text. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of a method for supporting question and answer through the computing power of an intelligent computing center provided in an embodiment of the present application;

[0019] Figure 2 It is a schematic diagram of a database construction process provided in an embodiment of the present application;

[0020] Figure 3 It is a structural diagram of a device for supporting question and answer through the computing power of an intelligent computing center provided in an embodiment of the present application;

[0021] Figure 4 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] The "intelligent computing center" mentioned in this application refers to a facility that provides the required computing power, data and algorithms for artificial intelligence applications (such as artificial intelligence deep learning model development, model training and model reasoning) by using large-scale heterogeneous computing resources, including general computing power (CPU) and intelligent computing power (GPU, FPGA, ASIC, etc.). The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from bottom-level computing power to top-level application enablement.

[0024] The "computing resources" mentioned in this application refer to technologies and facilities with information computing, transmission, storage and application capabilities required for the development of a digital society, including but not limited to computing resources such as CPUs and GPUs, network resources such as switches and routers, storage resources such as storage arrays and distributed storage, security resources such as firewalls and intrusion detection systems, and supporting and guarantee resources such as wind, fire, water and electricity.

[0025] The "computing power" mentioned in this application is the ability of computer equipment or computing / data center to process information. It is the ability of computer hardware and software to work together to perform certain computing needs. It is the computing power to achieve target result output by processing information data. It is a new type of productivity that integrates information computing power, network carrying capacity, and data storage capacity. It mainly provides services to society through computing power infrastructure.

[0026] The "computing power (CP)" described in this application is the ability of a data center server to process data and output results. It is a comprehensive indicator to measure the computing power of a data center, including general computing power, supercomputing power and intelligent computing power. The commonly used unit of measurement is the number of floating point operations per second (Floating Point Operations Per Second, FLOPS, where 1EFLOPS=10^18FLOPS). The larger the value, the stronger the comprehensive computing power. According to calculations, 1EFLOPS is approximately the computing power output of 5 Tianhe-2A or 500,000 mainstream server CPUs (Central Processing Unit) or 2 million mainstream notebooks. The calculation formula is: CP=CP 通用 +CP 智能 +CP 超级 In the embodiment of the present application, the computing power adopts the half-precision floating point computing power (FP16) of the graphics card.

[0027] The "carrying capacity" (Network Power, NP) described in this application is a manifestation of the data transmission capability of computing facilities, including comprehensive capabilities such as network architecture, network bandwidth, transmission latency, intelligent management and scheduling. Carrying capacity involves network transmission within and between data centers, and is a comprehensive indicator for measuring network transmission scheduling capabilities. In the embodiment of this application, carrying capacity uses video memory bandwidth. In the embodiment of this application, storage capacity uses video memory bandwidth.

[0028] The "Storage Power" (SP) described in this application refers to the comprehensive capabilities of a data center in terms of data storage capacity, performance, safety and reliability, and green and low-carbon. It is a comprehensive indicator for measuring the data storage capacity of a data center, including external storage devices such as storage arrays and server built-in storage devices. The commonly used unit of measurement for storage capacity is exabyte (EB, 1EB = 2^60bytes), and the commonly used unit of measurement for performance is the number of reads and writes per second per unit capacity (IOPS / TB, Input / Output Operations Per Second / TB). The disaster recovery ratio is an important manifestation of safety and reliability. In the embodiment of this application, storage power uses video memory capacity.

[0029] The "computing power infrastructure" described in this application is a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage capacity. It can realize centralized computing, storage, transmission and application of information, and presents characteristics such as diversity and ubiquity, intelligence and agility, security and reliability, green and low-carbon. It is of great significance to promote industrial transformation and upgrading, enable my country's scientific and technological innovation, meet people's better life and achieve high-efficiency social governance.

[0030] The “computing power” mentioned in this application includes general computing power CP 通用 、Intelligent computing power CP 智能 And super computing power CP 超级 .

[0031] The "general computing power" mentioned in this application refers to the computing power provided by servers based on CPU chips, which is used to support basic general computing such as cloud computing and edge computing.

[0032] The "intelligent computing power" mentioned in this application refers to the large-scale deployment of intelligent computing centers based on special chips such as GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit) for various innovative artificial intelligence applications, such as natural language processing, machine vision, etc.

[0033] The "supercomputing power" mentioned in this application is mainly the computing power provided by high-performance computing clusters such as supercomputers. It utilizes the centralized computing resources of multiple computer systems working in parallel and uses a dedicated operating system to handle extremely complex or data-intensive problems. It is mainly used for calculations in cutting-edge scientific fields, such as planetary simulation, drug molecule design, and gene analysis.

[0034] The vector data described in this application is a data set consisting of numerical vectors, where each vector represents a point or feature in a multidimensional space.

[0035] The knowledge graph described in this application is a graph used to display the knowledge development process and structural relationship. It describes knowledge resources and their carriers through visualization technology, and mines, analyzes, constructs, draws and displays knowledge and their mutual connections. The knowledge graph is also called a semantic network, which is used to represent the network of real-world entities and illustrate the relationships between entities.

[0036] With the development of artificial intelligence technology and computing power technology, the application scenarios of intelligent computing centers are also constantly expanding. When the intelligent computing center provides computing power support for the question-answering model, it is found that the multiple recalled text blocks obtained by the question-answering model in the retrieval and recall stage usually cannot cover all the content required for the answer, and the multiple recalled text blocks are independent of each other. This increases the complexity of the question-answering model's processing of text blocks, resulting in the accuracy of the answer text output by the question-answering model being lower than expected.

[0037] Based on this, the present application proposes a method, apparatus and related equipment for supporting question and answering through the computing power of an intelligent computing center, using the computing power allocated by the intelligent computing center to provide retrieval support based on a retrieval database composed of a mixture of multiple vector data and at least one knowledge graph, so that while multiple recalled text blocks are retrieved through the question text, the existence of the aforementioned knowledge graph is used to correspondingly obtain the logical associations between the multiple recalled text blocks, and all of them are input into the question and answer model, so that the question and answer model can output a more accurate answer text.

[0038] See also Figure 1 , Figure 1 It is a flowchart of a method for supporting question and answer through the computing power of an intelligent computing center provided in an embodiment of the present application.

[0039] like Figure 1 As shown, the method of supporting question answering by using the computing power of the intelligent computing center includes:

[0040] Step S1: Using the computing power allocated by the intelligent computing center, search in a pre-set search database according to the question text to obtain multiple recalled text blocks and text block connection information.

[0041] The text block connection information is used to represent the logical association between the multiple recalled text blocks.

[0042] The retrieval database includes multiple vector data formed based on multiple preset text blocks and at least one knowledge graph formed based on the multiple preset text blocks, each of the knowledge graphs includes multiple graph nodes, and in the retrieval database, the vector data and the graph nodes formed based on the same preset text block are linked to each other.

[0043] The above question text is used to represent the user's input text received by the question-answering model when the user performs question-answering interaction with the question-answering model.

[0044] The multiple preset text blocks are used to represent the business knowledge base of the target business scenario, and the target business scenario is the business scenario corresponding to the question text. For example, the target business scenario can be a weather forecast scenario, a map navigation scenario, a process consultation scenario, etc.

[0045] In the present application, a preset text block can form a vector data, and different preset text blocks form different vector data; a preset text block can form multiple knowledge triples, each of which includes two entity data and a relationship data for representing the logical relationship and / or semantic relationship between the two entity data. Several knowledge triples can form the aforementioned knowledge graph, and the entity data exists as a graph node in the knowledge graph.

[0046] For example, if the vector data formed based on a preset text block is set as vector data A, knowledge triple B and knowledge triple C can also be formed based on the preset text block, wherein knowledge triple B includes entity data a and entity data b, and knowledge triple C includes entity data a and entity data c, then vector data A will be linked to entity data a (that is, graph node a), entity data b (that is, graph node b) and entity data c (that is, graph node c).

[0047] Exemplarily, if any two different recalled text blocks among the multiple recalled text blocks are set as a first recalled text block and a second recalled text block, the text block connection information may indicate that the first recalled text block and the second recalled text block are in a parallel relationship / contextual relationship / inclusion relationship, etc.;

[0048] For example, when the first recall text block is used to describe the appearance characteristics of a product, and the second recall text block is used to describe the structural characteristics of the product, it can be considered that the first recall text block and the second recall text block are in a parallel relationship;

[0049] When the text content of the first recalled text block is an explanation of a certain noun concept / a certain paragraph in the text content of the second recalled text block, it can be considered that the first recalled text block and the second recalled text block are in a containment relationship;

[0050] When the first recall text block is used to explain the implementation measures of a certain technology, and the second recall text block is used to explain the application prospects, advantages and disadvantages of the technology, it can be considered that the first recall text block and the second recall text block are in a contextual relationship.

[0051] It should be understood that the logical association specifically represented by the text block connection information is not limited to the parallel relationship, contextual relationship, inclusion relationship, etc. mentioned in the above examples.

[0052] Step S2: input the multiple recalled text blocks and the text block connection information into the question-answering model for text generation to obtain the answer text.

[0053] The present application provides retrieval support based on a retrieval database composed of a mixture of multiple vector data and at least one knowledge graph. While retrieving multiple recalled text blocks through question text, the application can utilize the existence of the aforementioned knowledge graph to correspondingly obtain the logical associations between the multiple recalled text blocks, so as to achieve logical sorting of the multiple recalled text blocks. This can reduce the probability of the question-answering model misunderstanding the logical relationship between the multiple recalled text blocks, thereby enabling the question-answering model to output a more accurate answer text.

[0054] In one embodiment, the step S1: searching a preset search database according to the question text to obtain a plurality of recalled text blocks and the connection relationship between the plurality of recalled text blocks includes:

[0055] Step S11: searching the retrieval database according to the question text to obtain a plurality of recall vector data;

[0056] Step S12: determining, from the plurality of preset text blocks, a plurality of recalled text blocks corresponding to the plurality of recalled vector data respectively;

[0057] Step S13: Determine the text block connection information based on the multiple recalled text blocks and the at least one knowledge graph.

[0058] In this embodiment, multiple recalled text blocks are obtained by recalling vector data and combining the mapping relationship / association relationship between the recalled vector data and the preset text blocks. This can avoid the situation where redundant text blocks may be recalled when the knowledge graph retrieves the recalled text blocks, thereby ensuring the accuracy of the multiple recalled text blocks obtained.

[0059] It should be understood that the operations of the above steps S11 to S13 are all completed based on the computing power allocated by the intelligent computing center.

[0060] The recall text block corresponding to the recall vector data may be understood as a preset text block forming the recall vector data.

[0061] In one embodiment, the step S13: determining the text block connection information according to the multiple recalled text blocks and the at least one knowledge graph includes:

[0062] Step S131: in the at least one knowledge graph, obtaining node relationship information of a plurality of recall graph nodes corresponding to the plurality of recall text blocks, wherein the node relationship information is used to represent the semantic relationship between the plurality of recall graph nodes;

[0063] Step S132: Determine text block connection information according to the node relationship information and the corresponding relationship between the multiple recalled text blocks and the multiple recalled graph nodes.

[0064] In this application, a recalled text block corresponds to a set of graph nodes.

[0065] For any two different recall text blocks among the multiple recall text blocks, the logical association between the any two different recall text blocks can be determined by analyzing the semantic relationship and / or logical relationship between the two sets of graph node sets corresponding to the any two different recall text blocks in the at least one knowledge graph, and the text block connection information can be formed by traversing the multiple recall text blocks in pairs.

[0066] Among them, a set of graph nodes corresponding to a recalled text block can be understood as: a plurality of graph nodes corresponding to a plurality of knowledge triples formed based on the recalled text block in the knowledge graph.

[0067] In one example, the largest number of semantic relationships between the two groups of graph node sets can be determined as the logical association of the two different recalled text blocks. For example, if there are 5 parallel relationships and 1 contextual relationship between the two groups of graph node sets, then the two recalled text blocks corresponding to the two groups of graph node sets can be considered to be in a parallel relationship.

[0068] In another example, the two groups of graph node sets may also be analyzed through a trained network model to determine the logical association between any two different recalled text blocks.

[0069] In one embodiment, the step S11: searching the retrieval database according to the question text to obtain a plurality of recall vector data includes:

[0070] Step S111: vectorizing the question text to obtain question vector data;

[0071] Step S112: using the question vector data as a search condition, searching in the plurality of vector data to obtain a plurality of search recall vector data;

[0072] Step S113: when the graph node linked to the search and recall vector data is also linked to other vector data, and the multiple search and recall vector data do not include the other vector data, determining the other vector data as supplementary recall vector data;

[0073] Step S114: The multiple recall vector data include the retrieval recall vector data and the supplementary recall vector data.

[0074] In this embodiment, a plurality of retrieval recall vector data are first retrieved from the plurality of vector data by means of vector retrieval, and then based on the links between the plurality of vector data and a plurality of graph nodes, a reverse search for graph nodes is performed using the plurality of retrieval recall vector data. With the aid of the links of the graph nodes, other vector data that are not recalled by the vector retrieval process but are linked to the reverse searched graph nodes are supplementarily recalled as supplementary recall vector data, so as to ensure the comprehensiveness of the recalled vector data, thereby further improving the accuracy of the answer text output by the question-answering model.

[0075] For example, if the multiple retrieval recall vector data are set to be vector data 1 and vector data 3, and vector data 1 is linked to graph node 1 in the knowledge graph, and vector data 3 is linked to graph node 3 in the knowledge graph, and if graph node 1 is also linked to vector data 2, vector data 2 will be used as supplementary recall vector data for supplementary recall.

[0076] Exemplarily, the data similarity between each vector data and the problem vector data can be calculated among the multiple vector data, and the multiple vector data can be sorted in descending order based on the data similarity, and then the first N vector data in the sorted sequence can be recalled as the multiple retrieval recall vector data, where N is a positive integer.

[0077] In one embodiment, the step S11: searching the retrieval database according to the question text to obtain a plurality of recall vector data includes:

[0078] Step S115: According to the question keywords extracted from the question text, determine the question graph nodes matching the question keywords in the at least one knowledge graph;

[0079] Step S116: Determine a plurality of graph recall vector data according to the problem graph node and adjacent graph nodes of the problem graph node, wherein the plurality of recall vector data include the graph recall vector data.

[0080] In this embodiment, the question text is matched with the knowledge graph to determine several graph nodes related to the question text, and then based on the determined graph nodes and the links between the graph nodes and the vector data, a reverse search is performed for several vector data related to the question text to support the retrieval and recall of vector data based on the knowledge graph, enrich the retrieval and recall methods of vector data, and make the recalled vector data more comprehensive and accurate.

[0081] Exemplarily, the semantic similarity between each graph node and the text keyword in a plurality of graph nodes included in the knowledge graph can be calculated, and the plurality of graph nodes can be sorted in descending order based on the semantic similarity, and then the first M graph nodes in the sorted sequence can be determined as question graph nodes that match the question keyword, where M is a positive integer.

[0082] The adjacent graph nodes of the problem graph node should be understood as graph nodes whose connection depth with the problem graph node in the knowledge graph is less than or equal to a preset depth threshold.

[0083] For example, if the preset depth threshold is set to 2, the problem graph node k, graph node 4, graph node 5 and graph node 6 are connected in sequence, then graph node 4 and graph node 5 can be considered as adjacent graph nodes of the problem graph node k.

[0084] The above-mentioned preset depth threshold can be adaptively selected according to actual needs, and the present application does not limit the specific value of the preset depth threshold.

[0085] It should be pointed out that, in the process of searching the retrieval database according to the question text to obtain multiple recall vector data, the aforementioned vector retrieval method and the vector recall method based on the knowledge graph can be mixed and applied to improve the integrity of the recalled vector data. At this time, if the retrieval recall vector data and the supplementary recall vector data are set to form a first vector data set, and the multiple graph recall vector data form a second vector data set, then the vector data set formed by the multiple recall vector data can be understood as the union of the first vector data set and the second vector data set.

[0086] In one embodiment, before the step S1: searching a preset search database according to the question text to obtain a plurality of recalled text blocks and text block connection information, the method further comprises:

[0087] Step S01: performing vectorization processing on the plurality of preset text blocks respectively to obtain the plurality of vector data;

[0088] Step S02: extracting knowledge triples from the plurality of preset text blocks respectively to obtain the at least one knowledge graph;

[0089] Step S03: Among the multiple vector data and the at least one knowledge graph, the vector data and graph nodes corresponding to the same preset text block are linked.

[0090] In this embodiment, vectorization processing and knowledge triple extraction are performed on the same multiple preset text blocks respectively to obtain corresponding multiple vector data and at least one knowledge graph respectively, and then the vector data and graph nodes corresponding to the same preset text block are linked to achieve the reuse of the existing vector database construction process and knowledge graph database construction process, thereby reducing the difficulty of building a hybrid retrieval database.

[0091] In one example, if Figure 2 As shown, the preset documents (derived from the business knowledge base) can be segmented to obtain the aforementioned multiple preset text blocks, and then vectorization and knowledge triples can be extracted to respectively establish a vector database and a knowledge graph database, and finally the vectors in the vector database (i.e., vector data) and the nodes in the knowledge graph database (i.e., graph nodes) are linked with reference to the aforementioned link instructions to form a hybrid retrieval database.

[0092] The present application also provides a device that supports question-answering through the computing power of an intelligent computing center, such as Figure 3 As shown, the device 300 for supporting question and answer through intelligent computing center computing power includes:

[0093] A retrieval module 301 is used to use the computing power allocated by the intelligent computing center to search in a pre-set retrieval database according to the question text to obtain multiple recalled text blocks and text block connection information, wherein the text block connection information is used to represent the logical association between the multiple recalled text blocks;

[0094] A text generation module 302, for inputting the plurality of recalled text blocks and the text block connection information into the question-answering model to generate text and obtain an answer text;

[0095] Among them, the retrieval database includes multiple vector data formed based on multiple preset text blocks and at least one knowledge graph formed based on the multiple preset text blocks, each of the knowledge graphs includes multiple graph nodes, and in the retrieval database, the vector data and the graph nodes formed based on the same preset text block are linked to each other.

[0096] In one embodiment, the retrieval module 301 includes:

[0097] A vector recall unit, used for searching the retrieval database according to the question text to obtain a plurality of recall vector data;

[0098] A text block recall unit, configured to determine, from the plurality of preset text blocks, a plurality of recalled text blocks respectively corresponding to the plurality of recalled vector data;

[0099] An information determination unit is used to determine the text block connection information based on the multiple recalled text blocks and the at least one knowledge graph.

[0100] In one embodiment, the vector recall unit includes:

[0101] A vectorization processing subunit is used to perform vectorization processing on the question text to obtain question vector data;

[0102] A vector retrieval subunit, configured to use the question vector data as a retrieval condition to perform retrieval in the plurality of vector data to obtain a plurality of retrieval recall vector data;

[0103] A supplementary recall subunit, configured to determine the other vector data as supplementary recall vector data when the graph node linked to the retrieved recall vector data is also linked to other vector data and the multiple retrieved recall vector data do not include the other vector data;

[0104] Among them, the multiple recall vector data include the retrieval recall vector data and the supplementary recall vector data.

[0105] In one embodiment, the vector recall unit includes:

[0106] A node matching subunit, configured to determine, based on the question keywords extracted from the question text, a question graph node matching the question keywords in the at least one knowledge graph;

[0107] A reverse recall subunit is used to determine a plurality of graph recall vector data according to the problem graph node and adjacent graph nodes of the problem graph node, wherein the plurality of recall vector data include the graph recall vector data.

[0108] In one embodiment, the apparatus 300 further includes a database construction module, and the database construction module is specifically used to:

[0109] Performing vectorization processing on the plurality of preset text blocks respectively to obtain the plurality of vector data;

[0110] Extracting knowledge triples from the plurality of preset text blocks respectively to obtain the at least one knowledge graph;

[0111] In the multiple vector data and the at least one knowledge graph, the vector data and graph nodes corresponding to the same preset text block are linked.

[0112] In one embodiment, the information determination unit is specifically used to:

[0113] In the at least one knowledge graph, node relationship information of a plurality of recall graph nodes corresponding to the plurality of recall text blocks is obtained, wherein the node relationship information is used to represent a semantic relationship between the plurality of recall graph nodes;

[0114] The text block connection information is determined according to the node relationship information and the corresponding relationship between the multiple recalled text blocks and the multiple recalled graph nodes.

[0115] The device 300 for supporting question and answer through the computing power of an intelligent computing center provided in the embodiment of the present application can implement each process in the above-mentioned question and answer method embodiment. To avoid repetition, it will not be described here.

[0116] According to an embodiment of the present application, the present application also provides an electronic device and a readable storage medium.

[0117] Figure 4 A schematic block diagram of an example electronic device 400 that can be used to implement an embodiment of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, 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, 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 application described and / or required herein.

[0118] like Figure 4 As shown, the device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 to a random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0119] A number of components in the device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0120] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSP), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 401 performs the various methods and processes described above, such as a method for supporting question-answering through intelligent computing center computing power. For example, in some embodiments, the method for supporting question-answering through intelligent computing center computing power may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of the method for supporting question-answering through intelligent computing center computing power described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured in any other appropriate manner (eg, by means of firmware) to execute the method of supporting question-answering through the computing power of an intelligent computing center.

[0121] Various embodiments of the systems and techniques described above herein may 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), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may 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 may be a dedicated 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.

[0122] The program code for implementing the method of the present application can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, implements the functions / operations specified in the flow chart and / or block diagram. The program code can 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.

[0123] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. 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.

[0124] As used herein, the term "machine-readable medium" refers to any computer program product, apparatus, and / or device (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0125] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer 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 pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0126] The systems and techniques described herein may be implemented in a computing system that includes back-end 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 front-end 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 back-end components, middleware components, or front-end 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), and the Internet.

[0127] A computer 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 relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0128] The present application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the above Figure 1 The various processes of the method embodiment shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0129] 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 recorded in this application can be executed in parallel, sequentially or in different orders, as long as the expected results of the technical solution disclosed in this application can be achieved, and this document is not limited here.

[0130] The above specific implementations do not constitute a limitation on the protection scope of this application. 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 modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for supporting question answering through the computing power of an intelligent computing center, characterized in that: The method comprises: Step S1: using the computing power allocated by the intelligent computing center, searching in a pre-set search database according to the question text, and obtaining a plurality of recalled text blocks and text block connection information, wherein the text block connection information is used to represent the logical association between the plurality of recalled text blocks; Step S2: inputting the multiple recalled text blocks and the text block connection information into the question-answering model to generate text, thereby obtaining an answer text; Among them, the retrieval database includes multiple vector data formed based on multiple preset text blocks and at least one knowledge graph formed based on the multiple preset text blocks, each of the knowledge graphs includes multiple graph nodes, and in the retrieval database, the vector data and the graph nodes formed based on the same preset text block are linked to each other.

2. The method according to claim 1, characterized in that The step S1: searching a preset search database according to the question text to obtain a plurality of recalled text blocks and the connection relationship between the plurality of recalled text blocks, including: Step S11: searching the retrieval database according to the question text to obtain a plurality of recall vector data; Step S12: determining, from the plurality of preset text blocks, a plurality of recalled text blocks corresponding to the plurality of recalled vector data respectively; Step S13: Determine the text block connection information based on the multiple recalled text blocks and the at least one knowledge graph.

3. The method according to claim 2, characterized in that The step S11: searching the retrieval database according to the question text to obtain a plurality of recall vector data, including: Step S111: vectorizing the question text to obtain question vector data; Step S112: using the question vector data as a search condition, searching in the plurality of vector data to obtain a plurality of search recall vector data; Step S113: when the graph node linked to the search and recall vector data is also linked to other vector data, and the multiple search and recall vector data do not include the other vector data, determining the other vector data as supplementary recall vector data; Step S114: The multiple recall vector data include the retrieval recall vector data and the supplementary recall vector data.

4. The method according to claim 2, characterized in that: The step S11: searching the retrieval database according to the question text to obtain a plurality of recall vector data, including: Step S115: According to the question keywords extracted from the question text, determine the question graph nodes matching the question keywords in the at least one knowledge graph; Step S116: Determine a plurality of graph recall vector data according to the problem graph node and adjacent graph nodes of the problem graph node, wherein the plurality of recall vector data include the graph recall vector data.

5. The method according to any one of claims 1 to 4, characterized in that Before the step S1: searching a preset search database according to the question text to obtain a plurality of recalled text blocks and text block connection information, the method further comprises: Step S01: performing vectorization processing on the plurality of preset text blocks respectively to obtain the plurality of vector data; Step S02: extracting knowledge triples from the plurality of preset text blocks respectively to obtain the at least one knowledge graph; Step S03: Among the multiple vector data and the at least one knowledge graph, the vector data and graph nodes corresponding to the same preset text block are linked.

6. The method according to any one of claims 2 to 4, characterized in that: The step S13: determining the text block connection information according to the multiple recalled text blocks and the at least one knowledge graph, comprises: Step S131: in the at least one knowledge graph, obtaining node relationship information of a plurality of recall graph nodes corresponding to the plurality of recall text blocks, wherein the node relationship information is used to represent the semantic relationship between the plurality of recall graph nodes; Step S132: Determine text block connection information according to the node relationship information and the corresponding relationship between the multiple recalled text blocks and the multiple recalled graph nodes.

7. A device for supporting question and answer through the computing power of an intelligent computing center, characterized in that: The device comprises: A retrieval module, used to use the computing power allocated by the intelligent computing center to search in a pre-set retrieval database according to the question text, and obtain multiple recalled text blocks and text block connection information, wherein the text block connection information is used to represent the logical association between the multiple recalled text blocks; A text generation module, used for inputting the multiple recalled text blocks and the text block connection information into the question-answering model to generate text and obtain an answer text; Among them, the retrieval database includes multiple vector data formed based on multiple preset text blocks and at least one knowledge graph formed based on the multiple preset text blocks, each of the knowledge graphs includes multiple graph nodes, and in the retrieval database, the vector data and the graph nodes formed based on the same preset text block are linked to each other.

8. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of the method according to any one of claims 1 to 6 when executed by the processor.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 6.

Images