Intelligent question and answer method and system

Abstract

The invention belongs to the technical field of natural language processing, and provides an intelligent question answering method and system. The method comprises the following steps: extracting keyword information in a question and answer question to be processed, and matching the keyword information with a database to generate a matching success signal or a matching failure signal; for the matching failure signal, performing secondary identification analysis on the question to be processed to obtain question segmentation segments through further segmentation, and matching the question segmentation segments with a database to generate a matching result; the method comprises the following steps: performing secondary identification analysis on questions to be processed, namely determining segmentation points, segmenting the questions, analyzing a plurality of segmented segments obtained by segmentation, performing sorting calculation, generating display information according to a priority sorting value, and transmitting the display information to a corresponding user. According to the method, secondary identification analysis can be carried out on questions and answers, secondary matching is further carried out on a plurality of segmentation segments, and better display information can be obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of natural language processing, and in particular to an intelligent question-answering method and system. Background Art

[0002] Question answering system is an advanced form of information retrieval system, which can answer users' natural language questions accurately and concisely, so as to meet people's needs for obtaining information quickly and accurately. With the development of artificial intelligence and natural language processing, question answering system has attracted more and more attention.

[0003] In the existing related art (application number CN201510976272.X), an intelligent question-answering system is disclosed, which includes: one or more question input modules, the question input modules are used to input questions to be answered, and set source identification for the questions, the source identification contains at least two or more large user information, the large users also include small users, the large users are independent of each other, and the small users between different large users are independent; a question library; a question matching module; a general answer library; one or more professional answer libraries; a question answering module, according to the matching result of the question matching module and the source identification of the question, find the answer to the corresponding question in the professional answer library or the general answer library and feed back the corresponding answer to the answer demander according to the source identification. However, when some existing intelligent question-answering systems are used, when users input questions for interactive question-answering, there will be a problem that the system cannot recognize the user input, which will further cause the purpose of question-answering to fail to be achieved. Secondly, when displaying the question-answering results, a large amount of data is not comprehensively sorted, and it cannot be displayed according to the characteristics of the question-answering results, which will cause the problem of reduced user experience.

[0004] Therefore, it is necessary to provide an intelligent question answering method to solve the above problems. Summary of the invention

[0005] The present invention intends to provide an intelligent question-answering method and system to solve the technical problems in the prior art that the existing methods cannot identify the questions input by the user and thus cannot obtain replies, and there is no comprehensive organization of data characteristics to obtain more effective replies. The technical problems to be solved by the present invention are achieved through the following technical solutions.

[0006] The first aspect of the present invention proposes an intelligent question and answer method, including: analyzing the acquired question and answer questions to be processed, extracting keyword information from the question and answer questions to be processed, and matching the keyword information with a database to generate a match success signal or a match failure signal; for the match failure signal, performing a secondary recognition analysis on the question and answer questions to be processed, segmenting the question and answer questions to be processed according to the length of the keyword information to obtain question segments, and matching the question segments with a database to generate a matching result; the secondary recognition analysis of the question and answer questions to be processed includes comparing the length of the commonly used information and the length of the keyword information in the question and answer questions to be processed to determine the segmentation point, so as to perform question segmentation, and analyzing the multiple segments obtained by segmentation; analyzing the generated matching results, screening to obtain similar results and dissimilar results, and sorting and calculating the similar results and dissimilar results, and generating display information according to the priority sorting value, and transmitting the display information to the corresponding user.

[0007] The second aspect of the present invention provides an intelligent question-answering system, which executes the intelligent question-answering method described in the first aspect of the present invention, and the intelligent question-answering system includes: a first analysis model, which analyzes the acquired question-answering questions to be processed, extracts the keyword information in the question-answering questions to be processed, and matches the keyword information with the database to generate a matching success signal or a matching failure signal; a second analysis module, for the matching failure signal, performs a secondary recognition analysis on the question-answering questions to be processed, divides the question-answering questions to be processed according to the length of the keyword information to obtain question segments, and matches the question segments with the database to generate matching results; the secondary recognition analysis of the question-answering questions to be processed includes comparing the length of the commonly used information and the length of the keyword information in the question-answering questions to be processed to determine the segmentation point, so as to perform question segmentation, and analyze the multiple segments obtained by segmentation; a calculation module, which analyzes the generated matching results, screens similar results and dissimilar results, and sorts and calculates the similar results and dissimilar results, and generates display information according to the priority sorting value, and transmits the display information to the corresponding user.

[0008] The embodiments of the present invention include the following advantages:

[0009] Compared with the prior art, the present invention recognizes and analyzes the acquired question and answer topics. For situations where recognition is unavailable, especially situations where recognition results cannot be obtained in one recognition, the present invention can perform a secondary recognition analysis on the question and answer topics. Specifically, the segmentation points are determined, and different segmentation methods are performed according to different information in the question and answer topics. Multiple segmentation segments are further matched for a secondary time, and the matching results are comprehensively organized according to the characteristics of the matching results themselves. This allows for a more effective display based on the characteristics of the matching results, giving priority to displaying results with a high degree of matching, and making it convenient for users to view them. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a flowchart of an example of the intelligent question-answering method of the present invention;

[0011] Figure 2 It is a flowchart of an implementation method of the intelligent question-answering method of the present invention;

[0012] Figure 3 It is a schematic diagram of a flow chart of performing a vector space search in a vector database in the question-answering method of the present invention;

[0013] Figure 4 is a schematic structural diagram of an example of a question-answering system according to the present invention;

[0014] Figure 5 is a schematic structural diagram of an electronic device embodiment according to the present invention;

[0015] Figure 6 It is a schematic diagram of the structure of a computer-readable medium embodiment according to the present invention. DETAILED DESCRIPTION

[0016] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0017] In view of the above problems, the present invention proposes an intelligent question and answer method, which analyzes the acquired question and answer questions to be processed, extracts keyword information from the question and answer questions to be processed, and matches the keyword information with a database to generate a match success signal or a match failure signal; for the match failure signal, the question and answer questions to be processed are subjected to secondary recognition analysis, the question and answer questions to be processed are segmented according to the length of the keyword information to obtain question segments, and the question segments are matched with a database to generate a matching result; the secondary recognition analysis of the question and answer questions to be processed includes comparing the length of the commonly used information and the length of the keyword information in the question and answer questions to be processed to determine the segmentation point, so as to perform question segmentation, and analyzing the multiple segments obtained by the segmentation; analyzing the generated matching results, screening to obtain similar results and dissimilar results, and ranking and calculating the similar results and dissimilar results, and generating display information according to the priority ranking value, and transmitting the display information to the corresponding user.

[0018] Refer to the following Figures 1 to 3 , the contents of the present invention will be described in detail.

[0019] Figure 1 It is a flowchart of an example of the intelligent question-answering method of the present invention.

[0020] Next, in step S101, the acquired question and answer title to be processed is analyzed, keyword information in the question and answer title to be processed is extracted, and the keyword information is matched with the database to generate a match success signal or a match failure signal.

[0021] In a specific implementation, a question to be processed (e.g., "What is the temperature today", such as the interactive process of medical remote operation data, medical consultation, shopping Q&A, and urban intelligent construction Q&A) is obtained, and the keywords in the question to be processed are identified, and the commonly used information in the question to be processed is obtained, and the commonly used information here is represented by commonly used characters or words, and the commonly used information is removed from the question to be processed, and the remaining information is marked as keyword information, and then the keyword information is matched with the database. The database is used to store data related to the interactive process of medical remote operation data, shopping Q&A, urban intelligent construction Q&A, weather Q&A, etc.

[0022] In a specific implementation, for example, in the application direction of the medical field, in a municipal hospital, when a user is seeing a doctor, the user can perform automatic query processing through the self-service machine set up in the lobby. For example, the user's pending question and answer question is first obtained as "How to go to the registration office to register". Then, the self-service machine will analyze the pending question and answer question, clarify the text subject of the pending question and answer question, extract the corresponding keywords "registration office" and "registration", and search and match in the above database according to the extracted keywords, and obtain information related to the keywords. When obtaining, the search will be performed according to the order in which the keywords appear, with "registration office" being searched first, and then "registration" being searched and matched in related order.

[0023] For the search matching results obtained, a corresponding route is generated according to the location of the corresponding "registration office" and displayed to the corresponding personnel in combination with voice prompts.

[0024] Specifically, the keyword information corresponding to each application scenario (such as the interaction process of medical remote operation data, medical consultation, shopping Q&A, and urban intelligent construction Q&A) is extracted through the following steps:

[0025] Step S201: clarify the text topics related to each application scenario (specifically including the interaction process of medical remote operation data, medical consultation, shopping Q&A, and urban intelligent construction Q&A).

[0026] Before extracting keywords, you must first clarify the text theme of the question to be answered, that is, understand the core issue, main object or key event of the text.

[0027] Step S202: Analyze the text structure of the question to be processed.

[0028] Specifically analyze the title and subtitle of the question to be processed. The title is usually the core summary of the text, specifically containing important keywords. The subtitle is usually a text that further explains the topic and can also provide more keyword clues. The beginning and end of the text paragraph contain the key information of the text paragraph. Key sentences: Key sentences in the text, such as the conclusion sentence, the theme sentence, etc., usually contain important keywords.

[0029] Step S203: Analyze the nouns and verbs of the question to be processed.

[0030] Analyze and determine the nouns and verbs in the questions to be processed that correspond to various application scenarios (such as the interaction process of medical remote operation data, medical consultation, shopping questions and answers, and urban intelligent construction questions and answers).

[0031] In addition, the words that appear frequently in the text of the question and answer question to be processed are counted, and these words contain keywords.

[0032] Step S204: Verify the extracted keywords.

[0033] After extracting keywords, verify the extracted keywords. Combine the keywords to see if they can accurately summarize the main content of the text. If the keywords do not reflect the core of the text well, you need to re-extract or adjust the keywords.

[0034] For example, for the following text: "With the continuous development of artificial intelligence technology, machine learning algorithms have achieved remarkable results in image recognition, natural language processing and other fields. Deep learning, as an important branch of machine learning, has achieved efficient processing of professional data in urban intelligent construction questions and answers by constructing deep neural networks. The professional data includes road length, road location coordinates, building height, building area, etc."

[0035] Specifically, keywords are extracted according to the above steps S201 to S204. The extraction and analysis results include nouns in the above text, including "artificial intelligence", "machine learning", "deep learning", "urban intelligent construction Q&A", "professional data" and other words, which are keywords. Among them, "artificial intelligence", "machine learning" and "deep learning" are all professional terms. "Professional data" appears twice, and its frequency of appearance in the above text is relatively high, and it is also used as a keyword. The following nouns also included in the text are further explanations of the keyword "professional data": road length, road location coordinates, building height, building area. Therefore, they are also extracted and used as keywords.

[0036] In another specific implementation, the question to be processed is "In the medical field, what are the examination items for a certain disease?" First, determine that its applicable application direction is medical. Then, analyze the common information in the question to be processed, such as "in", "of", "which", etc. After removing these common information, the remaining "medical field", "certain disease", and "examination items" are marked as keyword information. Then, match them with the above-mentioned database. When the database stores disease information, examination item data, and related research results in various medical fields, the corresponding information is extracted as the answer text to the purpose of the question to be processed.

[0037] When the keyword information successfully matches the keyword and / or text segment in the database, a match success signal is generated, and the matching result corresponding to the successful match is obtained and transmitted to the matching result analysis module. On the contrary, when the keyword information fails to match the keyword and / or text segment in the database, a match failure signal is generated. The text segment information includes the subject name representing each application scenario, nouns and qualifiers that appear more than a set number of times (for example, twice, three times or more).

[0038] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0039] Next, in step S102, for the matching failure signal, a secondary recognition analysis is performed on the question and answer title to be processed, and the question and answer title to be processed is segmented according to the length of the keyword information to obtain question segments, and the question segments are matched with the database to generate a matching result; the secondary recognition analysis of the question and answer title to be processed includes comparing the length of the commonly used information and the length of the keyword information in the question and answer title to be processed to determine the segmentation point, so as to perform question segmentation, and analyze the multiple segments obtained by segmentation.

[0040] For the matching failure signal, a secondary recognition is performed by using the question to be processed (for example, what question to be processed, and an example to be given in combination with the application scenario) Figure 2 in "Secondary Matching") to determine keyword information and common information for topic segmentation (specifically corresponding to Figure 2 The “information segmentation” in the figure) and analyze the multiple segmented segments obtained by segmentation to determine the matching result of the secondary matching.

[0041] When the length of the common information in the question and answer topic to be processed is greater than the length of the keyword information, the common information is used as a segmentation point to segment the question and answer topic to be processed to generate a plurality of topic segment segments.

[0042] When the length of the commonly used information in the question and answer topic to be processed is less than or equal to the length of the keyword information, the question and answer topic to be processed is segmented using the keyword information as a segmentation point to generate a plurality of topic segments.

[0043] The question to be processed is segmented according to the length of the keyword information to obtain the question segmentation, and the question segmentation is matched with the keywords and / or text segments in the database to generate matching results. The text segment information includes the subject name representing each application scenario, nouns and qualifiers that appear more than a set number of times (for example, twice, three times or more).

[0044] Specifically, the number of common information and keyword information in the question and answer topic is recorded, and recorded as the length L1 of the common information and the length L2 of the keyword information (specifically the total number of words of the common information and the keyword information, also called the length L1 of the common information and the length L2 of the keyword information), and the length L1 of the common information and the length L2 of the keyword information are compared. When the length L1 of the common information is greater than the length L2 of the keyword information, the common information is used as the dividing point to divide the nursing question and answer topic to generate the topic segmentation. When the length L1 of the common information is less than the length L2 of the keyword information, the keyword information is used as the dividing point to divide the question and answer topic to generate the topic segmentation.

[0045] In a specific implementation, the question to be processed is "What is the temperature today?" First, the question to be processed as a whole is matched with the database. If there is a corresponding matching result, the corresponding matching result (for example, 25 degrees Celsius to 30 degrees Celsius) is directly obtained.

[0046] If the result is not successfully matched, then match it according to the segmented segments of the question generated by segmentation, such as segmenting into "today", "weather temperature", "for", and "how much".

[0047] Specifically, as shown in the above example, the keyword information is "today" and "weather temperature", and the corresponding common information is "how much" and "for".

[0048] like Figure 3 As shown, the following steps are performed to analyze the multiple segmented segments obtained by segmentation:

[0049] Step S301: Match the question to be processed with the database. If there is a corresponding matching result, the matching result is screened and matched with the database according to the generated question segments.

[0050] Step S302: label the generated question segments as i, where i=1, 2, ..., j, and obtain the length L1i of the commonly used information and the length L2i of the keyword information in each question segment, and calculate the proportion of the number of commonly used information in the question segment i, that is, the first proportion Z1i, and calculate the average proportion of keyword information in all question segments of the question and answer questions to be processed, that is, the average proportion Z2p.

[0051] Step S303: Eliminate the question segments corresponding to the first proportion value Z1i being greater than the proportion mean value Z2p, match the remaining question segments with the database, and determine whether there are corresponding matching results. When there are matching results, obtain the matching results, and transmit the obtained matching results to the matching result analysis module to generate matching success information; when there are no matching results, obtain an unmatched success result, and transmit the unmatched success result to the matching result analysis module to generate an unmatched success signal.

[0052] Specifically, the proportion of the commonly used information length in the question segments is calculated. For example, the number of question segments is 4, and the commonly used information length is 2. Therefore, the calculated proportion is 1 / 2. Similarly, the proportion of commonly used information length in all question segments is calculated. Then, the average proportion of keyword information in all question segments of the question and answer questions to be processed is calculated, that is, the mean proportion. The mean proportion is compared with the commonly used information proportion and screened. At the same time, the commonly used information proportion values ​​that are lower than the mean are screened.

[0053] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0054] Next, in step S103, the generated matching results are analyzed, similar results and dissimilar results are screened, and the similar results and dissimilar results are sorted and calculated, and display information is generated according to the priority sorting value, and the display information is transmitted to the corresponding user.

[0055] In a specific implementation, the generated matching results are obtained and labeled as n, where n=1, 2, ..., m. Next, the matching results are screened and the matching results that have the same information as the question and answer to be processed are recorded as similar results a, where a=1, 2, ..., o. The remaining matching results are recorded as non-similar results b, where b=1, 2, ..., p.

[0056] The number of words that are the same as the similar result a and the question to be processed is obtained and recorded as the number of words La. The ratio of the number of words La to the similar result a is calculated and recorded as

[0057] Calculate the number of displays Ya of similar result a in the historical display data of the specified historical time period, and obtain the sorting order Yaq corresponding to the number of displays Ya. Use the following expression to calculate the priority sorting value Ka corresponding to the similar result a:

[0058]

[0059] Among them, Ka represents the priority ranking value corresponding to the similar result a; La represents the number of words that are the same between the similar result a and the question to be processed; Ya represents the number of times the similar result a is displayed in the historical display data of the specified historical time period; Yaq represents the sorting order corresponding to the qth display number Ya, and q = 1, 2, ..., u.

[0060] Specifically, the historical designated time period includes one month, two months, three months or half a year.

[0061] The number of impressions refers to the number of impressions on the same display platform during a specified historical time period.

[0062] In another specific implementation, all non-similar results b are obtained, and non-similar results b are classified according to similarity ratios to obtain classified results b. g , and g = 1, 2, ..., e, calculate the proportion of the same number of words in the same type of classification results as in the question to be processed, and calculate the average proportion of the same type of classification results and record it as Calculate the mean proportion of all classification results in turn.

[0063] Calculate the number of non-similar result b impressions Yb in the historical impression data of the specified historical time period g , and get the number of impressions Yb g The corresponding sorting order Yb g q, and q = 1, 2, ..., u; the priority ranking value Kb corresponding to the non-similar result b is calculated using the following expression:

[0064]

[0065] Among them, Kb represents the priority ranking value corresponding to the non-similar result b; Yb g Indicates the number of non-similar result b displays in the historical display data of the specified historical time period; Yb g q represents the qth impression number Yb g The corresponding sorting order, and q = 1, 2, ..., u; Represents the mean proportion of the same type of classification results.

[0066] Specifically, the historical designated time period includes one month, two months, three months or half a year.

[0067] The number of impressions refers to the number of impressions on the same display platform during a specified historical time period.

[0068] It should be noted that the above is only described as an optional example and cannot be understood as a limitation of the present invention. In addition, the above drawings are only schematic illustrations of the processing included in the method according to an exemplary embodiment of the present invention, and are not intended to be limiting. It is easy to understand that the processing shown in the above drawings does not indicate or limit the time sequence of these processing. In addition, it is also easy to understand that these processing can be, for example, executed synchronously or asynchronously in multiple modules.

[0069] Compared with the prior art, the present invention recognizes and analyzes the acquired question and answer topics. For situations where recognition is unavailable, especially situations where recognition results cannot be obtained in one recognition, the present invention can perform a secondary recognition analysis on the question and answer topics. Specifically, the segmentation points are determined, and different segmentation methods are performed according to different information in the question and answer topics. Multiple segmentation segments are further matched for a secondary time, and the matching results are comprehensively organized according to the characteristics of the matching results themselves. This allows for a more effective display based on the characteristics of the matching results, giving priority to displaying results with a high degree of matching, and making it convenient for users to view them.

[0070] The following is a system embodiment of the present invention, which can be used to implement the method embodiment of the present invention. For details not disclosed in the system embodiment of the present invention, please refer to the method embodiment of the present invention.

[0071] Figure 4 4 is a schematic diagram of the structure of an example of an intelligent question-answering system according to the present invention.

[0072] like Figure 4 As shown, the second aspect of the present disclosure provides an intelligent question answering system 400 , which includes a first analysis model 410 , a second analysis module 420 and a calculation module 430 .

[0073] In a specific embodiment, the first analysis model 410 analyzes the acquired pending question and answer questions, extracts the keyword information in the pending question and answer questions, and matches the keyword information with the database to generate a match success signal or a match failure signal. The second analysis module 420 performs a secondary recognition analysis on the pending question and answer questions for the match failure signal, divides the pending question and answer questions according to the length of the keyword information to obtain question segments, and matches the question segments with the database to generate matching results; the secondary recognition analysis of the pending question and answer questions includes comparing the length of the commonly used information and the length of the keyword information in the pending question and answer questions to determine the segmentation point, so as to perform question segmentation, and analyze the multiple segments obtained by segmentation. The calculation module 430 analyzes the generated matching results, screens similar results and dissimilar results, and sorts and calculates the similar results and dissimilar results, and generates display information according to the priority ranking value, and transmits the display information to the corresponding user.

[0074] According to an optional implementation manner, the length of the commonly used information and the length of the keyword information in the question and answer topic to be processed are compared to determine the segmentation point, so as to perform topic segmentation.

[0075] Specifically, secondary recognition is performed on the pending question-and-answer questions with matching failure signals to determine keyword information and common information.

[0076] When the length of the common information in the question and answer topic to be processed is greater than the length of the keyword information, the common information is used as a segmentation point to segment the question and answer topic to be processed to generate multiple question segments. When the length of the common information in the question and answer topic to be processed is less than or equal to the length of the keyword information, the keyword information is used as a segmentation point to segment the question and answer topic to be processed to generate multiple question segments.

[0077] According to an optional implementation manner, analyzing the multiple segmented segments obtained by segmentation specifically includes:

[0078] S1: Match the question to be processed with the database. If there is a corresponding matching result, the matching result is screened and matched with the database according to the generated question segment;

[0079] S2: The generated question segments are labeled as i, where i=1, 2, ..., j, and the length L1i of the common information and the length L2i of the keyword information in each question segment are obtained, and the proportion of the number of common information in the question segment i, i.e., the first proportion Z1i, is calculated, and the average proportion of the keyword information in all the question segments of the question-and-answer question to be processed, i.e., the average proportion Z2p, is calculated;

[0080] S3: Eliminate the question segments corresponding to the first proportion value Z1i that is greater than the proportion mean value Z2p, match the remaining question segments with the database, and determine whether there are corresponding matching results. When there are matching results, obtain the matching results, and transmit the obtained matching results to the matching result analysis module to generate matching success information; when there are no matching results, obtain unmatched success results, and transmit the unmatched success results to the matching result analysis module to generate unmatched success signals.

[0081] According to an optional implementation manner, the generated matching results are analyzed to screen out similar results and dissimilar results, and the similar results and dissimilar results are ranked and calculated.

[0082] Specifically, all matching results are obtained and labeled as n, where n=1, 2, ..., m. Then, the matching results are screened, and the matching results with the same information as the question to be processed are recorded as similar results a, where a=1, 2, ..., o. The remaining matching results are recorded as non-similar results b, where b=1, 2, ..., p.

[0083] The number of words that are the same as the similar result a and the question to be processed is obtained and recorded as the number of words La. The ratio of the number of words La to the similar result a is calculated and recorded as

[0084] Calculate the number of displays Ya of similar result a in the historical display data of the specified historical time period, and obtain the sorting order Yaq corresponding to the number of displays Ya. Use the following expression to calculate the priority sorting value Ka corresponding to the similar result a:

[0085]

[0086] Among them, Ka represents the priority ranking value corresponding to the similar result a; La represents the number of words that are the same between the similar result a and the question to be processed; Ya represents the number of times the similar result a is displayed in the historical display data of the specified historical time period; Yaq represents the sorting order corresponding to the qth display number Ya, and q = 1, 2, ..., u.

[0087] According to an optional implementation manner, it further includes: obtaining all non-similar results b, and classifying the non-similar results b according to the similarity ratio to obtain the classification result b g , and g = 1, 2, ..., e, then calculate the proportion of the same number of words in the same type of classification results as in the question to be processed, and calculate the average proportion of the same type of classification results and record it as Calculate the mean of the proportions of all classification results in turn;

[0088] Calculate the number of non-similar result b impressions Yb in the historical impression data of the specified historical time period g , and get the number of impressions Yb g The corresponding sorting order Yb g q, and q = 1, 2, ..., u; the priority ranking value Kb corresponding to the non-similar result b is calculated using the following expression:

[0089]

[0090] Among them, Kb represents the priority ranking value corresponding to the non-similar result b; Yb g Indicates the number of non-similar result b displays in the historical display data of the specified historical time period; Yb g q represents the qth impression number Yb g The corresponding sorting order, and q = 1, 2, ..., u; Represents the mean proportion of the same type of classification results.

[0091] According to an optional implementation manner, the extraction of keyword information in the question and answer questions to be processed includes the following steps: clarifying the text topics corresponding to each application scenario; analyzing the text structure of the question and answer questions to be processed; analyzing the nouns and verbs of the question and answer questions to be processed; and verifying the extracted keywords.

[0092] Specifically, the historical designated time period includes one month, two months, three months or half a year.

[0093] When the secondary recognition analysis obtains an unmatched success result, the unmatched success result is output to the user.

[0094] It should be noted that, in this embodiment, the intelligent question answering method performed by the intelligent question answering system is the same as Figure 1 The intelligent question-answering methods in are roughly the same, so the description of the same parts is omitted.

[0095] Figure 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

[0096] like Figure 5 As shown, the electronic device is presented in the form of a general computing device. The processor may be one or more and work in coordination. The present invention does not exclude distributed processing, that is, the processor may be dispersed in different physical devices. The electronic device of the present invention is not limited to a single entity, but may also be the sum of multiple physical devices.

[0097] The memory stores a computer executable program, which is usually a machine-readable code. The computer-readable program can be executed by the processor to enable the electronic device to perform the method of the present invention, or at least part of the steps in the method.

[0098] The memory includes a volatile memory, such as a random access memory unit (RAM) and / or a cache memory unit, and may also be a non-volatile memory, such as a read-only memory unit (ROM).

[0099] Optionally, in this embodiment, the electronic device further includes an I / O interface, which is used for the electronic device to exchange data with an external device. The I / O interface can represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus structures.

[0100] It should be understood that Figure 5 The electronic device shown is only an example of the present invention, and the electronic device of the present invention may also include elements or components not shown in the above examples. For example, some electronic devices also include display units such as display screens, and some electronic devices also include human-computer interaction elements such as buttons, keyboards, etc. As long as the electronic device can execute the computer-readable program in the memory to implement the method of the present invention or at least part of the steps of the method, it can be considered as an electronic device covered by the present invention.

[0101] Through the above description of the implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by combining software with necessary hardware. Figure 6 As shown, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes a number of commands to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present invention.

[0102] The software product may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0103] The computer readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with a command execution system, device, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0104] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0105] The computer-readable medium carries one or more programs (for example, computer-executable programs). When the one or more programs are executed by a device, the computer-readable medium implements the method of the present disclosure.

[0106] Those skilled in the art will appreciate that the above modules can be distributed in the device according to the description of the embodiment, or can be changed accordingly and only used in one or more devices different from the embodiment. The modules of the above embodiments can be combined into one module, or further divided into multiple sub-modules.

[0107] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several commands to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present invention.

[0108] The exemplary embodiments of the present invention are specifically shown and described above. It should be understood that the present invention is not limited to the detailed structure, configuration or implementation method described herein; on the contrary, the present invention is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended claims.

Claims

1. An intelligent question-answering method, characterized in that: include: Analyze the acquired question-and-answer questions to be processed, extract keyword information from the question-and-answer questions to be processed, and match the keyword information with the database to generate a match success signal or a match failure signal; For the match failure signal, the question and answer title to be processed is subjected to secondary recognition analysis, the question and answer title to be processed is segmented according to the length of the keyword information to obtain the question segmentation, and the question segmentation segmentation is matched with the database to generate a matching result; the secondary recognition analysis of the question and answer title to be processed includes comparing the length of the common information and the length of the keyword information in the question and answer title to be processed to determine the segmentation point, so as to segment the title, and analyze the multiple segmentation segments obtained by segmentation; The generated matching results are analyzed, similar results and dissimilar results are screened, and the similar results and dissimilar results are sorted and calculated. At the same time, display information is generated according to the priority sorting value, and the display information is transmitted to the corresponding user.

2. The intelligent question-answering method according to claim 1, characterized in that: The step of comparing the length of the common information and the length of the keyword information in the question-and-answer topic to be processed to determine the segmentation point so as to segment the topic includes: Perform secondary recognition on pending question-and-answer questions with matching failure signals to determine keyword information and common information; When the length of the common information in the question and answer topic to be processed is greater than the length of the keyword information, the common information is used as a segmentation point to segment the question and answer topic to be processed to generate a plurality of topic segmentation segments; When the length of the commonly used information in the question and answer topic to be processed is less than or equal to the length of the keyword information, the question and answer topic to be processed is segmented using the keyword information as a segmentation point to generate a plurality of topic segments.

3. The intelligent question-answering method according to claim 1 or 2, characterized in that: The analyzing of the multiple segmented segments obtained by segmentation includes: S1: Match the question to be processed with the database. If there is a corresponding matching result, the matching result is screened and matched with the database according to the generated question segment; S2: The generated question segments are labeled as i, where i=1, 2, ..., j, and the length L1i of the common information and the length L2i of the keyword information in each question segment are obtained, and the proportion of the number of common information in the question segment i, i.e., the first proportion Z1i, is calculated, and the average proportion of the keyword information in all the question segments of the question-and-answer question to be processed, i.e., the average proportion Z2p, is calculated; S3: Eliminate the question segments corresponding to the first proportion value Z1i that is greater than the proportion mean value Z2p, match the remaining question segments with the database, and determine whether there are corresponding matching results. When there are matching results, obtain the matching results, and transmit the obtained matching results to the matching result analysis module to generate matching success information; when there are no matching results, obtain unmatched success results, and transmit the unmatched success results to the matching result analysis module to generate unmatched success signals.

4. The intelligent question-answering method according to claim 1, characterized in that: The generated matching results are analyzed, similar results and dissimilar results are obtained by screening, and the similar results and dissimilar results are sorted and calculated, including: Obtain all matching results and label them as n, where n=1, 2, ..., m. Then screen the matching results, and record the matching results with the same information as the question to be processed as similar results a, where a=1, 2, ..., o. The remaining matching results are recorded as non-similar results b, where b=1, 2, ..., p. The number of words that are the same as the similar result a and the question to be processed is obtained and recorded as the number of words La. The ratio of the number of words La to the similar result a is calculated and recorded as Calculate the number of displays Ya of similar result a in the historical display data of the specified historical time period, and obtain the sorting order PYa corresponding to the number of displays Ya. Use the following expression to calculate the priority sorting value Ka corresponding to the similar result a: Among them, Ka represents the priority ranking value corresponding to the similar result a; La represents the number of words that are the same between the similar result a and the question to be processed; Ya represents the number of times the similar result a is displayed in the historical display data of the specified historical time period; Yaq represents the sorting order corresponding to the qth display number Ya, and q = 1, 2, ..., u.

5. The intelligent question-answering method according to claim 1, characterized in that: Further including: Get all non-similar results b, and classify the non-similar results b according to the similarity ratio to obtain the classification result b g , and g = 1, 2, ..., e, then calculate the proportion of the same number of words in the same type of classification results as in the question to be processed, and calculate the average proportion of the same type of classification results and record it as Calculate the mean of the proportions of all classification results in turn; Calculate the number of non-similar result b impressions Yb in the historical impression data of the specified historical time period g , and get the number of impressions Yb g The corresponding sorting order Yb g q, and q = 1, 2, ..., u; the priority ranking value Kb corresponding to the non-similar result b is calculated using the following expression: Among them, Kb represents the priority ranking value corresponding to the non-similar result b; Yb g Indicates the number of non-similar result b displays in the historical display data of the specified historical time period; Yb g q represents the qth impression number Yb g The corresponding sorting order, and q = 1, 2, ..., u; Represents the mean proportion of the same type of classification results.

6. The intelligent question-answering method according to claim 4 or 5, characterized in that: The step of extracting keyword information from the question-and-answer topic to be processed includes the following steps: Clarify the text topics corresponding to each application scenario; Analyze the text structure of the question to be processed; Analyze the nouns and verbs in the question and answer questions to be processed; Verify the extracted keywords.

7. The intelligent question-answering method according to claim 6, characterized in that: The historical specified time period includes one month, two months, three months or half a year.

8. The intelligent question-answering method according to claim 1, characterized in that: When the secondary recognition analysis obtains an unmatched success result, the unmatched success result is output to the user.

9. An intelligent question-answering system, characterized in that: It executes the intelligent question-answering method according to claim 1, and the intelligent question-answering system comprises: The first analysis model analyzes the acquired question-and-answer topic to be processed, extracts keyword information from the question-and-answer topic to be processed, and matches the keyword information with the database to generate a match success signal or a match failure signal; The second analysis module, for the match failure signal, performs secondary recognition analysis on the question and answer title to be processed, divides the question and answer title to be processed according to the length of the keyword information to obtain question segments, and matches the question segments with the database to generate a matching result; the secondary recognition analysis on the question and answer title to be processed includes comparing the length of the common information and the length of the keyword information in the question and answer title to be processed to determine the segmentation point, so as to perform question segmentation, and analyze the multiple segmented segments obtained by segmentation; The calculation module analyzes the generated matching results, screens similar results and dissimilar results, and sorts and calculates the similar results and dissimilar results, generates display information according to the priority sorting value, and transmits the display information to the corresponding user.

10. The intelligent question-answering system according to claim 9, characterized in that: Further including: Perform secondary recognition on pending question-and-answer questions with matching failure signals to determine keyword information and common information; When the length of the common information in the question and answer topic to be processed is greater than the length of the keyword information, the common information is used as a segmentation point to segment the question and answer topic to be processed to generate a plurality of topic segmentation segments; When the length of the commonly used information in the question and answer topic to be processed is less than or equal to the length of the keyword information, the question and answer topic to be processed is segmented using the keyword information as a segmentation point to generate a plurality of topic segments.

Images