Data query method and system based on MA code and computing device

Abstract

The invention discloses a data query method and system based on MA codes and computing equipment, and is applied to the field of intelligent questions and answers. The method comprises the steps that each keyword in at least one keyword is converted into an MA code; and querying in a corpus based on the MA code of the at least one keyword to obtain an answer to the original question, the corpus storing a plurality of target corpora, and the target corpora carrying the MA code of the keyword. The MA code is introduced into the data query, and the accuracy of the answer fed back to the user in the intelligent question answering can be improved by utilizing the hierarchical structure of the MA code.

Description

Technical Field

[0001] The present application relates to the field of intelligent question answering, and in particular to a data query method, system and computing device based on MA code. Background Art

[0002] Intelligent Question Answering (IQA) is a core subfield in Natural Language Processing (NLP), which aims to design and develop systems that can parse, understand and answer natural language questions raised by users.

[0003] In the field of intelligent question answering, the data query method based on keyword matching is a common and important means of information retrieval. This method first converts the question input by the user into a form that can be processed by the computer, and then uses keyword matching technology to retrieve answers related to the question from the database.

[0004] However, due to the complexity of natural language, for example, the existence of multiple meanings of a word, the system may not be able to accurately match the question entered by the user with the answer in the database. Summary of the invention

[0005] In order to improve the accuracy of answers fed back to users in intelligent question and answering, the present invention provides a data query method, system and computing device based on MA code.

[0006] In the first aspect, the embodiment of the present application provides a data query method based on MA code. The method includes: obtaining a query request initiated by a user, the query request includes an original question input by the user; preprocessing the original question to obtain a target text; extracting at least one keyword from the target text; converting each of the at least one keyword into an MA code; querying in a corpus based on the MA code of the at least one keyword to obtain an answer to the original question, the corpus stores multiple target corpora, the target corpora carry MA codes of their keywords, and the answer includes at least one target corpus; feeding back the answer to the original question to the user.

[0007] In some embodiments, when the at least one keyword is multiple keywords, each keyword in the at least one keyword is converted into an MA code, including: determining one or more candidate MA codes for each keyword in the multiple keywords; combining the attribute fields of the candidate MA codes of the multiple keywords to obtain multiple attribute combinations, wherein each attribute combination is composed of multiple attribute fields corresponding one-to-one to the multiple keywords; for each of the multiple attribute combinations, performing semantic analysis on the multiple attribute fields in the attribute combination to obtain a consistency result of the attribute combination, wherein the consistency result reflects the semantic consistency of the multiple attribute fields; based on the consistency results of the multiple attribute combinations, screening out a target attribute combination from the multiple attribute combinations; screening out multiple target MA codes that meet the target attribute combination from the candidate MA codes of the multiple keywords, wherein the multiple target MA codes correspond one-to-one to the multiple keywords; and determining the MA codes of the multiple keywords based on the multiple target MA codes.

[0008] In some embodiments, determining one or more candidate MA codes for each of the multiple keywords includes: inputting each of the multiple keywords into a first encoding model to obtain one or more candidate MA codes for the keyword, wherein the first encoding model is a trained first machine learning model.

[0009] In some embodiments, each of the at least one keyword is converted into an MA code, including: generating an input instance based on the at least one keyword; wherein, when the at least one keyword is a single keyword, the input instance is the keyword; when the at least one keyword is multiple keywords, the input instance is a concatenation result of the multiple keywords; inputting the input instance into a second encoding model to obtain the MA code of the at least one keyword, the second encoding model being a trained second machine learning model.

[0010] In some embodiments, the query request also includes the user's identifier, and the answer to the original question is fed back to the user, including: determining whether the answer contains the target corpus in ciphertext form; in response to determining that the answer does not contain the target corpus in ciphertext form, feeding back the answer in full to the user; in response to determining that the answer contains the target corpus in ciphertext form, determining whether the user's identifier belongs to a set of authorized user identifiers; in response to determining that the user's identifier belongs to the set of authorized user identifiers, decrypting the target corpus in ciphertext form to obtain the target corpus in plaintext form; re-encrypting the target corpus in plaintext form with the user's public key to obtain a re-encrypted target corpus; replacing the target corpus in ciphertext form in the answer with the re-encrypted target corpus to obtain an updated answer; feeding back the updated answer to the user; in response to determining that the user's identifier does not belong to the set of authorized user identifiers, feeding back the answer excluding the target corpus in ciphertext form to the user.

[0011] In some embodiments, the MA code is generated based on the MA identification system of ISO / IEC15459 specification.

[0012] In the second aspect, an embodiment of the present application provides a data query system based on MA codes. The system includes an acquisition module, a preprocessing module, an extraction module, a conversion module, a query module and a feedback module. The acquisition module is used to obtain a query request initiated by a user, and the query request includes an original question input by the user. The preprocessing module is used to preprocess the original question to obtain a target text. The extraction module is used to extract at least one keyword from the target text. The conversion module is used to convert each of the at least one keyword into an MA code. The query module is used to query in a corpus based on the MA code of the at least one keyword to obtain an answer to the original question. The corpus stores multiple target corpora, and the target corpora carry the MA codes of their keywords, and the answer includes at least one target corpus. The feedback module is used to feed back the answer to the original question to the user.

[0013] In a third aspect, an embodiment of the present application provides a computing device, including a processor and a memory, wherein a computer program is stored in the memory. When the processor executes the computer program, the data query method based on the MA code and based on the object unified identification as described in the first aspect is implemented.

[0014] Compared with the prior art, the technical solution provided by the embodiments of the present application has the following advantages:

[0015] (1) Introducing MA codes into data queries and utilizing the hierarchical structure of MA codes can improve the accuracy of the answers fed back to users in intelligent question and answering.

[0016] (2) Re-encryption not only improves the security level, but also avoids the risk of a large amount of data in the corpus losing encryption protection due to the leakage of the unified key. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] Figure 1 An exemplary flow chart of a data query method based on MA code provided in an embodiment of the present application.

[0019] Figure 2 An exemplary flowchart of the process of generating MA codes for multiple keywords provided in an embodiment of the present application.

[0020] Figure 3 An exemplary flowchart of the answer feedback process based on user permissions provided in an embodiment of the present application.

[0021] Figure 4 An exemplary block diagram of a data query system for an MA code provided in an embodiment of the present application.

[0022] Figure 5 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the embodiments of the present application will be further described below. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0024] In the following description, many specific details are set forth to facilitate a full understanding of the present application, but the present application can also be implemented in other ways different from those described herein. Obviously, the embodiments in the specification are only part of the embodiments of the present application, not all of the embodiments.

[0025] Figure 1 100 is an exemplary flow chart of a data query method based on MA code provided in an embodiment of the present application. Process 100 may be executed by a processor. In some embodiments, process 100 may be executed by Figure 4 The system 400 shown in FIG. Figure 1 As shown, process 100 includes the following steps.

[0026] Step 110: obtaining a query request initiated by a user, wherein the query request includes an original question input by the user.

[0027] A user may initiate a query request through a user terminal, which includes but is not limited to a smart phone, a tablet computer, a laptop computer, and a desktop computer.

[0028] Step 120, preprocessing the original question to obtain a target text.

[0029] In some embodiments, the preprocessing includes text standardization, text cleaning, word segmentation, stop word removal, error correction, etc.

[0030] Step 130: extract at least one keyword from the target text.

[0031] Any known technology may be used to implement keyword extraction, for example, a keyword extraction method based on statistics, a keyword extraction method based on machine learning, a keyword extraction method based on graphs, etc.

[0032] Step 140: Convert each keyword in the at least one keyword into an MA code.

[0033] Exemplarily, the MA coding structure complies with the international standard ISO / IEC15459 "Technical Unique Identification for Automatic Identification and Data Capture of Information Technology", which is a hierarchical coding structure divided into three parts. The first part is the user identifier. The first node is the root identifier prefix "MA", which is a global code authorized by the three major organizations of ISO, CEN, and AIM. It also includes country, industry field, etc.; the second part is the object category, which is subdivided according to the industry field, including a hierarchical structure such as major categories and detailed categories; the third part is the individual code of the custom identification object, which is used to represent the attributes and codes of individuals in the detailed category. MA coding can be used in multiple fields, such as natural resources, three-dimensional real scenes, steel, language processing, etc. The three parts of MA coding reflect the hierarchical relationship from coarse to fine. For ease of understanding, in the MA code example provided in this article, the first part is simplified to MA, and the second and third parts are both expressed in natural language (Chinese). For details, see Figure 2 Related instructions.

[0034] In some embodiments, each of the at least one keyword can be converted into an MA code by a machine learning method. Specifically, an input instance can be generated based on the at least one keyword, wherein: when the at least one keyword is a single keyword, the input instance is the keyword; when the at least one keyword is multiple keywords, the input instance is the splicing result of the multiple keywords. Each keyword can be spliced ​​in sequence according to the positional relationship in the target text, and two adjacent keywords are connected by a preset connector. Furthermore, the input instance can be input into a second encoding model to obtain the MA code of the at least one keyword, and the second encoding model is a trained second machine learning model. Each training sample of the second encoding model includes a model input and a sample label, wherein the model input is a single keyword or a splicing result of multiple keywords, and the sample label is the MA code of one or more keywords.

[0035] In some embodiments, the second encoding model includes a neural network, for example, a deep learning network. In some embodiments, the second encoding model includes a recurrent neural network (RNN). RNN is a deep learning network whose structural feature is that the nodes between hidden layers have cyclic connections, which enables the network to capture contextual information in sequence data (for example, multiple keywords in the same text). Therefore, converting each keyword in the at least one keyword into an MA code through RNN helps to improve the encoding quality when multiple keywords are extracted, thereby helping to improve the accuracy of the answer.

[0036] In some embodiments, when the at least one keyword is multiple keywords, the keywords can be selected according to Figure 2 The process shown generates MA codes for the multiple keywords.

[0037] Step 150, querying in the corpus based on the MA code of the at least one keyword to obtain an answer to the original question.

[0038] The corpus stores a plurality of target corpora, each of which carries an MA code of a keyword, and the answer includes at least one target corpus.

[0039] The query process, namely the MA code matching process, can calculate the matching degree between the MA code of the keyword of each target corpus and the MA code of the at least one keyword, and determine at least one target corpus with the highest matching degree as the answer to the original question.

[0040] Exemplarily, the matching degree can be determined based on the ratio of the number of overlapping MA codes to the target number. The overlapping MA code refers to the MA code that appears in both the MA code of the keyword of the target corpus and the MA code of the at least one keyword, and the target number refers to the number of MA codes corresponding to the MA code of the at least one keyword. For example, the MA code of the at least one keyword is 5 MA codes in total, 4 of which appear in the MA code of the keyword of the target corpus, then the target number is 5, and the matching degree is 80%. Alternatively, the matching degree can be obtained by weighted calculation based on the overlapping MA codes that appear in the MA code of the keyword of the target corpus. Each MA code has a preset weight. For example, if only one overlapping MA code appears in the MA code of the keyword of the target corpus, the weight of the overlapping MA code is determined as the matching degree. For another example, if more than two overlapping MA codes appear in the MA code of the keyword of the target corpus, the sum of the weights of these overlapping MA codes is determined as the matching degree. The weighted calculation takes into account the importance differences of different keywords, which is conducive to improving the accuracy of the matching results.

[0041] Step 160: Feedback the answer to the original question to the user.

[0042] In some embodiments, in order to protect data privacy, some target corpora in the corpus need to restrict access rights. In this case, the feedback of answers needs to consider user permissions. Specifically, the process of answer feedback based on user permissions can be referred to Figure 3 and its related description.

[0043] Figure 2 The following is an exemplary flow chart of the MA code generation process for multiple keywords provided in the embodiment of the present application. Figure 2 As shown, step 140 includes the following steps.

[0044] Step 210, determining one or more candidate MA codes for each of the plurality of keywords.

[0045] The case where each keyword has only one candidate MA code is not considered here. In this case, the candidate MA code of each keyword is directly determined as its MA code.

[0046] In some embodiments, each of the multiple keywords is input into a first coding model to obtain one or more candidate MA codes for the keyword, and the first coding model is a trained first machine learning model. Each training sample of the first coding model may include a sample keyword and one or more candidate MA codes for the sample keyword, wherein the sample keyword is the model input and the one or more candidate MA codes for the sample keyword are sample labels. In some embodiments, the first coding model includes a neural network, for example, a deep learning network.

[0047] Step 220: Combining the attribute fields of the candidate MA codes of the multiple keywords to obtain multiple attribute combinations.

[0048] Each attribute combination is composed of a plurality of attribute fields corresponding one-to-one to the plurality of keywords.

[0049] As long as any keyword has multiple (two or more) candidate keywords, multiple (two or more) attribute combinations can be obtained through step 220. For example, the original question input by the user is "What are the benefits of millet and apple porridge?" After preprocessing, the keywords "millet", "apple", "porridge" and "effects" are extracted. Among them, the keyword "millet" has 2 candidate MA codes, recorded as "MA. Food. Millet" and "MA. Company. Xiaomi Technology", the keyword "apple" has 2 candidate MA codes, recorded as "MA. Fruit. Apple" and "MA. Company. Apple Technology", the keyword "porridge" has only 1 candidate MA code, recorded as "MA. Food. Porridge", and the keyword "effects" has only one candidate MA code, recorded as "MA. Item Category. Efficacy". Based on this, a total of 2×3×1×1=6 attribute combinations can be obtained.

[0050] Step 230: for each of the multiple attribute combinations, semantic analysis is performed on multiple attribute fields in the attribute combination to obtain a consistency result of the attribute combination.

[0051] The consistency result reflects the semantic consistency of multiple attribute fields. In some embodiments, the consistency result of each attribute combination is determined based on the semantic similarity of each pair of elements in the attribute combination (each attribute field is regarded as an element), for example, the average of the semantic similarities of each pair of elements is used as the consistency result. The consistency result can be in the form of a score (for example, a value range of 0 to 1) or in the form of a label (for example, consistent / inconsistent).

[0052] Step 240: based on the consistency results of the multiple attribute combinations, select a target attribute combination from the multiple attribute combinations.

[0053] It can be understood that step 240 is intended to screen out target attribute combinations whose internal elements have semantic consistency. Continuing with the previous example, among the six attribute combinations, through the hierarchical relationship in the MA code structure of the keywords, such as user identification and object category, or in the object category, the target attribute combination is screened in different hierarchical combinations, and "grain-fruit-porridge-item category" has semantic consistency, for example, the highest consistency score or the consistency label is consistent. Therefore, "grain-fruit-porridge-item category" is determined as the target attribute combination.

[0054] It is worth noting that if there are multiple target attribute combinations, one version of the answer can be fed back based on each target attribute combination. For example, if the consistency scores of two attribute combinations are both the highest scores or the consistency labels are both the same, two versions of the answer can be output. Of course, you can also feedback to the user that the original question does not contain enough information.

[0055] Step 250: Filter out a plurality of target MA codes that meet the target attribute combination from the candidate MA codes of the plurality of keywords.

[0056] The multiple target MA codes correspond to the multiple keywords one by one. Continuing with the previous example, the target attribute combination is "grain-fruit-porridge-item category". From the hierarchical relationship of the MA code, it can be known that the multiple target MA codes that meet the "grain-fruit-porridge-item category" are "MA. Grain. Millet", "MA. Fruit. Apple", "MA. Food. Porridge" and "MA. Item category. Efficacy".

[0057] Step 260: Determine the MA codes of the multiple keywords based on the multiple target MA codes.

[0058] For each of the multiple keywords, the target MA code corresponding to the keyword can be determined as its MA code. For example, "MA. Grain. Millet" can be determined as the MA code of the keyword "millet" for subsequent MA code matching.

[0059] Figure 3 An exemplary flow chart of the answer feedback process based on user rights provided in the embodiment of the present application. Figure 3 As shown, step 160 includes the following steps.

[0060] Step 310, determining whether the answer contains the target corpus in ciphertext form.

[0061] The target corpus in plain text and the target corpus in cipher text can be distinguished by different header information (also called encryption identifier), for example, "0" is used to identify the target corpus in plain text, and "1" is used to identify the target corpus in plain text. That is, it can be determined whether the answer contains the target corpus in cipher text according to the encryption identifier of each target corpus in the answer.

[0062] In response to determining that the answer does not include the target corpus in ciphertext form, the feedback module 460 executes step 320 .

[0063] Step 320: Feedback the complete answer to the user.

[0064] The answer in plain text form can be fully fed back to the user.

[0065] In response to determining that the answer includes the target corpus in ciphertext form, the feedback module 460 executes step 330 .

[0066] Step 330: Determine whether the user's identifier belongs to the authorized user identifier set.

[0067] Step 330 is to determine whether the user is an authorized user.

[0068] In response to determining that the user identifier belongs to the set of authorized user identifiers, the feedback module executes steps 340 to 370 .

[0069] Step 340: decrypt the target corpus in ciphertext form to obtain the target corpus in plaintext form.

[0070] Step 350: re-encrypt the target corpus in plain text form using the public key of the user to obtain a re-encrypted target corpus.

[0071] The public key of the user may be included in the query request of the user, or may be obtained according to the identification of the user.

[0072] In some embodiments, steps 340 and 350 may be performed in a trusted execution environment (TEE), so that data tampering and theft during the encryption and decryption process can be avoided, thereby enhancing data security.

[0073] Step 360, replacing the ciphertext target corpus in the answer with the re-encrypted target corpus to obtain an updated answer.

[0074] Step 370: Feedback the updated answer to the user.

[0075] After re-encryption, different authorized users can use their own private keys to decrypt the data. Compared with using a unified symmetric key for encryption and decryption, this approach not only improves the security level (the security of asymmetric encryption is higher than that of symmetric encryption), but also avoids the risk of a large amount of data in the corpus losing encryption protection due to the leakage of a unified key.

[0076] In response to determining that the user identifier does not belong to the set of authorized user identifiers, the feedback module executes step 380 .

[0077] Step 380: Feedback the answer of the target corpus without the ciphertext form to the user.

[0078] When the user is not an authorized user, the ciphertext portion of the answer is removed and the plaintext portion of the answer is fed back to the user.

[0079] Figure 4This is an exemplary block diagram of the data query system for MA codes provided in the embodiments of the present application. Figure 4 As shown, the system 400 includes an acquisition module 410 , a preprocessing module 420 , an extraction module 430 , a conversion module 440 , a query module 450 and a feedback module 460 .

[0080] The acquisition module 410 is used to acquire a query request initiated by a user, where the query request includes an original question input by the user.

[0081] The preprocessing module 420 is used to preprocess the original question to obtain the target text.

[0082] The extraction module is used for 430: extracting at least one keyword from the target text.

[0083] The conversion module is configured to: convert 440 each keyword of the at least one keyword into an MA code.

[0084] The query module 450 is used to query the corpus based on the MA code of the at least one keyword to obtain the answer to the original question. The corpus stores multiple target corpora, each of which carries the MA code of its keyword, and the answer includes at least one target corpus.

[0085] The feedback module 460 is used to feed back the answer to the original question to the user.

[0086] For more details about the system 400 and its modules, please refer to Figure 1 and related instructions.

[0087] Figure 5 This is a schematic diagram of the structure of a computing device provided in an embodiment of the present application. Figure 5 As shown, the computing device 500 includes a processor 510 and a memory 520. A computer program is stored in the memory 520. When the processor 510 executes the computer program, the data query method provided in the embodiment of the present application is implemented.

[0088] More details about the data query method can be found in Figure 1 and its related instructions, which will not be repeated here.

[0089] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusions, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to the process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including a..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0090] The above is only an embodiment of the present application, and the embodiment enables those skilled in the art to understand and implement the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and features disclosed herein.

Claims

1. A data query method based on MA code, characterized in that: include: Acquire a query request initiated by a user, wherein the query request includes an original question input by the user; Preprocessing the original question to obtain a target text; extracting at least one keyword from the target text; converting each of the at least one keyword into an MA code; Based on the MA code of the at least one keyword, a query is performed in a corpus to obtain an answer to the original question, wherein the corpus stores a plurality of target corpora, the target corpora carry the MA codes of their keywords, and the answer includes at least one target corpus; The answer to the original question is fed back to the user.

2. The method according to claim 1, characterized in that When the at least one keyword is a plurality of keywords, converting each keyword in the at least one keyword into an MA code comprises: determining one or more candidate MA codes for each keyword in the plurality of keywords; Combining the attribute fields of the candidate MA codes of the multiple keywords to obtain multiple attribute combinations, wherein each attribute combination is composed of multiple attribute fields corresponding to the multiple keywords one by one; For each of the multiple attribute combinations, perform semantic analysis on multiple attribute fields in the attribute combination to obtain a consistency result of the attribute combination, where the consistency result reflects the semantic consistency of the multiple attribute fields; Based on the consistency results of the multiple attribute combinations, a target attribute combination is screened out from the multiple attribute combinations; Filtering out a plurality of target MA codes that meet the target attribute combination from the candidate MA codes of the plurality of keywords, wherein the plurality of target MA codes correspond to the plurality of keywords one by one; The MA codes of the plurality of keywords are determined based on the plurality of target MA codes.

3. The method according to claim 2, characterized in that Determining one or more candidate MA codes for each of the plurality of keywords includes: Each of the multiple keywords is input into a first coding model to obtain one or more candidate MA codes for the keyword, and the first coding model is a trained first machine learning model.

4. The method according to claim 1, characterized in that Converting each keyword of the at least one keyword into an MA code comprises: Generate an input instance based on the at least one keyword; wherein, when the at least one keyword is a single keyword, the input instance is the keyword; when the at least one keyword is multiple keywords, the input instance is a concatenation result of the multiple keywords; The input instance is input into a second encoding model to obtain the MA code of the at least one keyword, and the second encoding model is a trained second machine learning model.

5. The method according to claim 1, characterized in that The query request also includes the identifier of the user; Feeding back the answer to the original question to the user, including: Determining whether the answer contains the target corpus in ciphertext form; In response to determining that the answer does not contain the target corpus in ciphertext form, feeding back the answer in full to the user; In response to determining that the answer includes the target corpus in ciphertext form, determining whether the user identifier belongs to a set of authorized user identifiers; In response to determining that the user's identifier belongs to the authorized user identifier set, decrypting the target corpus in ciphertext form to obtain the target corpus in plaintext form; re-encrypting the target corpus in plaintext form using the user's public key to obtain a re-encrypted target corpus; replacing the target corpus in ciphertext form in the answer with the re-encrypted target corpus to obtain an updated answer; and feeding back the updated answer to the user; In response to determining that the user identifier does not belong to the authorized user identifier set, an answer excluding the target corpus in ciphertext form is fed back to the user.

6. The method according to claim 1, characterized in that The MA code is generated based on the MA identification system of ISO / IEC15459 specification.

7. A data query system based on MA code, characterized in that: include: An acquisition module, used to acquire a query request initiated by a user, wherein the query request includes an original question input by the user; A preprocessing module, used for preprocessing the original question to obtain a target text; An extraction module, used to extract at least one keyword from the target text; a conversion module, configured to convert each keyword of the at least one keyword into an MA code; A query module, configured to query a corpus based on the MA code of the at least one keyword to obtain an answer to the original question, wherein the corpus stores a plurality of target corpora, the target corpora carry the MA codes of their keywords, and the answer includes at least one target corpus; The feedback module is used to: feed back the answer to the original question to the user.

8. The system according to claim 7, characterized in that When the at least one keyword is a plurality of keywords, converting each keyword in the at least one keyword into an MA code comprises: determining one or more candidate MA codes for each keyword in the plurality of keywords; Combining the attribute fields of the candidate MA codes of the multiple keywords to obtain multiple attribute combinations, wherein each attribute combination is composed of multiple attribute fields corresponding to the multiple keywords one by one; For each of the multiple attribute combinations, performing a correlation analysis based on multiple attribute fields in the attribute combination to obtain a correlation result of the attribute combination; Based on the correlation results of the multiple attribute combinations, a target attribute combination is screened out from the multiple attribute combinations; Filtering out a plurality of target MA codes that meet the target attribute combination from the candidate MA codes of the plurality of keywords, wherein the plurality of target MA codes correspond to the plurality of keywords one by one; The MA codes of the plurality of keywords are determined based on the plurality of target MA codes.

9. The system according to claim 7, characterized in that The query request also includes the identifier of the user; Feeding back the answer to the original question to the user, including: Determining whether the answer contains the target corpus in ciphertext form; In response to determining that the answer does not contain the target corpus in ciphertext form, feeding back the answer in full to the user; In response to determining that the answer includes the target corpus in ciphertext form, determining whether the user identifier belongs to a set of authorized user identifiers; In response to determining that the user's identifier belongs to the authorized user identifier set, decrypting the target corpus in ciphertext form to obtain the target corpus in plaintext form; re-encrypting the target corpus in plaintext form using the user's public key to obtain a re-encrypted target corpus; replacing the target corpus in ciphertext form in the answer with the re-encrypted target corpus to obtain an updated answer; and feeding back the updated answer to the user; In response to determining that the user identifier does not belong to the authorized user identifier set, an answer excluding the target corpus in ciphertext form is fed back to the user.

10. A computing device, characterized in that: The method comprises a processor and a memory, wherein a computer program is stored in the memory. When the processor executes the computer program, the data query method based on the MA code and based on the object unified identification as claimed in any one of claims 1 to 6 is implemented.

Images