Intelligent question-answering system for oil and gas storage and transportation industry

Abstract

The invention provides an intelligent question-answering system for the oil and gas storage and transportation industry. The intelligent question-answering system comprises a question-answering environment module, a question-answering mode module, a question-answering method module and a question-answering analysis module. Wherein the question-answering environment module is used for receiving a question-answering instruction input by a user and storing question-answering data. The question-answering mode module is used for determining a target question-answering mode in the question-answering modes based on the question-answering instruction. The question-answering method module is used for determining a reply paragraph associated with the question-answering instruction based on the target question-answering mode. The question-answer analysis module is used for generating reply statements corresponding to the question-answer instructions based on the reply paragraphs The intelligent question-answering system for the oil and gas storage and transportation industry provided by the invention can quickly and accurately provide reply statements including data resources required by a user to the user based on the question-answering instruction input by the user.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent question-answering system for the oil and gas storage and transportation industry. Background Art

[0002] At present, although the oil and gas storage and transportation industry has accumulated rich data resources, it faces problems such as many types of data, complex structures, loose connections with industry knowledge, and few business scenario correlations. Most people still find it difficult to obtain the data resources needed in the oil and gas storage and transportation industry.

[0003] Therefore, how to build an intelligent question-answering system for the oil and gas storage and transportation industry so that users can quickly and accurately obtain the data resources needed in the oil and gas storage and transportation industry through interaction with the intelligent question-answering system is an urgent problem to be solved. Summary of the invention

[0004] The present invention provides an intelligent question-answering system for the oil and gas storage and transportation industry, which can quickly and accurately provide users with reply statements including data resources required by users based on question-answering instructions input by users.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] The present invention provides an intelligent question-answering system for the oil and gas storage and transportation industry, including a question-answering environment module, a question-answering mode module, a question-answering method module, and a question-answering parsing module. The question-answering environment module is used to receive question-answering instructions input by a user and store question-answering data. The question-answering mode module is used to determine a target question-answering mode in each question-answering mode based on the question-answering instruction. The question-answering method module is used to determine a reply paragraph associated with the question-answering instruction based on the target question-answering mode. The question-answering parsing module is used to generate a reply sentence corresponding to the question-answering instruction based on the reply paragraph.

[0007] Among them, the task question-and-answer mode includes determining the reply paragraph in the question-and-answer data based on the similarity between the question-and-answer instruction and each preset question-and-answer instruction. The knowledge question-and-answer mode includes determining the reply paragraph in the question-and-answer data based on the semantics of the question-and-answer instruction. The retrieval question-and-answer mode includes determining the reply paragraph in the question-and-answer data based on the text corresponding to the question-and-answer instruction. The chat question-and-answer mode includes determining the reply paragraph based on the question-and-answer instruction, the preset sequence model and the self-attention mechanism.

[0008] Based on the above technical solution, the present invention can also be improved as follows.

[0009] Further, the question-answer mode module includes a first switching unit, and the first switching unit is used to:

[0010] Obtain the similarity between the question-and-answer instruction and each preset question-and-answer instruction. Based on the similarity between the question-and-answer instruction and the target preset question-and-answer instruction being greater than a first threshold, determine the task question-and-answer mode as the target question-and-answer mode. Wherein, the target preset question-and-answer instruction is the preset question-and-answer instruction with the greatest similarity to the question-and-answer instruction. Based on the similarity between the question-and-answer instruction and the target preset question-and-answer instruction being less than or equal to a first threshold, determine the knowledge question-and-answer mode as the target question-and-answer mode.

[0011] Furthermore, the question-and-answer method module also includes a task question-and-answer unit, which is used to determine the paragraph in the question-and-answer data corresponding to the target preset question-and-answer instruction as the reply paragraph when the target question-and-answer mode is the task question-and-answer mode.

[0012] Furthermore, the question-answering method module further includes a knowledge question-answering unit, which is used to:

[0013] When the target question-answering mode is the knowledge question-answering mode, the semantics of the question-answering instruction is parsed. Based on the semantics of the question-answering instruction, a query statement is constructed. When the question-answering data includes a paragraph corresponding to the query statement, the paragraph corresponding to the query statement is determined as the answer paragraph.

[0014] Furthermore, the question-answer mode module further includes a second switching unit, which is used to:

[0015] In the case where the question-and-answer data does not include a paragraph corresponding to the query statement, the text corresponding to the question-and-answer instruction is determined. The similarity between the text corresponding to the question-and-answer instruction and each paragraph included in the question-and-answer data is obtained. Based on the similarity between the text corresponding to the question-and-answer instruction and the target paragraph being greater than a second threshold, the retrieval question-and-answer mode is determined as the target question-and-answer mode. The target paragraph is the paragraph in the question-and-answer data that has the greatest similarity to the text corresponding to the question-and-answer instruction.

[0016] Furthermore, the question-answering method module also includes a retrieval question-answering unit, which is used to determine the target paragraph as a reply paragraph when the target question-answering mode is a retrieval question-answering mode.

[0017] Furthermore, the question-and-answer mode module also includes a third switching unit, which is used to determine the chat question-and-answer mode as the target question-and-answer mode based on the similarity between the text corresponding to the question-and-answer instruction and the target paragraph being less than or equal to a second threshold.

[0018] Furthermore, the question-answering method module further includes a chat question-answering unit, which is used to:

[0019] When the target question-answering mode is the small talk question-answering mode, the question-answering instruction is input into the preset sequence model, and the reply paragraph is determined based on the output of the preset sequence model and the self-attention mechanism. The preset sequence model is used to output at least one preset sequence sentence corresponding to the content based on the input content. The self-attention mechanism is used to determine the target sequence sentence in at least one preset sequence sentence based on a preset rule, and determine the target sequence sentence as the reply paragraph.

[0020] Furthermore, the preset rule includes determining the target sequence sentence based on the weight of each preset sequence sentence relative to the content output by the preset sequence model. Wherein, for any preset sequence sentence in the at least one preset sequence sentence, the weight of any preset sequence sentence relative to the content output by the preset sequence model is determined based on the similarity between any preset sequence sentence and the content.

[0021] Furthermore, the question-answering environment module includes a question-answering configuration unit and a question-answering data unit, wherein the question-answering configuration unit is used to receive a question-answering instruction input by a user, and the question-answering data unit is used to store question-answering data.

[0022] The beneficial effects of the present invention include:

[0023] Technical effect: By integrating advanced natural language processing and machine learning technologies, the present invention achieves accurate understanding and response to professional terms and complex queries in the oil and gas storage and transportation industry, thereby improving the accuracy and efficiency of information retrieval.

[0024] Application effect: In the daily operation of oil and gas storage and transportation, the present invention can provide real-time decision support, such as predicting equipment maintenance needs by analyzing historical data, reducing unexpected downtime, and improving the operating efficiency of storage and transportation facilities.

[0025] User benefits: Through interaction with the system provided by the present invention, users can quickly obtain professional operation guidance and troubleshooting suggestions, which reduces dependence on professionals and improves the convenience and autonomy of operation.

[0026] Economic benefits: According to simulation tests, the application of the system provided by the present invention is expected to improve the operating efficiency by at least 15%, which has significant economic value.

[0027] Data support: In actual application tests, the system provided by the present invention performs better than the existing technology in processing complex queries and providing decision support, with an accuracy rate of more than 95% and a response time shortened by at least 30%.

[0028] In summary, the present invention is not only technically innovative, but also demonstrates significant application value and social benefits in practical applications in the oil and gas storage and transportation industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A schematic diagram of the structure of the intelligent question-answering system for the oil and gas storage and transportation industry provided by the present invention;

[0030] Figure 2 A schematic diagram of a first process flow for processing data streams provided by the present invention;

[0031] Figure 3 A schematic diagram of a second process flow for processing data streams provided by the present invention;

[0032] Figure 4 A schematic diagram of a third process flow for processing data streams provided by the present invention;

[0033] Figure 5 A schematic diagram of a fourth process flow for processing data streams provided by the present invention;

[0034] Figure 6 A schematic diagram of the workflow of the intelligent question-answering system for the oil and gas storage and transportation industry provided by the present invention. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. Among them, in the description of the present application, unless otherwise specified, " / " indicates that the objects associated before and after are in an "or" relationship, for example, A / B can represent A or B; "and / or" in the present application is only a description of the association relationship of the associated objects, indicating that there can be three relationships, for example, A and / or B, which can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. And, in the description of the present application, unless otherwise specified, "multiple" refers to two or more than two. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple. In addition, in order to facilitate the clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, the words "first", "second" and the like are used to distinguish between the same items or similar items with substantially the same functions and effects. Those skilled in the art will understand that the words "first", "second" and the like do not limit the quantity and execution order, and the words "first", "second" and the like do not necessarily limit the differences. At the same time, in the embodiments of the present application, the words "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or design.

[0036] In the information and intelligent construction under the new model, how to connect the centralized data and knowledge center with the application and use of grassroots industry staff, so that the multi-faceted information data accumulated in the platform and the knowledge and wisdom of the vast number of experts can empower grassroots users and be applied and verified in different business scenarios, is a pain point that needs to be solved in the industry.

[0037] In this context, considering that the current domestic intelligent assistants are rarely used in other industries, and most of them appear in the form of customer service and knowledge question and answer, there is a lack of application of industry knowledge, and there are still major deficiencies in intelligent identification and reasoning and answering, the present invention provides an intelligent question and answer system for the oil and gas storage and transportation industry, which can construct a knowledge graph by mining and analyzing the five elements of "people", "things", "objects", "places" and "organizations", and provide answers to domain knowledge-based questions by combining natural language understanding and knowledge reasoning, and provide retrieval questions and answers as well as chat questions and answers through industry open knowledge.

[0038] See also Figure 1 The intelligent question-answering system for the oil and gas storage and transportation industry provided by the present invention includes a question-answering environment module, a question-answering mode module, a question-answering method module and a question-answering analysis module.

[0039] The question-answering environment module is used to receive the question-answering instructions input by the user and store the question-answering data. The question-answering mode module is used to determine the target question-answering mode in each question-answering mode based on the question-answering instructions. The question-answering method module is used to determine the reply paragraph associated with the question-answering instruction based on the target question-answering mode. The question-answering parsing module is used to generate a reply sentence corresponding to the question-answering instruction based on the reply paragraph.

[0040] Among them, the task question-and-answer mode includes determining the reply paragraph in the question-and-answer data based on the similarity between the question-and-answer instruction and each preset question-and-answer instruction. The knowledge question-and-answer mode includes determining the reply paragraph in the question-and-answer data based on the semantics of the question-and-answer instruction. The retrieval question-and-answer mode includes determining the reply paragraph in the question-and-answer data based on the text corresponding to the question-and-answer instruction. The chat question-and-answer mode includes determining the reply paragraph based on the question-and-answer instruction, the preset sequence model and the self-attention mechanism.

[0041] In some embodiments, see Figure 1 The question-and-answer mode module includes a first switching unit, which is used to obtain the similarity between the question-and-answer instruction and each preset question-and-answer instruction. Based on the similarity between the question-and-answer instruction and the target preset question-and-answer instruction being greater than a first threshold, the task question-and-answer mode is determined as the target question-and-answer mode. Among them, the target preset question-and-answer instruction is the preset question-and-answer instruction with the greatest similarity to the question-and-answer instruction. Based on the similarity between the question-and-answer instruction and the target preset question-and-answer instruction being less than or equal to the first threshold, the knowledge question-and-answer mode is determined as the target question-and-answer mode.

[0042] It should be noted that the first switching unit can obtain the similarity between the edit distance, the short text vector cosine similarity and each preset question and answer instruction based on the edit distance, the short text vector cosine, etc., which is not limited in the embodiment of the present application.

[0043] In some embodiments, see Figure 1 , the question-answering method module also includes a task question-answering unit. Figure 2 The task question-and-answer unit is used to determine the paragraph corresponding to the target preset question-and-answer instruction in the question-and-answer data as the reply paragraph when the target question-and-answer mode is the task question-and-answer mode.

[0044] In some embodiments, see Figure 1 The question-answering method module also includes a knowledge question-answering unit. Figure 3 The knowledge question-answering unit is used to parse the semantics of the question-answering instruction when the target question-answering mode is the knowledge question-answering mode. Based on the semantics of the question-answering instruction, a query statement is constructed. When the question-answering data includes a paragraph corresponding to the query statement, the paragraph corresponding to the query statement is determined as a reply paragraph.

[0045] The knowledge question-and-answer unit can integrate questions and answers of various types of questions, including factual questions, explanatory questions, relevance questions, comparative questions, etc., to meet the needs of users for diverse questions and answers. The knowledge question-and-answer unit gathers industry knowledge data, combines the characteristics of industry elements, analyzes the relationship between people, things, objects, and organizations, expresses them in the form of triples of entities, concepts, and relationships, constructs relational data, and forms a general industry knowledge graph.

[0046] Specifically, the knowledge question and answer unit can use semantic parsing to perform grammatical analysis on the question and answer instructions, convert the query into a logical expression, and obtain a query statement. For example, a natural language processing tool is used to segment the question and answer instructions, generate candidate entities, and label the element categories corresponding to the nodes. Based on syntactic dependency relationships, such as subject-predicate-object, verb-object structure, and main-subordinate clause structure, the query statement is constructed. After the knowledge question and answer unit obtains the query statement, it can use the query graph to generate multiple SPARQLs, and use part of the information in the data set to filter and optimize the multiple SPARQLs, including the elimination of ambiguity. Afterwards, the knowledge question and answer unit can execute SPARQL in the question and answer data to query and return the results.

[0047] In some embodiments, see Figure 1The question-and-answer mode module also includes a second switching unit, which is used to determine the text corresponding to the question-and-answer instruction when the question-and-answer data does not include a paragraph corresponding to the query statement. The similarity between the text corresponding to the question-and-answer instruction and each paragraph included in the question-and-answer data is obtained. Based on the similarity between the text corresponding to the question-and-answer instruction and the target paragraph being greater than a second threshold, the retrieval question-and-answer mode is determined as the target question-and-answer mode. The target paragraph is the paragraph in the question-and-answer data that has the greatest similarity to the text corresponding to the question-and-answer instruction.

[0048] Specifically, the second switching unit can find similar questions in the question-and-answer data based on the question-and-answer instruction and return the corresponding answers, using the inverted index + TF IDF + cosine similarity method to complete. The question-and-answer data is aggregated to build a document library based on the existing open data in the oil and gas storage and transportation industry, such as work systems, work processes, open emails, meeting reports, etc., and is stored in an indexed form to provide external retrieval services.

[0049] In some embodiments, see Figure 1 The question-answering method module also includes a retrieval question-answering unit. Figure 4 The retrieval question and answer unit is used to determine the target paragraph as the answer paragraph when the target question and answer mode is the retrieval question and answer mode.

[0050] Exemplarily, the second switching unit calculates the similarity between the question and answer quality and the content in the question and answer data, uses the similarity calculated by the matching model as a candidate reply, and outputs the matching score as a feature. The retrieval question and answer unit uses the ranking model to sort the candidate replies and returns the answer corresponding to the content with higher similarity to the user.

[0051] In some embodiments, see Figure 1 The question-and-answer mode module also includes a third switching unit, which is used to determine the chat question-and-answer mode as the target question-and-answer mode based on the similarity between the text corresponding to the question-and-answer instruction and the target paragraph being less than or equal to a second threshold.

[0052] In some embodiments, see Figure 1 , the question-answering method module also includes a chat question-answering unit. Figure 5 The chat question-and-answer unit is used to input the question-and-answer instruction into the preset sequence model when the target question-and-answer mode is the chat question-and-answer mode, and determine the reply paragraph based on the output of the preset sequence model and the self-attention mechanism. The preset sequence model is used to output at least one preset sequence sentence corresponding to the content based on the input content. The self-attention mechanism is used to determine the target sequence sentence in at least one preset sequence sentence based on a preset rule, and determine the target sequence sentence as the reply paragraph.

[0053] Among them, the chat question and answer mode provides open-domain chat questions and answers to meet any topic, and uses a preset sequence model + self-attention mechanism generation method to automatically generate answers for replies. Chat questions and answers use the document library as an algorithm training data set for model training and prediction.

[0054] In some embodiments, the preset rule includes determining the target sequence sentence based on the weight of each preset sequence sentence relative to the content output by the preset sequence model. Wherein, for any preset sequence sentence in the at least one preset sequence sentence, the weight of any preset sequence sentence relative to the content output by the preset sequence model is determined based on the similarity between any preset sequence sentence and the content.

[0055] In some embodiments, see Figure 1 The question-answering environment module includes a question-answering configuration unit and a question-answering data unit. The question-answering configuration unit is used to receive the question-answering instructions input by the user. The question-answering data unit includes a knowledge base, a document base, and a knowledge model. The knowledge base, the document base, and the knowledge model can all be used to provide question-answering data.

[0056] That is, see Figure 6 The present invention can determine whether to enter the task question and answer mode after receiving the question and answer instruction input by the user; if so, the question and answer instruction is processed based on the task question and answer mode. If not, it continues to determine whether to enter the knowledge question and answer mode; if so, the question and answer instruction is processed based on the knowledge question and answer mode. If not, it continues to determine whether to enter the search question and answer mode, and if so, the question and answer instruction is processed based on the search question and answer mode. If not, the question and answer instruction is processed based on the chat question and answer mode.

[0057] Based on the above content, it can be known that in the present invention, when satisfying a specific scenario, the user has a clear purpose and hopes to obtain specific satisfaction. The system mainly tracks the state of multiple rounds of dialogue and continuously provides feedback and corrections to understand the user's intention, thereby completing the target task or activity. In the data preparation stage, business analysis and data modeling are performed on specific application scenarios, and the model analysis results are used as answers to task questions. In the question and answer configuration stage, according to the key elements of the application scenario model and the context activity nodes, a dialogue process is constructed to configure questions, slot annotations (slots, entities), and replies (access to model results in the form of interfaces).

[0058] It can be seen that the present application can at least achieve the following technical effects:

[0059] Through the dialogue process design provided by the present invention, hierarchical progressive matching is achieved, and the accuracy of question responses is high.

[0060] The present invention provides a question-answering integration method with rich question-answering types, supporting task question-answering, knowledge question-answering, retrieval question-answering, and chat question-answering.

[0061] The present invention integrates a variety of industry knowledge, covering a wide range of knowledge including industry general knowledge, industry open knowledge, and industry scenario knowledge.

[0062] In some schemes, multiple embodiments of the present application can be combined, and the combined scheme can be implemented. Optionally, some operations in the process of each method embodiment are optionally combined, and / or the order of some operations is optionally changed. In addition, the execution order between the steps of each process is only exemplary and does not constitute a restriction on the execution order between the steps. There can also be other execution orders between the steps. It is not intended to indicate that the execution order is the only order in which these operations can be performed. A person of ordinary skill in the art will think of a variety of ways to reorder the operations described herein. In addition, it should be noted that the process details involved in a certain embodiment of this article are also applicable to other embodiments in a similar manner, or different embodiments can be used in combination.

[0063] In addition, some steps in the method embodiment may be equivalently replaced by other possible steps. Alternatively, some steps in the method embodiment may be optional and may be deleted in certain usage scenarios. Alternatively, other possible steps may be added to the method embodiment. Furthermore, the method embodiments may be implemented separately or in combination.

[0064] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above.

[0065] In the several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0066] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0067] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0068] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application shall be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An intelligent question-answering system for the oil and gas storage and transportation industry, characterized in that: include: A question-answering environment module, used for receiving question-answering instructions input by a user, and for storing question-answering data; The question and answer data is data related to oil and gas storage and transportation; A question-and-answer mode module, used to determine a target question-and-answer mode among various question-and-answer modes based on the question-and-answer instruction; the question-and-answer modes include a task question-and-answer mode, a knowledge question-and-answer mode, a search question-and-answer mode, and a chat question-and-answer mode; A question-answering method module, configured to determine a reply paragraph associated with the question-answering instruction based on the target question-answering mode; A question-and-answer parsing module, used to generate a reply sentence corresponding to the question-and-answer instruction based on the reply paragraph; Among them, the task question and answer mode includes determining the reply paragraph in the question and answer data based on the similarity between the question and answer instructions and each preset question and answer instruction; the knowledge question and answer mode includes determining the reply paragraph in the question and answer data based on the semantics of the question and answer instruction; the retrieval question and answer mode includes determining the reply paragraph in the question and answer data based on the text corresponding to the question and answer instruction; the chat question and answer mode includes determining the reply paragraph based on the question and answer instruction, a preset sequence model and a self-attention mechanism.

2. The system according to claim 1, characterized in that The question-answer mode module includes a first switching unit, and the first switching unit is used to: Obtaining the similarity between the question-and-answer instruction and each of the preset question-and-answer instructions; Based on the similarity between the question-and-answer instruction and the target preset question-and-answer instruction being greater than a first threshold, determining the task question-and-answer mode as the target question-and-answer mode; The target preset question-and-answer instruction is the preset question-and-answer instruction having the greatest similarity to the question-and-answer instruction; Based on the fact that the similarity between the question and answer instruction and the target preset question and answer instruction is less than or equal to the first threshold, the knowledge question and answer mode is determined as the target question and answer mode.

3. The system according to claim 2, characterized in that The question-answering method module further includes a task question-answering unit, which is used to: In a case where the target question-and-answer mode is the task question-and-answer mode, the paragraph in the question-and-answer data corresponding to the target preset question-and-answer instruction is determined as the reply paragraph.

4. The system according to claim 3, characterized in that The question-answering method module further includes a knowledge question-answering unit, which is used to: When the target question-and-answer mode is the knowledge question-and-answer mode, parsing the semantics of the question-and-answer instruction; Constructing a query statement based on the semantics of the question-answer instruction; In a case where the question and answer data includes a paragraph corresponding to the query sentence, the paragraph corresponding to the query sentence is determined as the answer paragraph.

5. The system according to claim 4, characterized in that The question-answer mode module further includes a second switching unit, wherein the second switching unit is configured to: In a case where the question and answer data does not include a paragraph corresponding to the query statement, determining a text corresponding to the question and answer instruction; Obtaining similarities between the text corresponding to the question-and-answer instruction and each paragraph included in the question-and-answer data; Based on the fact that the similarity between the text corresponding to the question and answer instruction and the target paragraph is greater than a second threshold, the retrieval question and answer mode is determined as the target question and answer mode; the target paragraph is the paragraph in the question and answer data that has the greatest similarity with the text corresponding to the question and answer instruction.

6. The system according to claim 5, characterized in that The question-answering method module further includes a retrieval question-answering unit, which is used to: When the target question-and-answer mode is the retrieval question-and-answer mode, the target paragraph is determined as the reply paragraph.

7. The system according to claim 6, characterized in that The question-answer mode module further includes a third switching unit, and the third switching unit is used to: Based on the similarity between the text corresponding to the question and answer instruction and the target paragraph being less than or equal to a second threshold, the casual chat question and answer mode is determined as the target question and answer mode.

8. The system according to claim 7, characterized in that The question-answering method module further includes a chat question-answering unit, which is used to: In the case where the target question-answering mode is the small talk question-answering mode, inputting the question-answering instruction into the preset sequence model, and determining the reply paragraph based on the output of the preset sequence model and the self-attention mechanism; The preset sequence model is used to output at least one preset sequence sentence corresponding to the content based on the input content; the self-attention mechanism is used to determine the target sequence sentence in the at least one preset sequence sentence based on preset rules, and determine the target sequence sentence as the reply paragraph.

9. The system according to claim 8, characterized in that The preset rules include: The target sequence sentence is determined based on the weight of each of the preset sequence sentences relative to the content; wherein, for any preset sequence sentence among the at least one preset sequence sentence, the weight of any preset sequence sentence relative to the content is determined based on the similarity between any preset sequence sentence and the content.

10. The system according to claim 9, characterized in that The question-answering environment module includes a question-answering configuration unit and a question-answering data unit. The question-answering configuration unit is used to receive question-answering instructions input by a user; and the question-answering data unit is used to store the question-answering data.

Images