Content retrieval method and device, computer equipment, storage medium and program product

Abstract

The invention relates to a content retrieval method and device, computer equipment, a storage medium and a program product. The method comprises the steps of obtaining a keyword of a target problem; according to the keyword, performing retrieval processing in a pre-established knowledge base constructed by at least one knowledge data in a text, a table and a picture through an index nested structure to obtain target retrieval content; and determining target reply content according to the target retrieval content. By obtaining the keyword of the target question and performing retrieval in the knowledge base with the index nested structure, the target retrieval content highly related to the question can be quickly and accurately positioned. In this way, blind search in a large amount of unorganized data is avoided, and the efficiency and accuracy of question answering are greatly improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a content retrieval method, apparatus, computer equipment, storage medium and program product. Background Art

[0002] With the widespread application of big models in various industries, they face the problem of insufficient professional knowledge in professional scenarios or industry segments. The knowledge of general big models is limited to training data, which mainly comes from public data on the Internet. It is impossible to obtain real-time, non-public or offline data. In addition, data security is crucial for enterprises and institutions. Uploading private domain data to a third-party platform for training will bring the risk of data leakage, making it difficult to balance data security and effectiveness for application solutions that rely entirely on the capabilities of general big models.

[0003] Retrieval enhancement generation technology has become an effective solution to the above problems. However, when faced with different forms of content such as tables and pictures in the text, current technology makes it difficult to effectively perform targeted processing and indexing to achieve comprehensive knowledge association and accurate retrieval. Summary of the invention

[0004] Based on this, it is necessary to provide a content retrieval method, apparatus, computer equipment, storage medium and program product that can solve the above-mentioned technical problems.

[0005] In a first aspect, the present application provides a content retrieval method, the method comprising:

[0006] Get the keywords of the target question;

[0007] Performing search processing in a pre-established knowledge base according to keywords to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0008] Determine the target reply content according to the target search content.

[0009] In one embodiment, the above-mentioned obtaining target search content in a pre-established knowledge base based on keywords includes:

[0010] Performing a preliminary search process in the knowledge base according to the keywords to obtain preliminary search content; the preliminary search process includes at least one of a similarity search process and a full-text search process;

[0011] When the first index information exists in the preliminary search content, a deep search process is performed according to the first index information to obtain the target search content.

[0012] In one embodiment, the deep search process is performed according to the first index information to obtain the target search content, including:

[0013] In each round of iterative retrieval process of the deep retrieval processing, a retrieval process is performed based on the first index information and the index nested structure of the knowledge base to obtain a retrieval result;

[0014] When the search result indicates the existence of the second index information, the next depth of search processing is performed based on the second index information and the index nesting structure of the knowledge base until the search is completed and the target search content is obtained.

[0015] In one embodiment, the method further comprises:

[0016] In each round of iterative retrieval, the depth parameter is obtained;

[0017] When the number of iterations meets the depth parameter, the retrieval is stopped, and the target retrieval content is determined according to the content retrieved when the retrieval is stopped.

[0018] In one embodiment, the process of constructing the index nested structure of the knowledge base includes:

[0019] Perform data structure segmentation processing on the pre-acquired knowledge data to obtain the index root directory;

[0020] Perform content segmentation on knowledge data to obtain index key-value pairs;

[0021] Performing replacement processing on preset content in the knowledge data to obtain replacement content; the replacement processing includes at least one of text replacement, table replacement and image replacement;

[0022] Based on the index nesting algorithm, the index root directory, index key value and replacement content are nested to obtain the index nesting structure of the knowledge base.

[0023] In one embodiment, the step of determining the target reply content according to the target search content includes:

[0024] Perform language style adjustment on the target search content to obtain the intermediate response content;

[0025] The intermediate reply content is simplified to obtain the target reply content.

[0026] In a second aspect, the present application also provides a content retrieval device. The device includes:

[0027] Keyword acquisition module, used to obtain keywords of target questions;

[0028] A retrieval module is used to perform retrieval processing in a pre-established knowledge base according to keywords to obtain target retrieval content; the knowledge base is constructed by using at least one kind of knowledge data in text, table and picture through an index nested structure;

[0029] The content determination module is used to determine the target reply content according to the target search content.

[0030] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0031] Get the keywords of the target question;

[0032] Performing search processing in a pre-established knowledge base according to keywords to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0033] Determine the target reply content according to the target search content.

[0034] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0035] Get the keywords of the target question;

[0036] Performing search processing in a pre-established knowledge base according to keywords to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0037] Determine the target reply content according to the target search content.

[0038] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0039] Get the keywords of the target question;

[0040] Performing search processing in a pre-established knowledge base according to keywords to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0041] Determine the target reply content according to the target search content.

[0042] The above-mentioned content retrieval method, apparatus, computer equipment, storage medium and program product, the present application obtains the keywords of the target question; performs retrieval processing based on the keywords in a pre-established knowledge base constructed by an index nested structure of at least one knowledge data in text, table and picture to obtain the target retrieval content; determines the target reply content based on the target retrieval content. By obtaining the keywords of the target question and searching in a knowledge base with an index nested structure, the target retrieval content that is highly relevant to the question can be quickly and accurately located. This method avoids blindly searching in a large amount of unorganized data and greatly improves the efficiency and accuracy of problem answering. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is an application environment diagram of a content retrieval method in an embodiment;

[0044] Figure 2 is a flow chart of a content retrieval method in one embodiment;

[0045] Figure 3 A schematic diagram of a large model application technology route in an embodiment;

[0046] Figure 4 A schematic diagram of a process for obtaining target search content in one embodiment;

[0047] Figure 5 A schematic diagram of a process for obtaining target search content in another embodiment;

[0048] Figure 6 is a flowchart of a content retrieval method in another embodiment;

[0049] Figure 7 A schematic diagram of a process of constructing an index nested structure of a knowledge base in one embodiment;

[0050] Figure 8 A schematic diagram of creating an index root directory of different domain knowledge in one embodiment;

[0051] Fig. 9 A schematic diagram of an index root directory structure in one embodiment;

[0052] Fig.10 A schematic diagram of standard content in an embodiment;

[0053] Fig.11 A schematic diagram of content that needs to be replaced in its entirety in one embodiment;

[0054] Fig.12 A schematic diagram of contents requiring table replacement in one embodiment;

[0055] Fig.13A schematic diagram of content that requires image replacement in an embodiment;

[0056] Fig.14 A schematic diagram of replacing a picture in an embodiment;

[0057] Fig.15 1 is a schematic diagram of determining target reply content according to target search content in an embodiment;

[0058] Fig.16 is a structural block diagram of a content retrieval device in one embodiment;

[0059] Fig.17 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0061] The content retrieval method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The terminal 102 obtains the keywords of the target question; retrieves and processes the knowledge base constructed by the index nested structure of at least one knowledge data in the text, table and picture in advance according to the keywords, and obtains the target search content; determines the target reply content according to the target search content. Among them, the terminal 102 can be but is not limited to various personal computers, laptops, smart phones and tablets. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0062] In an exemplary embodiment, Figure 2 As shown, the present application provides a content retrieval method, which is applied to Figure 1 The terminal 102 in the example is used as an example to illustrate, and the following steps are included:

[0063] S201, obtaining keywords of the target question.

[0064] Among them, "keywords" are representative words extracted from the target question and used for accurate retrieval in the knowledge base to quickly locate knowledge content related to the target question.

[0065] In the embodiment of the present application, after the user inputs the target question in the terminal, the system first performs natural language processing on the target question. Using lexical analysis, syntactic analysis and other technologies, key nouns, verbs, adjectives and the like in the target question are extracted as keywords. For example, for the question "What equipment is available for low pressure testing?", keywords such as "low pressure testing" and "equipment" can be extracted.

[0066] Another implementation method of the present application embodiment: a machine learning algorithm is used to train a large number of labeled questions and establish a question classification model. When the user enters the target question, the model classifies it according to the characteristics of the question and extracts the common key elements in this type of question as keywords. For example, for the above question, if the model classifies it as "microelectronic device testing category", then the common "microelectronic device", "low pressure test", "process" and so on in this category are extracted as keywords.

[0067] S202, performing a search process in a pre-established knowledge base according to the keywords to obtain target search content.

[0068] Among them, the "knowledge base" is a database constructed by index nesting structure of knowledge data in various forms such as text, tables and pictures, which is used to store and retrieve relevant knowledge content. "Index nesting structure" is a way to organize knowledge data according to different levels and relationships, and achieve the purpose of fast retrieval and deep mining of knowledge data by creating index root directories, building index key-value pairs and performing index replacement.

[0069] In the embodiments of the present application, for textual knowledge data, in addition to collecting from traditional channels such as industry standard documents and professional books, the latest knowledge content can also be obtained from online technical forums, academic paper databases, etc. The collected text is preprocessed, including operations such as removing stop words and stem extraction to improve the accuracy of subsequent retrieval. For tabular knowledge data, not only key information such as the title, column name, and row name of the table can be extracted, but also the data trends and relationships in the table can be analyzed to add richer descriptive information to the table in order to better match it with keywords. For picture knowledge data, image recognition technology can be used to automatically identify key objects and scenes in the picture, and add corresponding labels and descriptive information to the picture.

[0070] In the retrieval process, in addition to matching directly in the knowledge base based on keywords, you can also use semantic similarity calculation methods to find knowledge content with similar semantics to the keywords. For example, for the keyword "low pressure test", not only can you retrieve knowledge data containing the keyword, but you can also retrieve content with similar semantics such as "low pressure test" and "testing under low pressure environment". At the same time, you can combine the user's historical search records and behavior patterns to personalize the search results and improve the user's search experience.

[0071] Another implementation method of the retrieval process is to use parallel retrieval technology to search on multiple index nodes at the same time to improve the retrieval speed. After the relevant content is retrieved, the caching mechanism is used to cache the content so that it can be retrieved faster the next time. If the knowledge data in the knowledge base is found to be updated or changed during the retrieval process, the index is updated in time to ensure the accuracy and timeliness of the retrieval results.

[0072] S203, determining target reply content according to the target search content.

[0073] Among them, "target search content" refers to the knowledge content related to the target question retrieved from the knowledge base based on keywords, which may include text, tables, pictures, etc. "Target response content" is the content that answers the user's question after analyzing, sorting and processing the target search content, aiming to provide users with accurate, clear and useful information.

[0074] In the embodiment of the present application, the retrieved target search content is comprehensively analyzed, and the text summary generation technology is used to extract key information and generate a concise and clear summary as the target reply content. For example, for the retrieved long text content about low pressure test, the summary generation technology is used to extract the main steps, key equipment and precautions of the test, and form a short summary reply to the user.

[0075] Another implementation method of the present application: Combined with knowledge graph technology, the knowledge elements in the target search content are constructed into a knowledge graph, and then reasoning and querying are performed in the knowledge graph according to the user's question to obtain more accurate and comprehensive response content. For example, for questions about low-pressure testing of microelectronic devices, the retrieved relevant knowledge is constructed into a knowledge graph, and by performing path search and reasoning in the knowledge graph, more complex questions about the impact of testing on different types of microelectronic devices can be answered.

[0076] Specifically, the embodiments of the present application can also be based on the large model application technology route of RAG (Retrieval-Augmented Generation, retrieval enhanced generation architecture), such as Figure 3 After the user asks a question, the large model rewrites the question, uses the rewritten content to search the knowledge base, sorts the searched content, outputs the TopN results, and rewrites and outputs them through the LLM (Large Language Model).

[0077] The above-mentioned content retrieval method, apparatus, computer equipment, storage medium and program product, the present application obtains the keywords of the target question; performs retrieval processing based on the keywords in a pre-established knowledge base constructed by an index nested structure of at least one knowledge data in text, table and picture to obtain the target retrieval content; determines the target reply content based on the target retrieval content. By obtaining the keywords of the target question and searching in a knowledge base with an index nested structure, the target retrieval content that is highly relevant to the question can be quickly and accurately located. This method avoids blindly searching in a large amount of unorganized data and greatly improves the efficiency and accuracy of problem answering.

[0078] In an exemplary embodiment, based on the above embodiment, see Figure 4 The embodiment of the present application relates to a process of obtaining target search content in a pre-established knowledge base based on keywords, including the following steps:

[0079] S301, performing preliminary search processing in the knowledge base according to the keywords to obtain preliminary search content.

[0080] Among them, "preliminary search processing" is the process of quickly finding knowledge content related to keywords in the knowledge base, including similarity search processing and full-text search processing. "Similarity search processing" calculates the vector similarity between keywords and knowledge content to find semantically similar results. "Full-text search processing" is a comprehensive string matching search of all data in the knowledge base.

[0081] In the embodiment of the present application, similarity retrieval processing is adopted, and the vector space model or deep learning text embedding technology is used to convert the keywords and the knowledge content in the knowledge base into vector representations, and the similarity between the vectors is calculated. For example, for the keyword "low pressure test equipment", after it is converted into a vector, the similarity is calculated with the vectorized text, table and picture description in the knowledge base, and the knowledge content with higher similarity is selected as the preliminary retrieval result.

[0082] Another implementation method of the present application embodiment is to perform full-text search processing and conduct a word-by-word matching search on all knowledge data in the knowledge base. For example, regular expressions or string matching algorithms are used to search for text, table cell content, or image description information containing keywords. At the same time, fuzzy matching technology can be combined to handle spelling errors or similar expressions of keywords to improve the recall rate of the search.

[0083] S302: When the preliminary search content contains the first index information, a deep search process is performed according to the first index information to obtain the target search content.

[0084] Among them, "first index information" refers to clues found in the preliminary search content that can guide in-depth search, such as specific index numbers, keyword references, etc. "In-depth search processing" is the process of further exploring deeper and more detailed knowledge content based on the preliminary search to obtain more accurate target search content.

[0085] In the embodiment of the present application, when there is index information in the preliminary search results, such as a specific standard number, method name or document reference, a deep search is performed in the knowledge base based on the index information. For example, if the preliminary search results mention "see the low pressure test method in standard XYZ", the standard document is located in the knowledge base according to the standard number "XYZ", and the detailed content about the low pressure test is deeply searched. The index nested structure can be used to quickly jump to the relevant knowledge block and extract more specific and in-depth information.

[0086] Another implementation method of the embodiment of the present application: using an iterative search method to perform a deep search. First, perform the first deep search based on the first index information in the preliminary search results, and then analyze the content obtained from the first deep search to see if there is other relevant index information. If so, continue to perform the next round of deep search based on these new index information until a specific stop condition is met, such as reaching a preset search depth or no new index information is available. For example, when searching for low-pressure test content for microelectronic devices, it is possible to start from a standard number index in a preliminary search result, and gradually go deeper into different chapters, tables, and pictures in the standard, and continuously dig out more detailed knowledge content.

[0087] The embodiment of the present application combines similarity retrieval processing and full-text retrieval processing to perform preliminary retrieval, which can quickly and comprehensively filter out knowledge content related to keywords from the knowledge base, thereby improving retrieval efficiency and accuracy. When the preliminary retrieval content contains the first index information, a deep retrieval process can be performed to further explore more detailed and in-depth knowledge to meet the user's needs for in-depth understanding of specific issues. The multi-stage retrieval method ensures that the retrieval results have both breadth and depth.

[0088] In an exemplary embodiment, based on the above embodiment, see Figure 5 The embodiment of the present application relates to a process of performing deep search processing according to first index information to obtain target search content, including the following steps:

[0089] S401, in each round of iterative retrieval process of the deep retrieval process, a retrieval process is performed based on the first index information and the index nested structure of the knowledge base to obtain a retrieval result.

[0090] Among them, the "index nested structure" is a structure that organizes knowledge data according to different levels and relationships. It achieves the purpose of rapid retrieval and deep mining of knowledge data by creating index root directories, building index key-value pairs, and performing index replacement.

[0091] In an embodiment of the present application, first, the type and scope of the first index information are determined. For example, if the first index information is a specific standard number, then the corresponding standard document part is located in the index nested structure of the knowledge base according to this number. Then, a depth-first search or breadth-first search algorithm is used to search in this part and its related sub-indexes. For textual knowledge data, relevant content is found through keyword matching and semantic analysis; for tabular knowledge data, check whether the table title, column name, row name, etc. are related to the first index information; for image knowledge data, analyze whether the image description information and labels match the first index information. Finally, all retrieved relevant knowledge content is integrated into the retrieval results.

[0092] Another implementation method of the embodiment of the present application: using a machine learning algorithm to extract and analyze features of the first index information. Based on the extracted features, select the part of the index nested structure of the knowledge base that is most likely to contain relevant content for retrieval. For example, if the first index information is related to a specific technical field, then use the trained classification model to determine the index parts of the knowledge base related to the technical field, and focus on searching these parts. At the same time, it can be combined with a collaborative filtering algorithm to predict possible related knowledge content based on the user's historical search behavior and similar search records of other users, and search the index parts where these contents are located.

[0093] S402: When the search result indicates the existence of the second index information, a search process of the next depth is performed based on the second index information and the index nesting structure of the knowledge base until the search is completed and the target search content is obtained.

[0094] Among them, the "second index information" is the new index clues found from the search results during the deep search process, which can guide deeper searches. The "target search content" is the knowledge content collection that is highly relevant to user needs that is finally obtained through the deep search process.

[0095] In an embodiment of the present application, when new index information (i.e., second index information) appears in the search results, the above steps are repeated, and a deeper search is performed in the index nested structure of the knowledge base, guided by the second index information. For example, if a link citing other standard documents or an index number pointing to a specific chapter or table is found in the search results, the search is continued based on this new index information. This process is repeated until no new index information appears or the preset search depth is reached. Finally, all retrieved content is integrated into the target search content.

[0096] Another implementation method of the embodiment of the present application: Use knowledge graph technology to analyze and reason about the second index information in the search results. Match the second index information with the nodes in the knowledge graph to find relevant knowledge paths and relationships. Then, perform targeted searches in the index nested structure of the knowledge base based on these relationships. For example, if the second index information is a specific technical concept, then look for other concepts and entities related to the concept in the knowledge graph, and retrieve the index part where these related contents are located in the knowledge base. In this way, deep retrieval can be performed more efficiently, improving the accuracy and completeness of the retrieval results.

[0097] The embodiment of the present application performs deep mining based on the index nested structure and the continuously emerging new index information through multiple rounds of iterative retrieval, and can comprehensively and accurately obtain relevant content in the knowledge base.

[0098] In an exemplary embodiment, based on the above embodiment, see Figure 6 , the method of the embodiment of the present application further includes the following steps:

[0099] S501, in each round of iterative retrieval process, obtain depth parameters.

[0100] Among them, the "depth parameter" is a parameter used to control the depth of iterative retrieval, which determines how many rounds of retrieval are performed during the iterative retrieval process.

[0101] In the embodiment of the present application, the depth parameter can be preset to a fixed value, for example, setting the depth parameter to 3 means performing three rounds of iterative retrieval. At the beginning of each round of retrieval, the system automatically obtains the preset depth parameter. Alternatively, the depth parameter can be dynamically adjusted according to the needs of the user. For example, when the user enters a retrieval request, the user can specify the desired retrieval depth, and the system determines the depth parameter according to the user's specification.

[0102] Another implementation of the embodiment of the present application: the depth parameter can be dynamically calculated based on the structure and content of the knowledge base. For example, a suitable depth parameter is calculated by analyzing the level of index nesting in the knowledge base and the knowledge density of each level. If the knowledge level of a certain field in the knowledge base is deep and the content is rich, the value of the depth parameter can be increased accordingly to ensure that the knowledge in this field can be fully mined.

[0103] S502, when the number of iterations meets the depth parameter, the search is stopped, and the target search content is determined according to the content retrieved when the search is stopped.

[0104] In each round of iterative search, the system records the current number of iterations. When the number of iterations reaches the value set by the depth parameter, the search is stopped immediately. Then, all search results obtained when the search is stopped are integrated and screened to remove duplicate and irrelevant content to obtain the target search content.

[0105] For example, if the depth parameter is 3, the system stops searching at the end of the third round of searching, and sorts out the contents retrieved in the first three rounds and determines them as target search contents.

[0106] Another implementation of the embodiment of the present application: During the iterative search process, the system monitors the quality and relevance of the search results in real time. If in a certain round of iterative search, it is found that the quality of the search results has met the preset standard, or the relevance no longer increases significantly with the increase in the number of iterations, then even if the number of iterations has not reached the value of the depth parameter, the search can be stopped in advance. Then, the target search content is determined based on the current search results.

[0107] In the embodiment of the present application, by introducing the depth parameter, the search depth can be effectively controlled while ensuring the search accuracy, avoiding excessive search and wasting resources. At the same time, the target search content is determined according to the content when the search is stopped, ensuring the pertinence and practicality of the search results.

[0108] In an exemplary embodiment, based on the above embodiment, see Figure 7 The index nested structure construction process of the knowledge base in the embodiment of the present application includes the following steps:

[0109] S601, performing data structure segmentation processing on the pre-acquired knowledge data to obtain an index root directory; performing content segmentation processing on the knowledge data to obtain index key-value pairs; performing replacement processing on the preset content in the knowledge data to obtain replacement content.

[0110] Among them, the "index root directory" is the top-level classification of the knowledge base index structure, which is used to preliminarily classify and locate knowledge data. The "index key-value pair" is a unique identifier assigned to each knowledge block, which is used to quickly retrieve and locate specific knowledge content. The "replacement processing" is to modify and optimize specific content in the knowledge data to improve the retrievability and accuracy of knowledge.

[0111] In the embodiment of the present application, for the data structure segmentation process, the overall structural characteristics of the knowledge data are first analyzed to determine possible classification bases, such as knowledge fields, knowledge types, etc. Then, the knowledge data is divided into different parts based on these classification bases, and each part is used as an index root directory. For example, for knowledge data in the field of electronic technology, it can be divided into several index root directories such as hardware knowledge, software knowledge, and communication knowledge.

[0112] For content segmentation, natural language processing technology is used to analyze knowledge data and segment text content into meaningful segments. Each segment is regarded as a knowledge block and a unique index key-value pair is assigned to it. For example, for an article about electronic circuit design, it can be segmented into different paragraphs or chapters. Each paragraph or chapter is regarded as a knowledge block and an index key-value pair consisting of numbers and letters is assigned to it.

[0113] For replacement processing, preset content is searched in the knowledge data, such as specific keywords, table titles, image descriptions, etc. Then, replacement is performed according to pre-set rules. For example, for text replacement, specific keywords can be replaced with more standardized or easier to search expressions; for table replacement, the ambiguous expressions in the table title can be replaced with more specific descriptions; for image replacement, more detailed description information can be added to the image and the original simple description can be replaced.

[0114] S602, based on the index nesting algorithm, perform nesting processing on the index root directory, index key value and replacement content to obtain the index nesting structure of the knowledge base.

[0115] Among them, the "index nesting algorithm" is an algorithm used to construct a nested structure of knowledge base indexes. By nesting indexes at different levels, efficient organization and retrieval of knowledge data can be achieved.

[0116] In the embodiment of the present application, first, the level and structure of index nesting are determined. Different levels of index nesting structures can be designed according to the complexity of knowledge data and retrieval requirements. For example, a three-layer nested structure can be designed, with the first layer being the index root directory, the second layer being the index key-value pair, and the third layer being the specific knowledge content.

[0117] Then, the index root directory, index key-value pairs and replacement content are nested according to the designed hierarchical structure using the index nesting algorithm. The specific algorithm can use a recursive algorithm or an iterative algorithm to gradually nest the index of each level into the index of the previous level to form a complete index nesting structure.

[0118] Finally, the constructed index nesting structure is optimized and adjusted. The integrity and accuracy of the index can be checked, duplicate indexes and invalid indexes can be removed, and the quality and efficiency of the index can be improved.

[0119] Specifically, for industry domain knowledge such as books, standards, and specifications, an index dictionary is constructed. The specific operations are as follows:

[0120] Create index root directories for different fields of knowledge. For example, in standard 548C-2016 "Microelectronic Device Test Methods and Procedures", there are different methods, such as Figure 8 shown.

[0121] The constructed index root directory structure is shown below. The index can quickly index specific methods and their contents through index numbers, method numbers, and method names. Fig. 9 shown.

[0122] Index key value construction: for Fig.10 The standard content in the structure index and the corresponding key value. The construction method is as follows:

[0123] {"knowledges":{

[0124] "1.1":"Test Methods and Procedures for Microelectronic Devices"#Purpose\n1.1 Purpose\nThis test is conducted to simulate the low pressure conditions encountered by aircraft or other aircraft during high-altitude flight. The purpose of this test is to determine the ability of devices and materials to avoid electrical breakdown failure due to the reduction in dielectric strength of air and other insulating materials when the air pressure is reduced. Under low pressure conditions, even if complete voltage breakdown does not occur, harmful effects such as corona discharge, dielectric loss and ionization will be enhanced. This test can also be used to examine other effects under low pressure conditions, such as the effect of thin air on the operating characteristics of the device, such as changes in the dielectric constant of the insulating material and reduced heat dissipation capacity of the heating device. ".

[0125] "1.2":"Test Methods and Procedures for Microelectronic Devices"#Equipment\n1.2 Equipment\nThe equipment required for low pressure test includes vacuum pump, suitable sealed chamber (if necessary, the sealed chamber should be equipped with a device for observing the sample), pressure gauge for measuring the simulated height of the sealed chamber, microammeter or oscilloscope for detecting current ranging from DC to 30MHz, and other necessary test instruments. ".}

[0126] This construction method constructs each part of the content into a knowledge block, and uses the method's label "1" as the first index number and ".1" as the secondary index number. When the part is mentioned in the text, such as "see method 1001 purpose", the content of the part can be associated with the index.

[0127] Text replacement: For text content such as Fig.11 Replace the entire content in the box.

[0128] Replace with: Use a microammeter or oscilloscope to monitor the device terminals with the maximum voltage applied (see 1.4c)) in the range from DC to 30MHz to see if corona discharge current appears.

[0129] Table replacement: for text content such as Fig.12 Replace the content in the box with the table and rewrite the table.

[0130] Content replacement method in the text: The sample should be installed in a sealed room as required, and the indoor air pressure should be reduced to the specified value in the following test conditions:

[0131] Test condition A was carried out at a pressure of 58 kPa, corresponding to an altitude of 4572 m.

[0132] The pressure of test condition B is 30 kPa and the altitude is 9144 m.

[0133] Test condition C was conducted at a pressure of 12 kPa and an altitude of 15,240 m.

[0134] The pressure of test condition D is 4.4 kPa and the altitude is 21,336 m.

[0135] Test condition E was carried out at a pressure of 1.1 kPa, corresponding to an altitude of 30,480 m.

[0136] The pressure of test condition F is 0.15 kPa and the altitude is 45,720 m.

[0137] Image replacement: For text content such as Fig.13 and Fig.14 The picture information existing in the present application embodiment will be rewritten. Content replacement method in the text: The sample should continuously complete the specified number of cycles under the specified conditions. Use test condition C for at least 10 cycles.

[0138] Rewrite the contents of the graph:

[0139] Start at Room Temperature: The test starts at room temperature.

[0140] Cooling phase: The temperature is lowered to -65°C to -75°C and maintained for at least 10 minutes.

[0141] Heating phase: The temperature is raised to 150°C to 165°C and maintained for at least 10 minutes.

[0142] Cycle process:

[0143] First cycle: decrease from room temperature to -65°C to -75°C, hold for at least 10 minutes, then increase to 150°C to 165°C, hold for at least 10 minutes.

[0144] Second cycle: Repeat the steps of the first cycle.

[0145] The figure shows that the sample temperature gradually increases and decreases during the cycle, and the time spent at each extreme temperature point (minimum and maximum temperatures) is at least 10 minutes.

[0146] The schematic diagrams of the first and second cycles clearly indicate the change process from low temperature to high temperature and the residence time at each stage.

[0147] The test can end at room temperature or enter the next cycle, depending on the test requirements.

[0148] During the processing, all the Figure X " is rewritten as "<Figure>The index number of this method. The number of the figure under this number".

[0149] This replacement can achieve the goal of introducing relevant knowledge of different depths through layer-by-layer indexing during the process of paragraph comprehension.

[0150] The knowledge base index nested structure constructed by the embodiment of the present application can effectively organize and manage knowledge data and improve the retrievability and accuracy of knowledge. Through data structure segmentation, content segmentation and replacement processing, and the application of index nesting algorithm, multi-dimensional classification and positioning of knowledge data can be achieved, making it convenient for users to quickly and accurately find the required knowledge content. At the same time, the structure has good scalability and adaptability, and can be dynamically adjusted and optimized as knowledge continues to grow and change.

[0151] In an exemplary embodiment, based on the above embodiment, see Fig.15 The embodiment of the present application relates to a process of determining target reply content according to target search content, including the following S701 to S702. Among them:

[0152] S701, performing language style adjustment processing on the target search content to obtain intermediate response content.

[0153] Among them, the "intermediate reply content" is the reply content after the language style adjustment, and the content has not yet been simplified.

[0154] In the embodiment of the present application, the language characteristics and domain attributes of the target search content are analyzed. If the target search content comes from literature in a professional field, the language may be more rigorous and professional. According to the needs and background of the user, determine the appropriate language style adjustment strategy. For example, if the user is an ordinary person, convert professional terms into easy-to-understand expressions, and adjust the sentence structure to make it more in line with daily language habits.

[0155] Using lexical analysis, syntactic analysis, and semantic analysis in natural language processing technology, the target search content is processed sentence by sentence. Professional terms, complex sentence structures, and uncommon words are identified, and corresponding replacements and adjustments are made. For example, "integrated circuit" is replaced with "chip", and the complex sentence structure of "using... to carry out..." is adjusted to the simple sentence structure of "through... to...".

[0156] S702, simplifying the intermediate reply content to obtain the target reply content.

[0157] In the embodiment of the present application, the structure and logical relationship of the intermediate reply content are analyzed. Long sentences and complex paragraphs are identified, key information is extracted, and redundant content is removed. For example, for a paragraph describing a complex technical process, key steps and main concepts are extracted, and unnecessary detailed descriptions are removed.

[0158] Use text summarization technology to automatically summarize the intermediate reply content. You can use the extractive summary generation method to select important sentences and paragraphs from the intermediate reply content to form a summary; you can also use the generative summary generation method to generate concise summary text based on the semantics of the intermediate reply content.

[0159] The embodiment of the present application can transform professional and complex knowledge into easy-to-understand, concise and clear reply content by adjusting the language style and simplifying the content of the target search content, thereby improving the spreadability and usability of the knowledge. At the same time, this processing method can meet the needs of different user groups and expand the application scope of the knowledge search system.

[0160] In an exemplary embodiment, based on the above embodiment, the method of the embodiment of the present application further includes the following steps:

[0161] Step 1: Perform data structure segmentation processing on the pre-acquired knowledge data to obtain the index root directory; perform content segmentation processing on the knowledge data to obtain index key-value pairs; perform replacement processing on the preset content in the knowledge data to obtain replacement content; the replacement processing includes at least one of text replacement, table replacement and image replacement; based on the index nesting algorithm, perform nesting processing on the index root directory, index key value and replacement content to obtain the index nesting structure of the knowledge base.

[0162] Step 2: Get the keywords of the target question;

[0163] Step 3: Perform a preliminary search in the knowledge base based on the keywords to obtain preliminary search content;

[0164] Step 4: Obtaining depth parameters; in each round of iterative retrieval process of the deep retrieval processing, performing retrieval processing based on the first index information and the index nested structure of the knowledge base to obtain retrieval results;

[0165] Step 5: When the number of iterations meets the depth parameter, the search is stopped, and the target search content is determined according to the content retrieved when the search is stopped;

[0166] Step 6: Adjust the language style of the target search content to obtain the intermediate response content; simplify the intermediate response content to obtain the target response content.

[0167] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0168] Based on the same inventive concept, the embodiment of the present application also provides a content retrieval device for implementing the content retrieval method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more content retrieval device embodiments provided below can refer to the limitations on the content retrieval method above, and will not be repeated here.

[0169] In one embodiment, Fig.16 As shown, a content retrieval device 800 is provided, comprising:

[0170] A keyword acquisition module 801 is used to acquire keywords of a target question;

[0171] The retrieval module 802 is used to perform retrieval processing in a pre-established knowledge base according to keywords to obtain target retrieval content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0172] The content determination module 803 is used to determine the target reply content according to the target search content.

[0173] In one embodiment, the retrieval module includes:

[0174] A preliminary content determination unit is used to perform preliminary search processing in the knowledge base according to the keywords to obtain preliminary search content; the preliminary search processing includes at least one of similarity search processing and full-text search processing;

[0175] The deep search unit is used to perform deep search processing according to the first index information to obtain target search content when the preliminary search content has the first index information.

[0176] In one embodiment, the depth search unit comprises:

[0177] A retrieval subunit, used to perform retrieval processing based on the first index information and the index nested structure of the knowledge base to obtain retrieval results in each round of iterative retrieval process of the deep retrieval processing;

[0178] The deep search subunit is used to perform the next deep search process based on the second index information and the index nesting structure of the knowledge base when the search result indicates the existence of the second index information, until the search is completed and the target search content is obtained.

[0179] In one embodiment, the above device further comprises:

[0180] The parameter acquisition module is used to obtain the depth parameters in each round of iterative retrieval process;

[0181] The content determination module is used to stop the retrieval when the number of iterations meets the depth parameter, and determine the target retrieval content according to the content retrieved when the retrieval is stopped.

[0182] In one embodiment, the above device comprises:

[0183] A structure segmentation module is used to perform data structure segmentation processing on the pre-acquired knowledge data to obtain an index root directory;

[0184] The content segmentation module is used to perform content segmentation processing on knowledge data to obtain index key-value pairs;

[0185] A content replacement module is used to replace the preset content in the knowledge data to obtain replacement content; the replacement process includes at least one of text replacement, table replacement and image replacement;

[0186] The nesting module is used to perform nesting processing on the index root directory, index key value and replacement content based on the index nesting algorithm to obtain the index nesting structure of the knowledge base.

[0187] In one embodiment, the content determination module includes:

[0188] A style adjustment unit, used to perform language style adjustment processing on the target search content to obtain intermediate response content;

[0189] The content simplification unit is used to simplify the intermediate reply content to obtain the target reply content.

[0190] Each module in the above content retrieval device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0191] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Fig.17 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a content retrieval method is realized. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0192] Those skilled in the art will understand that Fig.17The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0193] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0194] Get the keywords of the target question;

[0195] Performing search processing in a pre-established knowledge base according to keywords to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0196] Determine the target reply content according to the target search content.

[0197] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0198] Performing a preliminary search process in the knowledge base according to the keywords to obtain preliminary search content; the preliminary search process includes at least one of a similarity search process and a full-text search process;

[0199] When the first index information exists in the preliminary search content, a deep search process is performed according to the first index information to obtain the target search content.

[0200] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0201] In each round of iterative retrieval process of the deep retrieval processing, a retrieval process is performed based on the first index information and the index nested structure of the knowledge base to obtain a retrieval result;

[0202] When the search result indicates the existence of the second index information, the next depth of search processing is performed based on the second index information and the index nesting structure of the knowledge base until the search is completed and the target search content is obtained.

[0203] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0204] In each round of iterative retrieval, the depth parameter is obtained;

[0205] When the number of iterations meets the depth parameter, the retrieval is stopped, and the target retrieval content is determined according to the content retrieved when the retrieval is stopped.

[0206] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0207] Perform data structure segmentation processing on the pre-acquired knowledge data to obtain the index root directory;

[0208] Perform content segmentation on knowledge data to obtain index key-value pairs;

[0209] Performing replacement processing on preset content in the knowledge data to obtain replacement content; the replacement processing includes at least one of text replacement, table replacement and image replacement;

[0210] Based on the index nesting algorithm, the index root directory, index key value and replacement content are nested to obtain the index nesting structure of the knowledge base.

[0211] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0212] Perform language style adjustment on the target search content to obtain the intermediate response content;

[0213] The intermediate reply content is simplified to obtain the target reply content.

[0214] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0215] Get the keywords of the target question;

[0216] Performing search processing in a pre-established knowledge base according to keywords to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0217] Determine the target reply content according to the target search content.

[0218] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0219] Performing a preliminary search process in the knowledge base according to the keywords to obtain preliminary search content; the preliminary search process includes at least one of a similarity search process and a full-text search process;

[0220] When the first index information exists in the preliminary search content, a deep search process is performed according to the first index information to obtain the target search content.

[0221] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0222] In each round of iterative retrieval process of the deep retrieval processing, a retrieval process is performed based on the first index information and the index nested structure of the knowledge base to obtain a retrieval result;

[0223] When the search result indicates the existence of the second index information, the next depth of search processing is performed based on the second index information and the index nesting structure of the knowledge base until the search is completed and the target search content is obtained.

[0224] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0225] In each round of iterative retrieval, the depth parameter is obtained;

[0226] When the number of iterations meets the depth parameter, the retrieval is stopped, and the target retrieval content is determined according to the content retrieved when the retrieval is stopped.

[0227] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0228] Perform data structure segmentation processing on the pre-acquired knowledge data to obtain the index root directory;

[0229] Perform content segmentation on knowledge data to obtain index key-value pairs;

[0230] Performing replacement processing on preset content in the knowledge data to obtain replacement content; the replacement processing includes at least one of text replacement, table replacement and image replacement;

[0231] Based on the index nesting algorithm, the index root directory, index key value and replacement content are nested to obtain the index nesting structure of the knowledge base.

[0232] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0233] Perform language style adjustment on the target search content to obtain the intermediate response content;

[0234] The intermediate reply content is simplified to obtain the target reply content.

[0235] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0236] Get the keywords of the target question;

[0237] Performing search processing in a pre-established knowledge base according to keywords to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data in text, table and picture through an index nested structure;

[0238] Determine the target reply content according to the target search content.

[0239] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0240] Performing a preliminary search process in the knowledge base according to the keywords to obtain preliminary search content; the preliminary search process includes at least one of a similarity search process and a full-text search process;

[0241] When the first index information exists in the preliminary search content, a deep search process is performed according to the first index information to obtain the target search content.

[0242] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0243] In each round of iterative retrieval process of the deep retrieval processing, a retrieval process is performed based on the first index information and the index nested structure of the knowledge base to obtain a retrieval result;

[0244] When the search result indicates the existence of the second index information, the next depth of search processing is performed based on the second index information and the index nesting structure of the knowledge base until the search is completed and the target search content is obtained.

[0245] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0246] In each round of iterative retrieval, the depth parameter is obtained;

[0247] When the number of iterations meets the depth parameter, the retrieval is stopped, and the target retrieval content is determined according to the content retrieved when the retrieval is stopped.

[0248] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0249] Perform data structure segmentation processing on the pre-acquired knowledge data to obtain the index root directory;

[0250] Perform content segmentation on knowledge data to obtain index key-value pairs;

[0251] Performing replacement processing on preset content in the knowledge data to obtain replacement content; the replacement processing includes at least one of text replacement, table replacement and image replacement;

[0252] Based on the index nesting algorithm, the index root directory, index key value and replacement content are nested to obtain the index nesting structure of the knowledge base.

[0253] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0254] Perform language style adjustment on the target search content to obtain the intermediate response content;

[0255] The intermediate reply content is simplified to obtain the target reply content.

[0256] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0257] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0258] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0259] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A content retrieval method, characterized in that: The method comprises: Get the keywords of the target question; Performing a search process in a pre-established knowledge base according to the keyword to obtain target search content; the knowledge base is constructed by at least one kind of knowledge data among text, table and picture through an index nested structure; The target reply content is determined according to the target search content.

2. The method according to claim 1, characterized in that: The step of obtaining target search content in a pre-established knowledge base according to the keyword includes: Performing a preliminary search process in the knowledge base according to the keywords to obtain preliminary search content; the preliminary search process includes at least one of a similarity search process and a full-text search process; In the case that the preliminary search content has the first index information, a deep search process is performed according to the first index information to obtain the target search content.

3. The method according to claim 2, characterized in that The performing deep search processing according to the first index information to obtain the target search content includes: In each round of iterative retrieval process of the deep retrieval process, a retrieval process is performed based on the first index information and the index nested structure of the knowledge base to obtain a retrieval result; In the case where the search result indicates the existence of the second index information, a search process of the next depth is performed based on the second index information and the index nesting structure of the knowledge base until the search is completed and the target search content is obtained.

4. The method according to claim 3, characterized in that The method further comprises: In each round of iterative retrieval, the depth parameter is obtained; When the number of iterations meets the depth parameter, the search is stopped, and the target search content is determined according to the content retrieved when the search is stopped.

5. The method according to claim 3, characterized in that: The index nested structure construction process of the knowledge base includes: Perform data structure segmentation processing on the pre-acquired knowledge data to obtain the index root directory; Performing content segmentation processing on the knowledge data to obtain index key-value pairs; Performing a replacement process on the preset content in the knowledge data to obtain replacement content; the replacement process includes at least one of text replacement, table replacement and image replacement; Based on an index nesting algorithm, the index root directory, the index key value and the replacement content are nested to obtain an index nesting structure of the knowledge base.

6. The method according to claim 1, characterized in that The step of determining target reply content according to the target search content includes: Performing language style adjustment processing on the target search content to obtain intermediate response content; The intermediate reply content is simplified to obtain the target reply content.

7. A content retrieval device, characterized in that: The device comprises: Keyword acquisition module, used to obtain keywords of target questions; A retrieval module, used to perform retrieval processing in a pre-established knowledge base according to the keyword to obtain target retrieval content; the knowledge base is constructed by at least one kind of knowledge data among text, table and picture through an index nested structure; The content determination module is used to determine the target reply content according to the target search content.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

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

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images