Object information acquisition method, device, storage medium, and program product

Abstract

The application discloses an object information acquisition method and device, a storage medium and a program product, and relates to the technical field of data processing. In the method, an electronic device can generate initial information of a target object in a dimension indicated by each dimension identifier according to a target request, and obtain at least one initial information; the initial information can be subjected to quality detection; if the quality detection result corresponding to the initial information is detection failure, the electronic device can update the initial information to obtain target information, and the quality detection result corresponding to the target information is detection success; the target information and / or the initial information with the detection success quality detection result can be subjected to merging processing to obtain object information of the target object. Through the method, the accuracy of the object information can be improved, and the accuracy of information acquisition can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to an object information acquisition method and device, a storage medium and a program product. BACKGROUND

[0002] With the deepening of digital transformation, information acquisition is becoming more and more difficult.

[0003] At present, information can be acquired based on a large language model (LLM). For example, an information acquisition requirement can be input to the LLM, and the LLM can acquire and output corresponding information from a preset database. However, in the above method, there may be inaccurate information acquisition, resulting in poor accuracy of information acquisition. SUMMARY

[0004] The present application provides an object information acquisition method and device, a storage medium and a program product to at least solve the problem of poor accuracy of information acquisition in related technologies.

[0005] In a first aspect, the present application provides an object information acquisition method, comprising:

[0006] Acquiring a target request, wherein the target request comprises a target object identifier of a target object and at least one dimension identifier;

[0007] Acquiring initial information of the target object in each dimension indicated by the dimension identifier from a database, to obtain at least one initial information;

[0008] Performing quality detection on each initial information to obtain a quality detection result, wherein the quality detection result is detection pass or detection fail;

[0009] For any one initial information, if the quality detection result corresponding to the initial information is detection fail, updating the initial information to obtain target information, wherein the quality detection result corresponding to the target information is detection pass;

[0010] Updating the database according to the target information and / or the initial information with the quality detection result of detection pass;

[0011] Merging the target information and / or the initial information with the quality detection result of detection pass to obtain object information of the target object.

[0012] In a second aspect, the present application provides an object information acquisition device, comprising an acquisition module, a detection module, an updating module and a merging module, wherein:

[0013] The acquisition module is configured to acquire a target request, the target request comprising a target object identifier of a target object and at least one dimension identifier;

[0014] The acquisition module is further configured to acquire initial information of the target object in each dimension indicated by the dimension identifier from a database, to obtain at least one initial information;

[0015] The detection module is configured to perform quality detection on each initial information to obtain a quality detection result, the quality detection result being detection pass or detection fail;

[0016] For any one initial information, if the quality detection result corresponding to the initial information is detection fail, the updating module is configured to update the initial information to obtain target information, the quality detection result corresponding to the target information being detection pass;

[0017] The updating module is further configured to update the database according to the target information and / or the initial information with the quality detection result being detection pass;

[0018] The merging module is configured to perform merging processing on the target information and / or the initial information with the quality detection result being detection pass to obtain object information of the target object.

[0019] In a third aspect, the present application provides an electronic device, comprising: a memory configured to store a computer program; and a processor configured to execute the computer program to implement the steps of any of the above object information acquisition methods.

[0020] In a fourth aspect, the present application further provides a computer readable storage medium, the computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the steps of any of the above object information acquisition methods.

[0021] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, wherein the computer program is executed by a processor to implement the steps of any of the above object information acquisition methods.

[0022] Through the technical solutions provided in the embodiments of the present application, the electronic device can obtain a target request, the target request including a target object identifier of a target object and at least one dimension identifier; can generate initial information of the target object in the dimensions indicated by the dimension identifiers to obtain at least one initial information; can perform quality detection on each initial information to obtain a quality detection result, the quality detection result being detection pass or detection fail; for any one initial information, if the quality detection result corresponding to the initial information is detection fail, the electronic device can update the initial information to obtain target information, the quality detection result corresponding to the target information being detection pass; and can perform merging processing on the target information and / or the initial information with the quality detection result being detection pass to obtain object information of the target object. In the above method, the initial information of the target object can be subjected to quality detection, and the object information of the target object can be generated after the quality detection result of each initial information is detection pass. Through the above method, the accuracy of the object information can be improved, and thus the accuracy of information acquisition can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present application, the drawings required to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some of the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0024] Figure 1 Flowchart of the object information acquisition method provided by the embodiments of the present application Figure One

[0025] Figure 2 Flowchart of the object information acquisition method provided by the embodiments of the present application Figure Two

[0026] Figure 3 Structure diagram of the object information acquisition device provided by the embodiments of the present application

[0027] Figure 4 Structure diagram of the electronic device provided by the present application DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0029] ​​It should be noted that in the description of the present application, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles or devices including a series of elements not only include those elements, but also include other elements not explicitly listed or inherent to such processes, methods, articles or devices. The terms "first", "second" and the like in the present application are used to distinguish similar objects, not to describe a specific order or sequence.

[0030] With the deepening of digital transformation, the traditional information acquisition method has problems such as low information acquisition efficiency. For example, the traditional information acquisition method includes manually performing network search, or reading reports and other methods.

[0031] At present, artificial intelligence technology represented by large language model (Large Language Model, LLM) has made breakthrough progress, and in related technologies, information can be acquired based on LLM, which improves the information acquisition efficiency to a certain extent. Specifically, the information acquisition demand can be input to the LLM, and the LLM can acquire and output corresponding information from the preset database. However, in the above method, there may be inaccurate information acquisition, resulting in poor accuracy of information acquisition.

[0032] Therefore, the embodiments of the present application provide an object information acquisition method, which can detect the quality of automatically acquired information, and reacquire if the quality detection fails, thereby improving the accuracy of information acquisition.

[0033] In order for those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0034] Figure 1 Flowchart of object information acquisition method provided by embodiments of the present application Figure One Please refer to Figure 1 The execution subject of the method can be an electronic device or a processor in the electronic device. For example, the electronic device can be a server or a memory, and the processor can be a central processing unit (Central Processing Unit, CPU). For ease of understanding, the execution subject will be taken as an electronic device in the following description. The method can include:

[0035] S101, acquire a target request, the target request including a target object identifier of a target object and at least one dimension identifier.

[0036] The target object identifier can be used to uniquely identify the target object. For example, the target object can be company A, and the target object identifier can be the company name of company A.

[0037] The dimension identifier can be used to indicate the dimension of the information acquisition. For example, the dimension identifier can be "technical strength", or "market strategy", etc.

[0038] The target request is used to request the object information of the target object in the dimension corresponding to the dimension identifier.

[0039] For example, assuming that the target object identifier included in the target request is "A company", and the dimension identifier is "technical strength", the target request can be used to request the technical strength information of A company.

[0040] In this embodiment, the electronic device can obtain the target request input by the user through the input operation, or the electronic device can receive the target request sent by another electronic device, and this embodiment does not limit the same.

[0041] S102, obtain the initial information of the target object in the dimension indicated by each dimension identifier from the database, to obtain at least one initial information.

[0042] It should be noted that the method of obtaining the initial information of the target object in the dimension indicated by each dimension identifier from the database is the same. In the embodiment of the application, the method of obtaining the initial information of the target object in the dimension indicated by the dimension identifier from the database is described by taking any one dimension identifier as an example.

[0043] The database can include a preset local database and an Internet database. That is, the initial information of the target object in the dimension indicated by the dimension identifier can be obtained from the preset local database and the Internet database.

[0044] The preset local database can be a database established locally in the electronic device.

[0045] The Internet database can be a database updated in real time according to the information on the Internet.

[0046] In this embodiment, for any one dimension identifier, the electronic device can obtain the first initial information from the preset local database based on the target object identifier and the dimension identifier, obtain the second initial information from the Internet database based on the target object identifier and the dimension identifier, and integrate and process the first initial information and the second initial information through the first preset model to obtain the initial information.

[0047] The preset local database can include a preset local map database and / or a preset local vector database.

[0048] The preset local map database can be used to store structured and clearly related knowledge.

[0049] The data mode corresponding to the data stored in the preset local map database can be: object-[has]-information-[includes]-key content, and the data mode can also be referred to as: Object-[Has Report]-Report-[Mentions]-Concept. The object, the information, and the key content can be nodes in the preset local map database, the "has" can be a relationship between the object and the information, and the "includes" can be a relationship between the information and the key content.

[0050] In addition, for any piece of data in the preset local map database, the data can correspond to a unique information ID. The information ID can be generated by the "object" and the "information" corresponding to the data. Specifically, the information ID can be generated by the following formula:

[0051]

[0052] wherein, represents an MD5 hash algorithm, represents a string concatenation operation, represents an object identifier of the object, represents a dimension identifier corresponding to the information included in the data.

[0053] The preset local vector database can be used to store unstructured semantic information. The information can be converted into a high-dimensional vector, and the high-dimensional vector can be stored in the preset local vector database.

[0054] The first preset model can be a pre-trained LLM. For example, the first preset model can be a Qianwen3rd Generation (Qwen3) -235B model, etc. Wherein, 235B is used to indicate the parameter size of the model.

[0055] In this embodiment, on the one hand, information already collected in the preset local database can be obtained, and on the other hand, the latest data can be obtained from the Internet database, which breaks through the timeliness limitation of LLM obtaining information from the inherent database in the related art, and makes the timeliness of information acquisition better.

[0056] It should be noted that the specific method of integrating the first initial information and the second initial information by the first preset model can be referred to in the Figure 2 embodiment, which will not be repeated here.

[0057] In this embodiment, if the target request includes multiple dimension identifiers, the electronic device can generate initial information corresponding to the multiple dimension identifiers in parallel, so that the information acquisition efficiency is higher.

[0058] S103, quality detection is performed on each initial information to obtain a quality detection result, the quality detection result being detection pass or detection fail.

[0059] It should be noted that the method of quality detection on each initial information is the same. In the embodiment of the present application, the method of quality detection on initial information is described by taking any one initial information as an example.

[0060] In this embodiment, for any one initial information, the electronic device can detect the initial information from multiple detection dimensions through the second preset model to obtain improvement suggestions and multiple detection scores; obtain the weight corresponding to each detection dimension in the multiple detection dimensions; determine the quality score corresponding to the initial information according to the multiple detection scores and the weight corresponding to each detection dimension; and determine the quality detection result according to the quality score.

[0061] The second preset model can be an LLM that is pre-set for detecting information quality.

[0062] The second preset model and the first preset model can be different LLMs. The first preset model can select an LLM with strong information integration capability, and the cost of this model is generally higher. The second preset model can select an LLM with faster running speed and lower cost because the cost of the detection task is relatively small. In this way, different LLMs can be selected for different tasks to ensure performance and make the cost of the model lower.

[0063] For example, the second preset model can be a Qwen3-30B model, etc. Wherein, 30B is used to indicate the parameter size of the model.

[0064] This embodiment does not limit the detection dimension, as long as the initial information can be detected from multiple dimensions.

[0065] Optionally, the detection dimension can include information relevance, information integrity, information depth, original text matching degree, and information structure and fluency.

[0066] The information relevance can be used to evaluate the relevance of the initial information, the target object identifier, and the corresponding dimension identifier.

[0067] The information integrity can be used to evaluate whether the first initial information and the second initial information are fully utilized when generating the initial information.

[0068] The information depth can be used to evaluate whether valuable analysis opinions are provided in the initial information.

[0069] The original text matching degree can be used to evaluate whether there is guessing content in the initial information that deviates from the first initial information and the second initial information.

[0070] The information structure and fluency can be used for evaluation, the organization of the initial information and the professionalism of the language expression.

[0071] It should be noted that the second preset model can output a detection score corresponding to each detection dimension.

[0072] The weight corresponding to each detection dimension can be pre-set according to actual needs, which is not limited in the embodiment.

[0073] The improvement suggestion can be an evaluation of the initial information or a suggestion for modification. For example, the improvement suggestion can be "the initial information quality is very high" or "the initial information does not highlight the core theme of 'technical strength', and it is suggested to rewrite and focus on the technical breakthrough of the research and development investment mentioned in the first initial information and the second initial information".

[0074] In the embodiment, the quality score corresponding to the initial information satisfies the following formula:

[0075]

[0076] wherein, is the quality score, N is the total number of detection dimensions, i is 1 to N, is the weight corresponding to the i-th detection dimension, is the detection score of the i-th detection dimension.

[0077] In the embodiment, after determining the quality score corresponding to the initial information, the quality detection result corresponding to the initial information can be determined according to the quality score and the quality threshold.

[0078] Specifically, if the quality score is greater than or equal to the quality threshold, it is determined that the quality detection result is passed; if the quality score is less than the quality threshold, it is determined that the quality detection result is not passed.

[0079] In S104, for any one initial information, if the quality detection result corresponding to the initial information is not passed, the initial information is updated to obtain target information, and the quality detection result corresponding to the target information is passed.

[0080] In the embodiment, if the quality detection result corresponding to the initial information is not passed, the electronic device can update the initial information according to the initial information and the improvement suggestion by using the first preset model to obtain target information.

[0081] Specifically, the electronic device can input the initial information, the improvement suggestion, and the update prompt word into the first preset model, the update prompt word is used to instruct the first preset model to update the initial information according to the improvement suggestion; and the target information can be obtained from the first preset model.

[0082] Optionally, the electronic device can input the first initial information and the second initial information to the first preset model when inputting the initial information, the improvement suggestion, and the update prompt to the first preset model.

[0083] It should be noted that after obtaining the target information, the electronic device can also determine the quality detection result of the target information through the second preset model. If the quality detection result corresponding to the target information is detection failure, the electronic device can continue to update the target information through the first preset model until the quality detection result corresponding to the target information is detection success.

[0084] The following exemplary process of generating target information is described:

[0085] (1) Determine the quality detection result corresponding to the initial information: the electronic device can compare the quality score of the initial information with the quality threshold.

[0086] (2) Loop triggering: if the quality score is less than the quality threshold, the electronic device determines that the quality detection result corresponding to the initial information is detection failure, and can start the update process. The electronic device can send the initial information, the first initial information, the second initial information, and the improvement suggestion to the first preset model again, and instruct the first preset model to update the initial information.

[0087] (3) Guided update: during the update process, the electronic device can switch to an update prompt (Rewrite Prompt) designed specifically for the update scenario. The prompt can be different from the prompt for generating the initial information. The prompt can set the first preset model as a senior analyst who is good at listening to improvement suggestions and making deep modifications, and can take the improvement suggestion output by the second preset model as the most core input and the update instruction that must be followed. The above method can change the first preset model from a free creation mode to a corrector with clear guidance for solving specific problems, so that the update efficiency of the initial information is higher, and the update accuracy is higher.

[0088] (4) Loop termination: the update process of the initial information can be repeatedly executed until the quality score of the target information generated at a certain time is greater than or equal to the quality threshold. In addition, in order to prevent accidental infinite loop execution, the maximum number of repetitions can also be set as a condition for ending the update.

[0089] In this embodiment, by performing quality detection on the initial information, the accuracy of the finally generated object information can be higher. Compared with directly outputting information in related art, the accuracy of the object information is improved.

[0090] S105, updating the database according to the target information, and / or the initial information whose quality detection result is detection success.

[0091] In this embodiment, for any one target information, the database can be updated according to the target information. For any one initial information with a quality detection result of detection pass, the database can be updated according to the initial information with the quality detection result of detection pass.

[0092] It should be noted that the method of updating the database according to any one target information or any one initial information with a quality detection result of detection pass is the same. In this embodiment, the method of updating the database according to the target information is taken as an example to be described.

[0093] In this embodiment, the database can include a preset local database, and the preset local database includes a preset local map database and a preset local vector database. The electronic device can update the preset local map database according to the target information, and update the preset local vector database according to the target information.

[0094] First, the process of updating the preset local map database according to the target information is described.

[0095] When updating the preset local map database according to the target information, the electronic device can extract at least one target key content from the target information, and update the preset local map database according to the at least one target key content.

[0096] When extracting the at least one target key content, the electronic device can input the target information into a third preset model, and obtain at least one keyword from the third preset model; and the at least one keyword is processed to obtain at least one target key content. The third preset model is used to extract a key word in the target information according to a preset theme, and the character length of the keyword is less than or equal to a length threshold. The length threshold can be set according to actual needs.

[0097] The third preset model can be a pre-trained LLM. For example, the third preset model can also be a Qwen3-235B model, etc.

[0098] For example, the third preset model extracts at least one keyword, which can include the following steps:

[0099] (1) Role-based Prompting: The electronic device can invoke the third preset model and assign a highly specialized "information extraction specialist" role to the third preset model. The system prompt words corresponding to the role of the instruction are not generation or analysis, but "recognition and extraction". The electronic device can instruct the third preset model through the prompt words: from the input target information, identify and extract no more than W keywords that best represent the core content of the target information. W can be an integer greater than or equal to 1, for example, W can be equal to 3. The keywords can be nouns or noun phrases, but the keywords are not too general words such as companies, markets, etc. For example, the keywords can be AI server, supply chain risk, etc.

[0100] (2) Mandatory structured output: In order to enable automatic extraction of keywords and lower error rate, the prompt words can include mandatory output format requirements. The electronic device can provide clear examples and instructions to indicate that the keywords output by the third preset model are in the JavaScript Object Notation (JSON) string list format without additional explanatory text. For example, the keywords output by the third preset model can be: ["AI server", "self-developed product"]. This strict constraint on the output format can make the keyword extraction error rate lower.

[0101] (3) Supplement and deduplication: After the electronic device obtains at least one keyword from the third preset model, it can supplement and deduplicate the at least one keyword. First, the electronic device can add the corresponding dimension identifier to the list of at least one keyword to ensure the most context association. Second, the electronic device can perform a deduplication operation on the list of at least one keyword to obtain at least one target key content corresponding to the target information. The deduplication operation can ensure the uniqueness of each target key content.

[0102] In this embodiment, when updating the preset local database according to the at least one target key content, for any one target key content, the electronic device can update the corresponding target node in the preset local graph database according to the target key content; according to the relationship between the at least one target key content, update the relationship between the at least one target node corresponding to the at least one target key content.

[0103] When the target key content is used to update the target node in the preset local map database, the electronic device can determine whether the target key content corresponds to a target node in the preset local map database, and whether the similarity between the target key content and the target node is greater than or equal to a similarity threshold; if the target key content corresponds to a target node in the preset local map database, the target node is updated according to the target key content; if the target key content does not correspond to a target node in the preset local map database, it is determined that the target node is the same as the target key content, and the target node is added to the preset local map database.

[0104] The similarity threshold can be set according to actual needs. For example, the similarity threshold can be 99%.

[0105] Specifically, if the preset local map database contains a target node with a similarity greater than the similarity threshold to the target key content, the electronic device can replace the target node with the target key content. If the preset local map database does not contain a target node with a similarity greater than the similarity threshold to the target key content, the electronic device can add the target key content to the preset local map database.

[0106] For example, when at least one target key content is written to the preset local map database, the electronic device can use the MERGE command of the Cypher query language to perform the operation of "updating if existing, creating if not existing". The MERGE command can create or update the target object corresponding node, the target information corresponding node, and the at least one target key content corresponding node in the preset local map database. The idempotency of the MERGE command ensures that even if it is executed repeatedly, it will not produce duplicate nodes in the preset local map database. Secondly, the electronic device can also use the MERGE command to create or update the "owning" relationship between the target object corresponding node and the target information corresponding node, and create or update the "including" relationship between the target information corresponding node and all related key content corresponding nodes. In this way, the structured knowledge of "which object" owns "target information" and "which key content" is included in "target information" can be accurately described in the preset local map database.

[0107] Secondly, the process of updating the preset local vector database according to the target information is described.

[0108] When the target information is used to update the preset local vector database, the electronic device can divide the target information into a plurality of text blocks of a preset size; determine the semantic vectors of the text blocks to obtain a plurality of semantic vectors; and update the preset local vector database according to the plurality of semantic vectors.

[0109] Specifically, for any semantic vector, the electronic device can store the semantic vector and the metadata corresponding to the semantic vector into the preset local vector database; wherein the metadata corresponding to the semantic vector includes: the target object identifier, and the target information ID corresponding to the target object identifier and the dimension identifier.

[0110] For example, when updating the preset local vector database according to the target information, the electronic device can first divide the target information into a plurality of small text blocks with overlapping semantics and coherent semantics through a text segmenter. Secondly, the electronic device can calculate the semantic vector of each text block through an embedding model, and add the vector and the corresponding metadata to the preset local vector database. The metadata may, for example, be the target object identifier of the target object and the corresponding information ID in the preset local map database. After adding the corresponding information ID to the preset local vector database, the information ID can be used as the association between the preset local vector database and the preset local map database.

[0111] In this embodiment, by updating the preset local map database and the preset local vector database, the stored knowledge in the preset local map database and the preset local vector database can be continuously enriched. Further, the analysis capabilities of the preset local map database and the preset local vector database can be continuously enhanced.

[0112] In this embodiment, storing the target key content into the preset local map database can enrich the data in the preset local map database on the one hand, and the storage space occupied by the target key content is smaller than the storage space occupied by the target information, which can reduce the storage cost of the preset local map database.

[0113] In this embodiment, storing the plurality of semantic vectors corresponding to the target information into the preset local vector database can achieve the purpose of continuously updating the preset local vector database, and the content in the preset local vector database can be continuously enriched.

[0114] S106, merging the target information and / or the initial information with a quality detection result passing to obtain object information of the target object.

[0115] Specifically, the electronic device can determine the output format parameter of the object information according to the configuration file, and the output format parameter is used to indicate the output format of the object information. The output format parameter can also be referred to as "output_format". For example, the output format of the object information can be word format or excel format, etc. In this embodiment, the output format parameter in the configuration file can be updated as needed, so that the generation of the object information has high flexibility.

[0116] In this embodiment, the generation of the object information can include at least the following steps:

[0117] (1) Document structured construction: First, the electronic device can create a document and set a standard header, footer and page number for the document. Second, the electronic device can dynamically generate a large title containing the target object identifier and the generation date of the object information. Optionally, the electronic device can also generate a clickable table of contents for the document.

[0118] (2) Content chapter integration: The electronic device can traverse all dimension identifiers in the target request, and can add the target information corresponding to each dimension identifier or the quality detection result as initial information that passes the detection to the document as an independent chapter, and can add a title to each chapter.

[0119] (3) Source appendix generation: To ensure the professionalism and credibility of the object information, the electronic device can collect the uniform resource identifiers (URLs) of all original information sources used in the entire generation process, and format these URLs into a clear and standardized "source reference" appendix at the end of the document.

[0120] (4) Document naming and saving: The electronic device can generate a unique and information-rich file name based on the target object identifier and the current timestamp accurate to the second. For example, the file name can be "A Company Insight Report 20250702 144540.docx". The electronic device can also save the complete document to the output directory specified by the user to achieve the purpose of generating the object information of the target object.

[0121] (5) Synchronous generation of archive file: While generating the object information of the target object, the electronic device also performs a parallel archiving task to ensure complete traceability and auditability of the entire research process. Specifically, the electronic device can separately organize and save all second initial information collected from the Internet database during the generation of the object information as an additional archive file. This archive file is an important support for the object information. In addition, if manual review or other analysis is required, the archive file can also provide corresponding convenience.

[0122] The object information acquisition method provided in the embodiment can be used to acquire a target request, the target request comprising a target object identifier and at least one dimension identifier; initial information of the target object in the dimensions indicated by the dimension identifiers can be generated to obtain at least one initial information; quality detection can be performed on each initial information to obtain a quality detection result, the quality detection result being detection pass or detection fail; for any one initial information, if the quality detection result corresponding to the initial information is detection fail, the initial information can be updated to obtain target information, the quality detection result corresponding to the target information being detection pass; and the target information and / or the initial information with the quality detection result of detection pass can be merged to obtain object information of the target object. In the method, quality detection can be performed on the initial information of the target object, and the object information of the target object can be generated after the quality detection result of each initial information is detection pass. Through the method, the accuracy of the object information can be improved, and thus the accuracy of information acquisition can be improved.

[0123] In addition, in the technical solution provided in the embodiment, the electronic device can automatically acquire the initial information corresponding to each dimension identifier, automatically perform quality detection on each initial information, and generate the object information of the target object. The manual intervention in the information acquisition process is reduced, and the human resources are saved.

[0124] On the basis of the above-described embodiments, the method of acquiring the initial information will be described below in combination with Figure 2 It should be noted that the method of acquiring the initial information corresponding to each dimension identifier is the same, and the method of acquiring the initial information corresponding to any one dimension identifier will be described in the embodiments of the present application.

[0125] Figure 2 The flowchart of the object information acquisition method provided in the embodiments of the present application is shown in Figure Two Please refer to Figure 2 The execution subject of the method can be an electronic device or a processor in the electronic device. For example, the electronic device can be a server or a memory, and the processor can be a CPU. For the convenience of understanding, the execution subject will be taken as an electronic device in the following description. The method can comprise the following steps.

[0126] S201, acquiring graph database information from a preset local graph database according to the target object identifier and the dimension identifier.

[0127] In this embodiment, the electronic device can determine a target information ID corresponding to the target object identifier and the dimension identifier through a hash algorithm; based on the target information ID, obtain historical information of the target object in the corresponding dimension from a preset local graph database, the information ID of the historical information of the target object in the corresponding dimension and the target information ID have a similarity greater than or equal to a first threshold; based on the target object identifier and the dimension identifier, obtain associated information of the target object from the preset local graph database, the associated information of the target object is used to indicate the historical information corresponding to the dimension identifier; wherein the graph database information includes: the historical information of the target object in the corresponding dimension, and the associated information of the target object.

[0128] Specifically, the electronic device can obtain a preset instruction, the preset instruction is used to obtain historical information from a preset local graph database, the preset instruction includes an information ID filling bit; write the target information ID into the information ID filling bit to obtain an updated instruction; according to the updated instruction, obtain the historical information of the target object in the corresponding dimension from the preset local graph database.

[0129] The preset instruction can be determined according to a query instruction of the preset local graph database, which is not limited in this embodiment.

[0130] For example, the preset instruction can be:

[0131] “MATCH (r:Report {_: …}) RETURN r.content LIMIT 1”. Wherein “_” can be an information ID filling bit.

[0132] The updated instruction can be:

[0133] “MATCH (r:Report {id: …}) RETURN r.content LIMIT 1”. Wherein “id” can be a target information ID.

[0134] In the actual implementation process, the electronic device can determine the target information ID, and can send a preset instruction to the preset local graph database. The preset instruction can include the target information ID. The preset local graph database can find the historical information of the target object in the corresponding dimension in the preset local graph database according to the target information ID.

[0135] In this embodiment, the electronic device can generate the target information ID through the following formula:

[0136]

[0137] Wherein, represents the MD5 hash algorithm, represents the string concatenation operation.

[0138] In this embodiment, the electronic device can find the candidate information including the dimension identifier in the preset local graph database; and delete the information corresponding to the target object from the candidate information according to the target object identifier, to obtain the associated information of the target object.

[0139] For example, the electronic device can query the candidate information including the dimension identifier through a query instruction, which can be, for example:

[0140] “MATCH (c:Company) - [:HasReport] -> (r:Report) - [:Mentions] -> (p:Concept {name: "dimension identifier"}).

[0141] The electronic device can delete the information corresponding to the target object from the candidate information through a filtering instruction, to obtain the associated information of the target object. For example, the filtering instruction can be:

[0142] WHERE c.name <> "object identifier"

[0143] RETURN c.name, r.topic, r.content”.

[0144] Specifically, the electronic device can find all candidate information including the dimension identifier in the preset local graph database according to the dimension identifier; and can find the node of the corresponding object along the "has" relationship of the candidate information. It should be understood that after finding the above candidate information, at least one data in the above candidate information can be determined according to the data mode of the preset local database. The preset local graph database can also filter out the data corresponding to the target object from the at least one data through the "WHERE clause", and the remaining data is the associated information of the target object.

[0145] S202, obtaining vector database information from the preset local vector database according to the dimension identifier.

[0146] In this embodiment, the electronic device can map the dimension identifier into a query vector in a high-dimensional space, and can send the query vector to the preset local vector database. For example, the high-dimensional space can be 1024 dimensions. The preset local vector database can find an existing vector with a cosine similarity less than or equal to a cosine similarity threshold to the query vector through semantic similarity retrieval in the preset local vector database, and can send the existing vector to the electronic device. The electronic device can determine the existing vector as the vector database information.

[0147] The cosine similarity can be calculated by the following formula:

[0148]

[0149] wherein, may be a query vector, may be an existing vector, may be a cosine similarity between the query vector and the existing vector.

[0150] In this embodiment, by pre-setting the local vector database, vector database information similar in concept but different in literal can be effectively obtained, so that the object information is more abundant.

[0151] S203, generating first initial information based on the graph database information and the vector database information.

[0152] In this embodiment, the first initial information can be generated by the following formula:

[0153]

[0154] wherein, is the first initial information, is a formatting function, is used to process texts from different sources into a unified format, refers to a splicing operation of the text content, is historical information of the target object in the corresponding dimension, is the vector database information, is the associated information of the target object.

[0155] It should be noted that the specific description of the formatting function and the splicing operation can be referred to related technologies, which will not be repeated here.

[0156] In this embodiment, by simultaneously setting the pre-set local graph database and the pre-set local vector database, the information acquisition requirements of different types can be met, so that the technical solution provided by the present application can be used in more scenarios.

[0157] In this embodiment, the historical information of the target object in the corresponding dimension, the vector database information, and the associated information of the target object can be three different kinds of context information. Based on the three kinds of information, the first initial information can be generated, so that the information density and information depth of the first initial information are better, and the first initial information is more abundant.

[0158] S204, obtaining second initial information from an Internet database based on the target object identifier and the dimension identifier.

[0159] In this embodiment, the electronic device can obtain at least one instruction template, the instruction template comprising a first filling bit corresponding to a target object identifier and a second filling bit corresponding to a dimension identifier; for any one instruction template, the target object identifier is written into the first filling bit, and the dimension identifier is written into the second filling bit to obtain a corresponding network search instruction; and the second initial information is obtained from an Internet database according to the network search instruction corresponding to each instruction template.

[0160] For example, after the target object identifier is written into the first filling bit and the dimension identifier is written into the second filling bit, the obtained network search instruction can be: "target object identifier" + "dimension identifier".

[0161] Optionally, the instruction template further comprises at least one preset keyword, the preset keyword being used to indicate an information type of the second initial information, the information type comprising a news type or a report type.

[0162] For example, after the target object identifier is written into the first filling bit and the dimension identifier is written into the second filling bit, the obtained network search instruction can be: "target object identifier" + "dimension identifier" + "news". The "news" can be the preset keyword. The "news" is used to indicate a tendency to find corresponding news releases or media reports.

[0163] For another example, after the target object identifier is written into the first filling bit and the dimension identifier is written into the second filling bit, the obtained network search instruction can be: "target object identifier" + "dimension identifier" + "company introduction". The "company introduction" can be the preset keyword. The "company introduction" is used to indicate a tendency to find summary information published on an official website.

[0164] For another example, after the target object identifier is written into the first filling bit and the dimension identifier is written into the second filling bit, the obtained network search instruction can be: "target object identifier" + "dimension identifier" + "filetype:pdf". The "filetype:pdf" can be the preset keyword. The "filetype:pdf" is used to indicate a tendency to find a research report in a pdf format published on an official website.

[0165] In this embodiment, for any one network search instruction, the electronic device can obtain network information corresponding to the network search instruction from an Internet database; and the second initial information comprises the network information corresponding to each network search instruction.

[0166] It should be noted that the method of obtaining the network information corresponding to the network search instruction from the Internet database by the electronic device according to each network search instruction is the same. In this embodiment, the method of obtaining the network information corresponding to the network search instruction from the Internet database is described by taking any one network search instruction as an example.

[0167] Specifically, the electronic device can obtain at least one initial URL according to the network search instruction; perform deduplication processing on the at least one initial URL to obtain at least one target URL; obtain web page content corresponding to each target URL and a publication timestamp; and encapsulate the web page content corresponding to each target URL and the publication timestamp to obtain network information corresponding to the network search instruction.

[0168] For example, when obtaining the web page content corresponding to the target URL, the electronic device can access the target URL through a headless browser framework. In this way, the JavaScript code in the web page corresponding to the target URL can be executed completely, so that the web page content generated by relying on front-end dynamic rendering technology can be captured, overcoming the limitations of traditional Hyper Text Transfer Protocol (HTTP) crawlers. For example, the headless browser framework can be Playwright, and the front-end dynamic rendering technology can be Single-Page Application (SPA).

[0169] If the target URL points to a HyperText Markup Language (HTML) file, the electronic device can use a web text collection library (trafilatura) library specially designed for news, blog, and other article pages to identify and extract the core text area of the corresponding web page. In this way, a large amount of irrelevant noise such as navigation bar, footer, comment area, etc. can be effectively removed.

[0170] If the target URL points to a non-HTML file, the electronic device can automatically identify the file type according to the response header or file extension, and call the corresponding parsing library. For a PDF file, the electronic device can use the PyMuPDF library to extract text from the binary stream in memory; for a Word file, the electronic device can use the python-docx library for parsing.

[0171] When obtaining the publication timestamp corresponding to the target URL, the electronic device can obtain the publication timestamp corresponding to the target URL through a timestamp extraction algorithm. The timestamp extraction algorithm can refer to related technologies, which will not be described here.

[0172] S205, integrating the first initial information and the second initial information through a first preset model to obtain initial information.

[0173] In this embodiment, the electronic device can determine, according to the configuration file, a single string processing length of the first preset model; divide the first initial information and the second initial information into a plurality of to-be-processed information blocks according to the single string processing length, the length of each to-be-processed information block being equal to the single string processing length; process each to-be-processed information block through the first preset model to obtain a target information block corresponding to each to-be-processed information block; and integrate and process the target information blocks corresponding to each to-be-processed information block through the first preset model to obtain the initial information.

[0174] The configuration file can be a configuration file for generating object information of a target object. The configuration file can include parameter values of a plurality of parameters, and the parameter values of the parameters can be default values, or the user can reconfigure the parameter values of the parameters in the configuration file according to actual needs.

[0175] It should be understood that the single string processing length of the first preset model can be one parameter in the configuration file.

[0176] It should be noted that the configuration file can further include other parameters. For example, the other parameters can include one or more of an identifier of the first preset model, an identifier of the second preset model, an identifier of the third preset model, a quality threshold, and an output format parameter.

[0177] When dividing the to-be-processed information blocks, for the first initial information or the second initial information whose content exceeds the single string processing length of the first preset model, the electronic device can automatically call a text segmenter, and can cut the first initial information or the second initial information into a plurality of semantically coherent to-be-processed information blocks in a chunk overlap manner. For example, the text segmenter can be a recursive character text segmenter.

[0178] For any one to-be-processed information block, the electronic device can generate a target information block corresponding to the to-be-processed information block through the first preset model.

[0179] In this embodiment, by dividing the to-be-processed information blocks and processing each to-be-processed information block respectively, a large analysis task can be decomposed into a plurality of independent simple small tasks, so that the success rate of the task is higher.

[0180] When integrating and processing the target information blocks corresponding to each to-be-processed information block through the first preset model, the plurality of target information blocks can be deeply fused to generate the initial information.

[0181] The object information acquisition method provided in the embodiment can obtain graph database information from a preset local graph database according to the target object identifier and the dimension identifier, can obtain vector database information from a preset local vector database according to the dimension identifier, can generate first initial information based on the graph database information and the vector database information, can obtain second initial information from an Internet database based on the target object identifier and the dimension identifier, and can integrate and process the first initial information and the second initial information through a first preset model to obtain initial information. In the method, the electronic device can obtain the first initial information from the preset local database on one hand, and can obtain the first initial information from the Internet database on the other hand, and can generate the initial information according to the first initial information and the second initial information. The initial information generated through the above method is relatively rich and ensures the timeliness of the initial information.

[0182] Through the above description of the embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software and a general hardware platform as required, and of course, can also be implemented by hardware, but in many cases, the former is a better implementation.

[0183] Figure 3 The object information acquisition device provided in the embodiment of the present application is shown in a structural schematic diagram. As shown in the figure, the object information acquisition device 10 can include an acquisition module 11, a detection module 12, an update module 13 and a merging module 14, wherein: Figure 3

[0184] The acquisition module 11 is configured to acquire a target request, wherein the target request includes a target object identifier of a target object and at least one dimension identifier.

[0185] The acquisition module 11 is further configured to acquire initial information of the target object in each dimension indicated by the dimension identifier from a database to obtain at least one initial information.

[0186] The detection module 12 is configured to perform quality detection on each initial information to obtain a quality detection result, wherein the quality detection result is detection pass or detection fail.

[0187] For any one initial information, if the quality detection result corresponding to the initial information is detection fail, the update module 13 is configured to update the initial information to obtain target information, wherein the quality detection result corresponding to the target information is detection pass.

[0188] The update module 13 is further configured to update the database according to the target information and / or the initial information with the quality detection result of detection pass.

[0189] ​The merging module 14 is configured to perform merging processing on the target information and the initial information whose quality detection result is passed to obtain object information of the target object.

[0190] The object information acquisition apparatus provided in the embodiment can perform the method shown in any of the method embodiments, and has similar implementation principles and technical effects. Details are not described herein again.

[0191] In a possible implementation, the database includes a preset local database and an Internet database, and the acquisition module 11 is specifically configured to:

[0192] acquire first initial information from the preset local database based on the target object identifier and the dimension identifier;

[0193] acquire second initial information from the Internet database based on the target object identifier and the dimension identifier;

[0194] integrate the first initial information and the second initial information based on the first preset model to obtain the initial information.

[0195] In a possible implementation, the preset local database includes a preset local map database and a preset local vector database, and the acquisition module 11 is specifically configured to:

[0196] acquire map database information from the preset local map database based on the target object identifier and the dimension identifier;

[0197] acquire vector database information from the preset local vector database based on the dimension identifier;

[0198] generate the first initial information based on the map database information and the vector database information.

[0199] In a possible implementation, the acquisition module 11 is specifically configured to:

[0200] determine a target information ID corresponding to the target object identifier and the dimension identifier by using a hash algorithm;

[0201] acquire historical information of the target object in the corresponding dimension from the preset local map database based on the target information ID, and a similarity between an information ID of the historical information of the target object in the corresponding dimension and the target information ID is greater than or equal to a first threshold value;

[0202] acquire associated information of the target object from the preset local map database based on the target object identifier and the dimension identifier, and the associated information of the target object is used to indicate historical information corresponding to the dimension identifier;

[0203] In a possible implementation, the acquisition module 11 is specifically configured to:

[0204] The preset instruction is used to obtain historical information from a preset local map database, and the preset instruction includes an information ID filling bit;

[0205] The target information ID is written into the information ID filling bit to obtain an updated instruction;

[0206] According to the updated instruction, historical information of the target object in the corresponding dimension is obtained from the preset local map database.

[0207] In a possible implementation, the obtaining module 11 is specifically configured to:

[0208] Find the candidate information including the dimension identifier in the preset local map database;

[0209] According to the target object identifier, the information corresponding to the target object is deleted from the candidate information to obtain the associated information of the target object.

[0210] The graph database information includes the historical information of the target object in the corresponding dimension and the associated information of the target object.

[0211] In a possible implementation, the first initial information satisfies:

[0212]

[0213] wherein, is the first initial information, is the formatting function, is a splicing operation of text content, is the historical information of the target object in the corresponding dimension, is the vector database information, is the associated information of the target object.

[0214] In a possible implementation, the obtaining module 11 is specifically configured to:

[0215] Obtain at least one instruction template, the instruction template including a first filling bit corresponding to the target object identifier and a second filling bit corresponding to the dimension identifier;

[0216] For any one instruction template, the target object identifier is written into the first filling bit, and the dimension identifier is written into the second filling bit to obtain a corresponding network search instruction;

[0217] For any one network search instruction, the network information corresponding to the network search instruction is obtained from the Internet database;

[0218] The second initial information includes the network information corresponding to each network search instruction.

[0219] In a possible implementation, the instruction template further includes at least one preset keyword, and the preset keyword is used to indicate an information type of the second initial information.

[0220] In a possible implementation, the acquisition module 11 is specifically configured to:

[0221] acquire at least one initial website address according to the network search instruction;

[0222] perform deduplication processing on the at least one initial website address to obtain at least one target website address;

[0223] acquire webpage content corresponding to each target website address and a publishing time stamp;

[0224] encapsulate the webpage content corresponding to each target website address and the publishing time stamp to obtain network information corresponding to the network search instruction.

[0225] In a possible implementation, the merging module 14 is specifically configured to:

[0226] determine a single-character string processing length of the first preset model according to the configuration file;

[0227] divide the first initial information and the second initial information into a plurality of to-be-processed information blocks according to the single-character string processing length, and a length of each to-be-processed information block is equal to the single-character string processing length;

[0228] process each to-be-processed information block through the first preset model to obtain a target information block corresponding to each to-be-processed information block;

[0229] perform splicing processing on the target information block corresponding to each to-be-processed information block through the first preset model to obtain the initial information.

[0230] In a possible implementation, for any one initial information, the detection module 12 is specifically configured to:

[0231] detect the initial information from a plurality of detection dimensions through a second preset model to obtain an improvement suggestion and a plurality of detection scores;

[0232] acquire a weight corresponding to each detection dimension in the plurality of detection dimensions;

[0233] determine a quality score corresponding to the initial information according to the plurality of detection scores and the weight corresponding to each detection dimension;

[0234] determine a quality detection result according to the quality score.

[0235] In a possible implementation, the updating module 13 is specifically configured to:

[0236] The initial information, the improvement suggestion, and an update prompt word are input into the first preset model, and the update prompt word is used to indicate that the first preset model updates the initial information according to the improvement suggestion;

[0237] The target information is obtained from the first preset model.

[0238] In a possible implementation, the database includes a preset local database, and the preset local database includes a preset local map database and a preset local vector database; for any one target information, the updating module 13 is specifically configured to:

[0239] update the preset local map database according to the target information; and

[0240] update the preset local vector database according to the target information.

[0241] In a possible implementation, the updating module 13 is specifically configured to:

[0242] extract at least one target key content from the target information;

[0243] update the preset local map database according to the at least one target key content.

[0244] In a possible implementation, the updating module 13 is specifically configured to:

[0245] input the target information into a third preset model, and the third preset model is used to extract a keyword in the target information according to a preset theme, and a character length of the keyword is less than or equal to a length threshold;

[0246] obtain at least one keyword from the third preset model;

[0247] perform a deduplication processing on the at least one keyword to obtain at least one target key content.

[0248] In a possible implementation, the updating module 13 is specifically configured to:

[0249] for any one target key content, update a corresponding target node in the preset local map database according to the target key content;

[0250] update a relationship between at least one target node corresponding to the at least one target key content according to the relationship between the at least one target key content.

[0251] In a possible implementation, the updating module 13 is specifically configured to:

[0252] determine whether a target node corresponding to the target key content exists in the preset local map database, and a similarity between the target key content and the target node is greater than or equal to a similarity threshold;

[0253] If a target node corresponding to the target's key content exists in the preset local graph database, the target node is updated based on the target's key content;

[0254] If the target node corresponding to the target key content does not exist in the preset local graph database, determine that the target node is the same as the target key content, and add the target node to the preset local graph database.

[0255] In one possible implementation, update module 13 is specifically used for:

[0256] Divide the target information into multiple text blocks of a preset size;

[0257] Determine the semantic vector of each text block to obtain multiple semantic vectors;

[0258] Update the preset local vector database based on multiple semantic vectors.

[0259] In one possible implementation, update module 13 is specifically used for:

[0260] For any given semantic vector, the semantic vector and its corresponding metadata are stored in a pre-defined local vector database.

[0261] The metadata corresponding to the semantic vector includes: the target object identifier, and the target information ID corresponding to the target object identifier and the dimension identifier.

[0262] The object information acquisition device provided in this embodiment can execute the method shown in any of the above method embodiments. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0263] It should be noted that the division of functional units in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0264] Figure 4 A schematic diagram of the structure of the electronic device provided in this application. Figure 4 As shown, the electronic device 20 provided in this embodiment includes at least one processor 201 and a memory 202. Optionally, the electronic device 20 further includes a communication component 203. The processor 201, memory 202, and communication component 203 are connected via a bus.

[0265] In the implementation process, the at least one processor 201 executes the computer execution instructions stored in the memory 202, so that the at least one processor 201 executes the above-mentioned object information acquisition method embodiment.

[0266] The specific implementation process of the processor 201 can refer to the above-mentioned method embodiments, which have similar implementation principles and technical effects, and will not be repeated here.

[0267] In the above-mentioned embodiments, it should be understood that the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the application can be directly embodied as hardware processor execution, or executed by hardware and software modules in the processor.

[0268] The memory can include a random access memory (RAM), and can also include a non-volatile memory (NVM), for example, at least one disk memory.

[0269] The bus can be an industry standard architecture (ISA) bus, a peripheral component (PCI) bus, or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of the present application does not limit only one bus or one type of bus.

[0270] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program, wherein the computer program is configured to execute the steps in any of the above-mentioned object information acquisition method embodiments when running.

[0271] In an example embodiment, the computer readable storage medium described above can include, but is not limited to, a U disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store computer programs.

[0272] Embodiments of the present application also provide a computer program product, which includes a computer program, and the computer program, when executed by a processor, implements the steps in any of the object information acquisition method embodiments described above.

[0273] Embodiments of the present application also provide another computer program product, which includes a non-volatile computer readable storage medium, and the non-volatile computer readable storage medium stores a computer program, and the computer program, when executed by a processor, implements the steps in any of the object information acquisition method embodiments described above.

[0274] The skilled in the art can further realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been described in the above description in a general manner. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. The skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0275] The above describes in detail the object information acquisition method, device, storage medium and program product provided by the present application. The principles and implementation modes of the present application are described by applying specific examples in this paper, and the above example descriptions are only used to help understand the method of the present application and its core idea. It should be pointed out that, for those skilled in the art, without departing from the principles of the present application, some improvements and modifications can be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. An object information acquisition method characterized by comprising: The method comprises: obtaining a target request, wherein the target request comprises a target object identifier of a target object and at least one dimension identifier; obtaining initial information of the target object in a dimension indicated by each dimension identifier from a database, to obtain at least one initial information; detecting the initial information from a plurality of detection dimensions by a second preset model to obtain improvement suggestions and a plurality of detection scores; obtaining a weight corresponding to each detection dimension in the plurality of detection dimensions; determining a quality score corresponding to the initial information according to the plurality of detection scores and the weight corresponding to each detection dimension; determining a quality detection result according to the quality score, wherein the quality detection result is detection pass or detection fail; for any one initial information, if the quality detection result corresponding to the initial information is detection fail, inputting the initial information, the improvement suggestions and an update prompt word into a first preset model, wherein the update prompt word is used to instruct the first preset model to update the initial information according to the improvement suggestions; obtaining target information from the first preset model, wherein a quality detection result corresponding to the target information is detection pass; updating the database according to the target information and / or initial information with a quality detection result of detection pass; merging the target information and / or initial information with a quality detection result of detection pass to obtain object information of the target object.

2. The method of claim 1, wherein, The database comprises a preset local database and an Internet database; for any one dimension identifier, obtaining initial information of the target object in a dimension indicated by the dimension identifier from a database comprises: obtaining first initial information from the preset local database based on the target object identifier and the dimension identifier; obtaining second initial information from the Internet database based on the target object identifier and the dimension identifier; integrating the first initial information and the second initial information by a first preset model to obtain the initial information.

3. The method of claim 2, wherein, The preset local database comprises a preset local map database and a preset local vector database; obtaining first initial information from the preset local database based on the target object identifier and the dimension identifier comprises: obtaining map database information from the preset local map database according to the target object identifier and the dimension identifier; obtaining vector database information from the preset local vector database according to the dimension identifier; generating the first initial information based on the map database information and the vector database information.

4. The method of claim 3, wherein, Obtaining map database information from the preset local map database according to the target object identifier and the dimension identifier comprises: determining a target information ID corresponding to the target object identifier and the dimension identifier by a hash algorithm; obtaining historical information of the target object in a corresponding dimension from the preset local map database based on the target information ID, wherein a similarity between an information ID of the historical information of the target object in the corresponding dimension and the target information ID is greater than or equal to a first threshold. obtain, based on the target object identifier and the dimension identifier, associated information of the target object from the preset local map database, the associated information of the target object being used to indicate historical information corresponding to the dimension identifier; The graph database information includes historical information of the target object in the corresponding dimension and the associated information of the target object.

5. The method of claim 4, wherein, Obtaining, based on the target information ID, historical information of the target object in the corresponding dimension from the preset local map database includes: Obtaining a preset instruction, the preset instruction being used to obtain historical information from the preset local map database, the preset instruction including an information ID filling bit; Write the target information ID into the information ID filling bit to obtain an updated instruction; According to the updated instruction, obtain the historical information of the target object in the corresponding dimension from the preset local map database.

6. The method of claim 4, wherein, Obtaining, based on the target object identifier and the dimension identifier, associated information of the target object from the preset local map database includes: Finding candidate information including the dimension identifier in the preset local map database; According to the target object identifier, delete the information corresponding to the target object from the candidate information to obtain the associated information of the target object.

7. The method of claim 4, wherein, The first initial information satisfies: Wherein, the first initial information is The first initial information is The formatting function is The splicing operation of text content is The historical information of the target object in the corresponding dimension is The vector database information is The association information of the target object is 8. The method of claim 2, wherein, Obtaining, based on the target object identifier and the dimension identifier, second initial information from the Internet database includes: Obtaining at least one instruction template, the instruction template including a first filling bit corresponding to the target object identifier and a second filling bit corresponding to the dimension identifier; For any one instruction template, write the target object identifier into the first filling bit and write the dimension identifier into the second filling bit to obtain a corresponding network search instruction; For any one network search instruction, obtain network information corresponding to the network search instruction from the Internet database; The second initial information includes network information corresponding to each network search instruction.

9. The method of claim 8, wherein, The instruction template further includes at least one preset keyword, the preset keyword being used to indicate the information type of the second initial information.

10. The method of claim 8, wherein, Obtaining network information corresponding to the network search instruction from the Internet database includes: According to the network search instruction, obtain at least one initial website; Perform deduplication processing on the at least one initial website to obtain at least one target website; Obtain web page content and a publication time stamp corresponding to each target website; Package the web page content and the publication time stamp corresponding to each target website to obtain network information corresponding to the network search instruction.

11. The method of claim 2, wherein, Integrating the first initial information and the second initial information through a first preset model to obtain the initial information includes: According to a configuration file, determine a single string processing length of the first preset model; According to the single string processing length, divide the first initial information and the second initial information into a plurality of to-be-processed information blocks, the length of the to-be-processed information block being equal to the single string processing length; The first preset model is used to process each to-be-processed information block to obtain a target information block corresponding to each to-be-processed information block; The first preset model is used to splice the target information block corresponding to each to-be-processed information block to obtain the initial information.

12. The method of claim 1, wherein, The database includes a preset local database, and the preset local database includes a preset local map database and a preset local vector database; for any target information, the database is updated according to the target information, including: The preset local map database is updated according to the target information; and The preset local vector database is updated according to the target information.

13. The method of claim 12, wherein, The preset local map database is updated according to the target information, including: At least one target key content is extracted from the target information; The preset local map database is updated according to the at least one target key content.

14. The method of claim 13, wherein, At least one target key content is extracted from the target information, including: The target information is input into a third preset model, and the third preset model is used to extract a keyword in the target information according to a preset theme, and a character length of the keyword is less than or equal to a length threshold; At least one keyword is obtained from the third preset model; The at least one keyword is de-duplicated to obtain the at least one target key content.

15. The method according to claim 13 or 14, characterized in that, The preset local map database is updated according to the at least one target key content, including: For any target key content, a target node corresponding to the target key content is updated in the preset local map database according to the target key content; According to the relationship between the at least one target key content, the relationship between at least one target node corresponding to the at least one target key content is updated.

16. The method of claim 15, wherein, The target node corresponding to the target key content is updated in the preset local map database, including It is determined whether the target node corresponding to the target key content exists in the preset local map database, and a similarity between the target key content and the target node is greater than or equal to a similarity threshold; If the target node corresponding to the target key content exists in the preset local map database, the target node is updated according to the target key content; If the target node corresponding to the target key content does not exist in the preset local map database, it is determined that the target node is the same as the target key content, and the target node is added to the preset local map database.

17. The method of claim 12, wherein, The preset local vector database is updated according to the target information, including: The target information is divided into a plurality of text blocks of a preset size; A semantic vector of each text block is determined to obtain a plurality of semantic vectors; The preset local vector database is updated according to the plurality of semantic vectors.

18. The method of claim 17, wherein, The preset local vector database is updated according to the plurality of semantic vectors, including: For any semantic vector, the semantic vector and metadata corresponding to the semantic vector are stored in the preset local vector database; The metadata corresponding to the semantic vector includes the target object identifier and a target information ID corresponding to the target object identifier and the dimension identifier.

Images