Processing method and device of intelligent question answering system, storage medium and processor

Abstract

The application discloses a processing method and device of an intelligent question and answer system, a storage medium and a processor. The method comprises the following steps: obtaining initial question information of a target object, and analyzing and processing the initial question information by using a target network model to obtain target question information; converting the target question information into a first text vector according to a first target weight table corresponding to a standard question and answer data set of the intelligent question and answer system and a second target weight table corresponding to a historical question and answer data set of the target object; performing similarity calculation on the first text vector and a second text vector corresponding to each question and answer data in the standard question and answer data set respectively to determine a highest similarity value; and feeding back target question and answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold. Through the application, the problem of low question and answer accuracy of an intelligent customer service question and answer system in related technologies is solved.

Description

Technical Field

[0001] This application relates to the field of intelligent question answering technology, and more specifically, to a processing method and apparatus, storage medium and processor for an intelligent question answering system. Background Technology

[0002] With the development of internet and artificial intelligence technologies, intelligent customer service Q&A technology has rapidly developed in various fields, and its value has been recognized by society. Although intelligent customer service Q&A systems have been accepted by the public and applied in the financial industry, their problem-solving capabilities and customer experience still need improvement. Currently, the main challenges of intelligent customer service Q&A systems in the financial industry are as follows: First, human natural language has various ambiguities, requiring the accumulation of a large amount of knowledge and analysis of dialogue through context and scenario. For complex sentences, intelligent robots face certain difficulties in answering. Second, for users, each user has different language and input habits. Simply comparing and analyzing the user's questions with a standard question-and-answer database to obtain a corresponding answer cannot provide personalized solutions, and users cannot obtain a good user experience.

[0003] There is currently no effective solution to the problem that intelligent customer service question-and-answer systems often only compare user questions with a standard question-and-answer database to obtain answers, resulting in low accuracy. Summary of the Invention

[0004] The main objective of this application is to provide a processing method, apparatus, storage medium, and processor for an intelligent question-and-answer system, in order to solve the problem in related technologies where, when answering user questions, the system simply compares and analyzes the user's question with a standard question-and-answer database to obtain the corresponding answer, resulting in a low accuracy rate for intelligent customer service question-and-answer systems.

[0005] To achieve the above objectives, according to one aspect of this application, a processing method for an intelligent question-answering system is provided. The method includes: acquiring initial question information of a target object, and analyzing and processing the initial question information using a target network model to obtain target question information; converting the target question information into a first text vector based on a first target weight table corresponding to a standard question-answer dataset of the intelligent question-answering system and a second target weight table corresponding to a historical question-answer dataset of the target object, wherein the first target weight table consists of words in the standard question-answer dataset and their corresponding weight values, and the second target weight table consists of words in the historical question-answer dataset and their corresponding weight values; calculating the similarity between the first text vector and a second text vector corresponding to each question-answer data in the standard question-answer dataset, and determining the highest similarity value; if the highest similarity value is greater than a preset threshold, then feeding back the target question-answer data corresponding to the highest similarity value to the target object, wherein the target question-answer data is the question-answer data in the standard question-answer dataset.

[0006] Furthermore, before obtaining the initial question information of the target object and analyzing and processing the initial question information using the target network model to obtain the target question information, the method further includes: obtaining the standard question-and-answer dataset and cleaning the standard question-and-answer dataset to obtain a processed standard question-and-answer dataset; inputting the processed standard question-and-answer dataset into the target classifier for classification to obtain a standard question-and-answer dataset for each type of question and answer; and training the initial neural network model using the standard question-and-answer dataset for each type of question and answer to obtain the target network model.

[0007] Furthermore, before converting the target question information into a first text vector based on the first target weight table corresponding to the standard question-answer dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answer dataset of the target object, the method further includes: processing the standard question-answer dataset of each type of question-answer using a first preset algorithm to obtain the weight table corresponding to the standard question-answer dataset of each type of question-answer and the vocabulary database corresponding to the standard question-answer dataset of each type of question-answer.

[0008] Furthermore, before converting the target question information into a first text vector based on the first weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and the second weight table corresponding to the historical question-and-answer dataset of the target object, the method further includes: acquiring the historical question-and-answer dataset and cleaning the historical question-and-answer dataset to obtain a processed historical question-and-answer dataset; classifying the processed historical question-and-answer dataset using the target classifier to obtain a historical question-and-answer dataset for each type of question; and processing the historical question-and-answer dataset for each type of question to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question.

[0009] Further, processing the historical question-and-answer dataset for each type of question and answer to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question and answer includes: processing the historical question-and-answer dataset for each type of question and answer using a second preset algorithm based on the weight table corresponding to the standard question-and-answer dataset for each type of question and answer and the vocabulary database corresponding to the standard question-and-answer dataset for each type of question and answer to obtain an initial weight table for the historical question-and-answer dataset for each type of question and answer; and correcting the weight values ​​in the initial weight table using a Gaussian filter to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question and answer.

[0010] Further, converting the target question information into a first text vector based on the first target weight table corresponding to the standard question-answer dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answer dataset of the target object includes: obtaining the first target weight table from the weight table corresponding to the standard question-answer dataset of each type of question-answer based on the type of the target question information; obtaining the second target weight table from the latest weight table corresponding to the historical question-answer dataset of each type of question-answer based on the type of the target question information; and converting the target question information into a first text vector based on the first target weight table, the second target weight table, and the vocabulary database.

[0011] Furthermore, before calculating the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset to determine the highest similarity value, the method further includes: processing the standard question and answer dataset for each type of question and answer to obtain a text vector set corresponding to the standard question and answer dataset for each type of question and answer; and obtaining the second text vector from the text vector set corresponding to the standard question and answer dataset for each type of question and answer based on the type of the target question information.

[0012] Furthermore, after feeding back the target question-and-answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold, the method further includes: updating the standard question-and-answer dataset and the historical question-and-answer dataset based on the target question information and the target question-and-answer data.

[0013] To achieve the above objectives, according to another aspect of this application, a processing apparatus for an intelligent question-answering system is provided. The apparatus includes: a first acquisition unit, configured to acquire initial question information of a target object and analyze and process the initial question information using a target network model to obtain target question information; a conversion unit, configured to convert the target question information into a first text vector based on a first target weight table corresponding to a standard question-answer dataset of the intelligent question-answering system and a second target weight table corresponding to a historical question-answer dataset of the target object, wherein the first target weight table consists of words in the standard question-answer dataset and their corresponding weight values, and the second target weight table consists of words in the historical question-answer dataset and their corresponding weight values; a calculation unit, configured to calculate the similarity between the first text vector and a second text vector corresponding to each question-answer data in the standard question-answer dataset, and determine the highest similarity value; and a feedback unit, configured to, if the highest similarity value is greater than a preset threshold, feed back the target question-answer data corresponding to the highest similarity value to the target object, wherein the target question-answer data is the question-answer data in the standard question-answer dataset.

[0014] Furthermore, the device further includes: a second acquisition unit, configured to acquire the standard question-and-answer dataset and perform data cleaning on the standard question-and-answer dataset before acquiring the initial question information of the target object and analyzing and processing the initial question information using a target network model to obtain the target question information; a first classification unit, configured to input the processed standard question-and-answer dataset into a target classifier for classification to obtain a standard question-and-answer dataset for each type of question and answer; and a training unit, configured to learn and train the initial neural network model using the standard question-and-answer dataset for each type of question and answer to obtain the target network model.

[0015] Furthermore, the device further includes: a first processing unit, configured to process the standard question-and-answer dataset of each type of question and answer using a first preset algorithm before converting the target question information into a first text vector based on a first target weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and a second target weight table corresponding to the historical question-and-answer dataset of the target object, thereby obtaining a weight table corresponding to the standard question-and-answer dataset of each type of question and answer and a vocabulary database corresponding to the standard question-and-answer dataset of each type of question and answer.

[0016] Furthermore, the device further includes: a third acquisition unit, configured to acquire the historical question-and-answer dataset and perform data cleaning on the historical question-and-answer dataset to obtain a processed historical question-and-answer dataset before converting the target question information into a first text vector based on a first weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and a second weight table corresponding to the historical question-and-answer dataset of the target object; a second classification unit, configured to classify the processed historical question-and-answer dataset using the target classifier to obtain a historical question-and-answer dataset for each type of question; and a second processing unit, configured to process the historical question-and-answer dataset for each type of question to obtain a latest weight table corresponding to the historical question-and-answer dataset for each type of question.

[0017] Further, the second processing unit includes: a processing module, used to process the historical question-and-answer dataset of each type of question and answer according to the weight table corresponding to the standard question-and-answer dataset of each type of question and answer and the vocabulary library corresponding to the standard question-and-answer dataset of each type of question and answer, through a second preset algorithm, to obtain an initial weight table of the historical question-and-answer data of each type of question and answer; and a correction module, used to correct the weight values ​​in the initial weight table through a Gaussian filter, to obtain the latest weight table corresponding to the historical question-and-answer dataset of each type of question and answer.

[0018] Further, the conversion unit includes: a first selection module, configured to obtain a first target weight table from the weight table corresponding to the standard question-and-answer dataset for each type of question and answer based on the type of the target question information; a second selection module, configured to obtain a second target weight table from the latest weight table corresponding to the historical question-and-answer dataset for each type of question and answer based on the type of the target question information; and a conversion module, configured to convert the target question information into a first text vector based on the first target weight table, the second target weight table, and the vocabulary library.

[0019] Furthermore, the apparatus further includes: a third processing unit, configured to process the standard question-and-answer dataset for each type of question and answer before calculating the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question-and-answer dataset to determine the highest similarity value, thereby obtaining a text vector set corresponding to the standard question-and-answer dataset for each type of question and answer; and a selection unit, configured to obtain the second text vector from the text vector set corresponding to the standard question-and-answer dataset for each type of question and answer based on the type of the target question information.

[0020] Furthermore, the device further includes: an update unit, configured to update the standard question-and-answer dataset and the historical question-and-answer dataset based on the target question information and the target question-and-answer data after feeding back the target question-and-answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold.

[0021] To achieve the above objectives, according to one aspect of this application, a computer-readable storage medium is also provided, the storage medium storing a program, wherein, when the program is executed, the device where the storage medium is located controls the execution of the processing method of the intelligent question-answering system described in any one of the above claims.

[0022] To achieve the above objectives, according to one aspect of this application, a processor is also provided, the processor being configured to run a program, wherein the program, when running, executes the processing method of the intelligent question-answering system described in any of the preceding claims.

[0023] This application employs the following steps: First, initial question information of the target object is obtained, and the initial question information is analyzed and processed using a target network model to obtain target question information. Second, based on the first target weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and the second target weight table corresponding to the historical question-and-answer dataset of the target object, the target question information is converted into a first text vector. The first target weight table consists of words in the standard question-and-answer dataset and their corresponding weight values, while the second target weight table consists of words in the historical question-and-answer dataset and their corresponding weight values. Third, the first text vector is compared with the second text vector corresponding to each question-and-answer data in the standard question-and-answer dataset to determine the highest similarity value. Fourth, if the highest similarity value is greater than a preset threshold, the target question-and-answer data corresponding to the highest similarity value is fed back to the target object. The target question-and-answer data is the question-and-answer data in the standard question-and-answer dataset. This addresses the problem in related technologies where, when answering user questions, the system simply compares and analyzes the user's question with a standard question-and-answer database to obtain the corresponding answer, resulting in low accuracy of the intelligent customer service question-and-answer system. First, after obtaining the initial question information of the target object, it is analyzed and processed into target question information through the target network model. Then, according to the first target weight table corresponding to the standard question and answer dataset and the second target weight table corresponding to the historical question and answer dataset of the target object, the target question information is converted into a first text vector. Then, the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset is calculated. When the highest similarity value is greater than a preset threshold, the corresponding question and answer data is fed back to the target object. This fully takes into account the different language habits and input habits of each user, thereby improving the question and answer accuracy of the intelligent customer service question and answer system. Attached Figure Description

[0024] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0025] Figure 1 This is a flowchart of a processing method for an intelligent question-answering system provided according to an embodiment of this application;

[0026] Figure 2 This is a flowchart of an optional intelligent question-answering system processing method provided according to an embodiment of this application;

[0027] Figure 3 This is a schematic diagram of the processing device of the intelligent question-answering system provided according to the embodiments of this application. Detailed Implementation

[0028] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0029] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0031] The present invention will now be described in conjunction with preferred implementation steps. Figure 1 This is a flowchart of a processing method for an intelligent question-answering system provided according to an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0032] Step S101: Obtain the initial problem information of the target object, and use the target network model to analyze and process the initial problem information to obtain the target problem information.

[0033] Specifically, when a user asks a question (i.e., obtaining the initial question information of the target audience as described above), this initial question information is input into the target network model (e.g., a trained Long Short-Term Neural Network). The target network model analyzes and processes the initial question information to obtain the target question information. For example, if a user inputs the word "borrow," the target network model will convert "borrow" to "loan" to better compare and analyze it with a standard question-and-answer dataset. Each user has different language and input habits, so preliminary analysis and processing through the target network model can effectively improve the accuracy of the intelligent customer service question-and-answer system.

[0034] Step S102: Based on the first target weight table corresponding to the standard question and answer dataset of the intelligent question answering system and the second target weight table corresponding to the historical question and answer dataset of the target object, the target question information is converted into a first text vector. The first target weight table consists of words in the standard question and answer dataset and the weight values ​​corresponding to the words in the standard question and answer dataset, and the second target weight table consists of words in the historical question and answer dataset and the weight values ​​corresponding to the words in the historical question and answer dataset.

[0035] Specifically, based on the first target weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and the second target weight table corresponding to the historical question-and-answer dataset of the target object, the obtained target question information is converted into a corresponding text vector (i.e., the first text vector mentioned above). It should be noted that the first text vector also includes the weight values ​​corresponding to the words in the target question information. The weight value of each word is the product of the weight value of the word in the first target weight table and the weight value of the word in the second target weight table.

[0036] Step S103: Calculate the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset, and determine the highest similarity value.

[0037] Specifically, each question and answer data in the standard question and answer dataset is converted into a corresponding second text vector. Then, the similarity between the first text vector and the second text vector is calculated to obtain multiple similarity values. These similarity values ​​are then sorted in descending order, and the question and answer data corresponding to the highest similarity value is selected.

[0038] Step S104: If the highest similarity value is greater than the preset threshold, the target question-and-answer data corresponding to the highest similarity value is fed back to the target object, wherein the target question-and-answer data is the question-and-answer data in the standard question-and-answer dataset.

[0039] Specifically, the highest similarity value obtained is compared with a preset threshold (e.g., 90%). If it is higher than the threshold, the question and answer data corresponding to the highest similarity value is fed back to the user; if it is lower than the threshold, the user is directly transferred to a human customer service representative to handle and answer the question.

[0040] In summary, after obtaining the initial question information of the target object, it is analyzed and processed into target question information through a target network model. Then, based on the first target weight table corresponding to the standard question-and-answer dataset and the second target weight table corresponding to the historical question-and-answer dataset of the target object, the target question information is converted into a first text vector. Next, the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question-and-answer dataset is calculated. When the highest similarity value is greater than a preset threshold, the corresponding question and answer data is fed back to the target object. This fully considers the different language habits and input habits of each user, thereby improving the question-and-answer accuracy of the intelligent customer service question-and-answer system.

[0041] To standardize and normalize user-submitted questions, the target network model is obtained as follows: a standard question-and-answer dataset is acquired and cleaned to obtain a processed standard question-and-answer dataset; the processed standard question-and-answer dataset is input into the target classifier for classification to obtain a standard question-and-answer dataset for each category; the initial neural network model is trained using the standard question-and-answer dataset for each category to obtain the target network model.

[0042] Specifically, the standard question-and-answer dataset of the current intelligent question-and-answer system is obtained. This dataset is then cleaned (e.g., invalid question-and-answer data is removed) to obtain a processed standard question-and-answer dataset. A target classifier is then used to classify this dataset to improve subsequent processing efficiency. The categories can be set according to actual needs; for example, financial institutions can set the following categories: loan category dataset, deposit category dataset, and other categories. The initial neural network model is then trained using the standard question-and-answer dataset for each category to obtain the target network model. Analyzing and processing the initial question information using the target network model effectively improves the question-and-answer accuracy of the intelligent customer service question-and-answer system.

[0043] Obtaining the first target weight table corresponding to the standard question-and-answer dataset is crucial for this application. Therefore, in the processing method of the intelligent question-and-answer system provided in this application embodiment, the first target weight table is obtained by the following steps: processing the standard question-and-answer dataset of each type of question and answer through a first preset algorithm to obtain the weight table corresponding to the standard question-and-answer dataset of each type of question and answer and the vocabulary database corresponding to the standard question-and-answer dataset of each type of question and answer.

[0044] Specifically, the TF-IDF algorithm (the first preset algorithm mentioned above) is used to process the standard question-and-answer dataset for each question-and-answer category, resulting in the ES inverted index table (i.e., the weight table mentioned above) and the corresponding vocabulary for each category's standard question-and-answer dataset. The TF-IDF values ​​in the ES inverted index table represent the weight values ​​corresponding to words in the standard question-and-answer dataset for each category. The TF-IDF algorithm is a statistical method that effectively evaluates the importance of a word to a document. Using the above method, the weight table and the corresponding vocabulary for each category's standard question-and-answer dataset can be obtained quickly and accurately.

[0045] Obtaining the second target weight table corresponding to the historical question-and-answer dataset is crucial for this application. Therefore, in the processing method of the intelligent question-and-answer system provided in this application embodiment, the second target weight table is obtained by the following steps: acquiring the historical question-and-answer dataset and cleaning the historical question-and-answer dataset to obtain the processed historical question-and-answer dataset; classifying the processed historical question-and-answer dataset using a target classifier to obtain the historical question-and-answer dataset for each type of question and answer; and processing the historical question-and-answer dataset for each type of question and answer to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question and answer.

[0046] Specifically, the historical question-and-answer dataset is first cleaned to obtain a processed historical question-and-answer dataset. A target classifier is then used to classify this processed dataset, ensuring the categories match those in the labeled question-and-answer dataset. Then, the historical question-and-answer dataset for each category is processed again to obtain the latest weight table for each category.

[0047] Since important keywords that users habitually use may not appear frequently in a user's historical question-and-answer dataset, the weight values ​​of these important keywords are readjusted to improve the accuracy of the latest weight table corresponding to the historical question-and-answer dataset. This involves increasing the weight of these important keywords while decreasing the weight of less important keywords, thereby providing targeted and personalized question-and-answer services to each user. In the processing method of the intelligent question-and-answer system provided in this application embodiment, the latest weight table corresponding to the historical question-and-answer dataset for each type of question-and-answer is obtained in the following way: Based on the weight table corresponding to the standard question-and-answer dataset for each type of question-and-answer and the vocabulary library corresponding to the standard question-and-answer dataset for each type of question-and-answer, the historical question-and-answer dataset for each type of question-and-answer is processed using a second preset algorithm to obtain an initial weight table for the historical question-and-answer data of each type of question-and-answer; the weight values ​​in the initial weight table are corrected using a Gaussian filter to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question-and-answer.

[0048] Specifically, the initial weight values ​​are adjusted using the density function of the normal distribution in a Gaussian filter. A sliding window weight value is set according to actual needs (e.g., 0.01 or 0.05). For important words, the weight values ​​are recalculated by combining the words on either side of them with the set sliding window weight value, resulting in the latest weight values ​​for the important words. Through these steps, the accuracy of obtaining the latest weight table corresponding to the historical question-and-answer dataset for each type of question can be effectively improved, thereby achieving targeted and personalized question-and-answer services for each user.

[0049] To quickly obtain the first text vector, the following steps are used to convert the target question information into a first text vector: based on the type of the target question information, a first target weight table is obtained from the weight table corresponding to the standard question-and-answer dataset for each type of question and answer; based on the type of the target question information, a second target weight table is obtained from the latest weight table corresponding to the historical question-and-answer dataset for each type of question and answer; based on the first target weight table, the second target weight table, and the vocabulary database, the target question information is converted into a first text vector.

[0050] Specifically, based on the type of the user's target question information, a first target weight table and a second target weight table of the corresponding type are selected. Then, based on the first target weight table, the second target weight table, and the vocabulary database, the target question information is converted into a first text vector. Selecting the corresponding target weight table based on the type of target question information can effectively improve the speed of obtaining the first text vector, thereby improving the question-and-answer efficiency of the intelligent customer service question-and-answer system.

[0051] In order to calculate the similarity between the first text vector and the second text vector, the processing method of the intelligent question answering system provided in this application embodiment adopts the following steps to obtain the second text vector: process the standard question answering dataset for each type of question answering to obtain the text vector set corresponding to the standard question answering dataset for each type of question answering; and obtain the second text vector from the text vector set corresponding to the standard question answering dataset for each type of question answering according to the type of target question information.

[0052] Specifically, firstly, each question and answer data point in the standard question and answer dataset for each type of question and answer is converted into a corresponding text vector. Then, based on the type of the target question information, a second text vector is obtained from the text vector set. Selecting the second text vector by category can effectively improve the question and answer efficiency of the intelligent customer service question and answer system.

[0053] Since users will continue to ask questions, in order to improve the accuracy of the answers, it is necessary to update the historical question-and-answer dataset and the standard question-and-answer dataset. In the processing method of the intelligent question-and-answer system provided in this application embodiment, the following method is adopted: update the standard question-and-answer dataset and the historical question-and-answer dataset according to the target question information and the target question-and-answer data.

[0054] In an optional embodiment, the standard question-and-answer dataset and the historical question-and-answer dataset are updated using target question information and target question-and-answer data. In most cases, this involves updating the weight values ​​of various words to improve subsequent question matching and answering of user queries more accurately. The weight values ​​of words can be updated using the following formula: Among them, W xi(h+1) For the new weight of the target words, W xi(h) is the original weight of the target word, TF is the number of times the target word appears in the target dataset, and IDF is the inverse document frequency of the target dataset.

[0055] In an alternative embodiment, such as Figure 2 The flowchart shows an optional processing method for an intelligent question-answering system. For a user's historical question-answering dataset: Step 1, preprocess (i.e., data cleaning), classify, etc., the historical question-answering dataset to obtain an initial weight table corresponding to each category of historical question-answering dataset; Step 2, correct the weight values ​​of important words using a Gaussian filter to obtain the latest weight table of each category of historical question-answering dataset.

[0056] For the standard question-and-answer dataset: Step 1, preprocess and classify the standard question-and-answer dataset to obtain the standard question-and-answer dataset for each category; Step 2, process the standard question-and-answer dataset for each category using the TF-IDF algorithm to obtain the weight table and vocabulary corresponding to the standard question-and-answer dataset for each category; Step 3, train the LSTM neural network model using the standard question-and-answer dataset for each category to obtain the target network model.

[0057] The process is as follows: Step 1, obtain the user's question; Step 2, analyze and process the question using the target network model to obtain the target question information; Step 3, based on the latest weight table of the historical question-and-answer dataset for each category and the corresponding weight table of the standard question-and-answer dataset for each category, convert the target question information into a first text vector and convert the standard question-and-answer dataset for each category into a second text vector; Step 4, calculate the similarity between the first text vector and the second text vector; Step 5, if the highest similarity value is higher than the threshold, the question-and-answer data corresponding to the highest similarity value is fed back to the user; Step 6, if the highest similarity value is lower than the threshold, the user's question is answered manually.

[0058] The above methods can be used to provide more efficient and accurate answers to questions from different users.

[0059] The processing method of the intelligent question-answering system provided in this application embodiment obtains the initial question information of the target object and analyzes and processes the initial question information using a target network model to obtain the target question information. Based on the first target weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answering dataset of the target object, the target question information is converted into a first text vector. The first target weight table consists of words in the standard question-answering dataset and their corresponding weight values, while the second target weight table consists of words in the historical question-answering dataset and their corresponding weight values. The first text vector is then compared with the second text vector corresponding to each question-answering data in the standard question-answering dataset to determine the highest similarity value. If the highest similarity value is greater than a preset threshold, the target question-answering data corresponding to the highest similarity value is fed back to the target object. The target question-answering data is the question-answering data in the standard question-answering dataset. This solves the problem in related technologies where, when answering user questions, the system simply compares and analyzes the user's question with a standard question-answering database to obtain the corresponding answer, resulting in a low accuracy rate for intelligent customer service question-answering systems. First, after obtaining the initial question information of the target object, it is analyzed and processed into target question information through the target network model. Then, according to the first target weight table corresponding to the standard question and answer dataset and the second target weight table corresponding to the historical question and answer dataset of the target object, the target question information is converted into a first text vector. Then, the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset is calculated. When the highest similarity value is greater than a preset threshold, the corresponding question and answer data is fed back to the target object. This fully takes into account the different language habits and input habits of each user, thereby improving the question and answer accuracy of the intelligent customer service question and answer system.

[0060] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0061] This application also provides a processing apparatus for an intelligent question-answering system. It should be noted that the processing apparatus for the intelligent question-answering system in this application can be used to execute the processing method for the intelligent question-answering system provided in this application. The processing apparatus for the intelligent question-answering system provided in this application will be described below.

[0062] Figure 3This is a schematic diagram of a processing device for an intelligent question-answering system according to an embodiment of this application. Figure 3 As shown, the device includes: a first acquisition unit 301, a conversion unit 302, a calculation unit 303, and a feedback unit 304.

[0063] The first acquisition unit 301 is used to acquire the initial problem information of the target object and to analyze and process the initial problem information using the target network model to obtain the target problem information.

[0064] The conversion unit 302 is used to convert the target question information into a first text vector based on the first target weight table corresponding to the standard question and answer dataset of the intelligent question answering system and the second target weight table corresponding to the historical question and answer dataset of the target object. The first target weight table consists of words in the standard question and answer dataset and the weight values ​​corresponding to the words in the standard question and answer dataset, and the second target weight table consists of words in the historical question and answer dataset and the weight values ​​corresponding to the words in the historical question and answer dataset.

[0065] The calculation unit 303 is used to calculate the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset, and to determine the highest similarity value.

[0066] Feedback unit 304 is used to feed back the target question-and-answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold. The target question-and-answer data is the question-and-answer data in the standard question-and-answer dataset.

[0067] The processing device of the intelligent question-answering system provided in this application embodiment acquires the initial question information of the target object through the first acquisition unit 301, and analyzes and processes the initial question information using a target network model to obtain target question information; the conversion unit 302 converts the target question information into a first text vector according to the first target weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answering dataset of the target object. The first target weight table consists of words in the standard question-answering dataset and the weight values ​​corresponding to those words, while the second target weight table consists of words in the historical question-answering dataset and historical question-answering data. The dataset consists of weight values ​​corresponding to words; the calculation unit 303 calculates the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset to determine the highest similarity value; if the highest similarity value is greater than a preset threshold, the feedback unit 304 feeds back the target question and answer data corresponding to the highest similarity value to the target object. The target question and answer data is the question and answer data in the standard question and answer dataset. This solves the problem in related technologies where, when answering questions raised by users, the system simply compares and analyzes the questions raised by users with the standard question and answer database to obtain the corresponding answers, resulting in a low accuracy rate of questions and answers in the intelligent customer service question and answer system. First, after obtaining the initial question information of the target object, it is analyzed and processed into target question information through the target network model. Then, according to the first target weight table corresponding to the standard question and answer dataset and the second target weight table corresponding to the historical question and answer dataset of the target object, the target question information is converted into a first text vector. Then, the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset is calculated. When the highest similarity value is greater than a preset threshold, the corresponding question and answer data is fed back to the target object. This fully takes into account the different language habits and input habits of each user, thereby improving the question and answer accuracy of the intelligent customer service question and answer system.

[0068] Optionally, in the processing apparatus of the intelligent question-answering system provided in the embodiments of this application, the apparatus further includes: a second acquisition unit, used to acquire a standard question-answer dataset and perform data cleaning on the standard question-answer dataset before acquiring the initial question information of the target object and analyzing and processing the initial question information using a target network model to obtain the target question information; a first classification unit, used to input the processed standard question-answer dataset into a target classifier for classification to obtain a standard question-answer dataset for each type of question; and a training unit, used to learn and train the initial neural network model using the standard question-answer dataset for each type of question to obtain the target network model.

[0069] Optionally, in the processing apparatus of the intelligent question-answering system provided in the embodiments of this application, the apparatus further includes: a first processing unit, configured to process the standard question-answering dataset of each type of question-answering according to a first target weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and a second target weight table corresponding to the historical question-answering dataset of the target object before converting the target question information into a first text vector, thereby obtaining a weight table corresponding to the standard question-answering dataset of each type of question-answering and a vocabulary database corresponding to the standard question-answering dataset of each type of question-answering.

[0070] Optionally, in the processing apparatus of the intelligent question-answering system provided in this application embodiment, the apparatus further includes: a third acquisition unit, configured to acquire a historical question-answering dataset and perform data cleaning on the historical question-answering dataset before converting the target question information into a first text vector based on a first weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and a second weight table corresponding to the historical question-answering dataset of the target object; a second classification unit, configured to classify the processed historical question-answering dataset using a target classifier to obtain a historical question-answering dataset for each type of question-answering; and a second processing unit, configured to process the historical question-answering dataset for each type of question-answering to obtain the latest weight table corresponding to the historical question-answering dataset for each type of question-answering.

[0071] Optionally, in the processing device of the intelligent question-answering system provided in this application embodiment, the second processing unit includes: a processing module, used to process the historical question-answering dataset of each type of question-answering according to the weight table corresponding to the standard question-answering dataset of each type of question-answering and the vocabulary library corresponding to the standard question-answering dataset of each type of question-answering, through a second preset algorithm, to obtain an initial weight table of the historical question-answering data of each type of question-answering; and a correction module, used to correct the weight values ​​in the initial weight table through a Gaussian filter, to obtain the latest weight table corresponding to the historical question-answering dataset of each type of question-answering.

[0072] Optionally, in the processing apparatus of the intelligent question-answering system provided in this application embodiment, the conversion unit 302 includes: a first selection module, used to obtain a first target weight table from the weight table corresponding to the standard question-answer dataset for each type of question and answer based on the type of the target question information; a second selection module, used to obtain a second target weight table from the latest weight table corresponding to the historical question-answer dataset for each type of question and answer based on the type of the target question information; and a conversion module, used to convert the target question information into a first text vector based on the first target weight table, the second target weight table, and the vocabulary library.

[0073] Optionally, in the processing apparatus of the intelligent question-answering system provided in the embodiments of this application, the apparatus further includes: a third processing unit, configured to process the standard question-answering dataset for each type of question-answering before calculating the similarity between the first text vector and the second text vector corresponding to each question-answering data in the standard question-answering dataset to determine the highest similarity value; and a selection unit, configured to obtain the second text vector from the text vector set corresponding to the standard question-answering dataset for each type of question-answering based on the type of the target question information.

[0074] Optionally, in the processing device of the intelligent question-answering system provided in the embodiments of this application, the device further includes: an update unit, used to update the standard question-answering dataset and the historical question-answering dataset based on the target question information and the target question-answering data after feeding back the target question-answering data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold.

[0075] The processing device of the intelligent question-answering system includes a processor and a memory. The first acquisition unit 301, conversion unit 302, calculation unit 303 and feedback unit 304 mentioned above are all stored in the memory as program units. The processor executes the program units stored in the memory to realize the corresponding functions.

[0076] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and the processing of the intelligent question-answering system can be achieved by adjusting the kernel parameters.

[0077] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0078] This invention provides a computer-readable storage medium storing a program thereon, which, when executed by a processor, implements a processing method for an intelligent question-answering system.

[0079] This invention provides a processor for running a program, wherein the program executes a processing method for an intelligent question-answering system. The processor executes the following steps when running the program: acquiring initial question information of a target object and analyzing and processing the initial question information using a target network model to obtain target question information; converting the target question information into a first text vector based on a first target weight table corresponding to the standard question-answer dataset of the intelligent question-answering system and a second target weight table corresponding to the historical question-answer dataset of the target object, wherein the first target weight table consists of words in the standard question-answer dataset and their corresponding weight values, and the second target weight table consists of words in the historical question-answer dataset and their corresponding weight values; calculating the similarity between the first text vector and the second text vector corresponding to each question-answer data in the standard question-answer dataset to determine the highest similarity value; if the highest similarity value is greater than a preset threshold, feeding back the target question-answer data corresponding to the highest similarity value to the target object, wherein the target question-answer data is the question-answer data in the standard question-answer dataset.

[0080] Optionally, before obtaining the initial question information of the target object and analyzing and processing the initial question information using the target network model to obtain the target question information, the method further includes: obtaining a standard question-and-answer dataset and cleaning the standard question-and-answer dataset to obtain a processed standard question-and-answer dataset; inputting the processed standard question-and-answer dataset into the target classifier for classification to obtain a standard question-and-answer dataset for each type of question and answer; and training the initial neural network model using the standard question-and-answer dataset for each type of question and answer to obtain the target network model.

[0081] Optionally, before converting the target question information into a first text vector based on the first target weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and the second target weight table corresponding to the historical question-and-answer dataset of the target object, the method further includes: processing the standard question-and-answer dataset of each type of question and answer using a first preset algorithm to obtain the weight table corresponding to the standard question-and-answer dataset of each type of question and answer and the vocabulary database corresponding to the standard question-and-answer dataset of each type of question and answer.

[0082] Optionally, before converting the target question information into a first text vector based on the first weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and the second weight table corresponding to the historical question-and-answer dataset of the target object, the method further includes: obtaining a historical question-and-answer dataset and cleaning the historical question-and-answer dataset to obtain a processed historical question-and-answer dataset; classifying the processed historical question-and-answer dataset using a target classifier to obtain a historical question-and-answer dataset for each type of question; and processing the historical question-and-answer dataset for each type of question to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question.

[0083] Optionally, processing the historical question-and-answer dataset for each type of question-and-answer to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question-and-answer includes: processing the historical question-and-answer dataset for each type of question-and-answer using a second preset algorithm based on the weight table corresponding to the standard question-and-answer dataset for each type of question-and-answer and the vocabulary database corresponding to the standard question-and-answer dataset for each type of question-and-answer, to obtain the initial weight table for the historical question-and-answer data of each type of question-and-answer; and correcting the weight values ​​in the initial weight table using a Gaussian filter to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question-and-answer.

[0084] Optionally, converting target question information into a first text vector based on a first target weight table corresponding to the standard question-answer dataset of the intelligent question-answering system and a second target weight table corresponding to the historical question-answer dataset of the target object includes: obtaining a first target weight table from the weight table corresponding to the standard question-answer dataset of each type of question based on the type of target question information; obtaining a second target weight table from the latest weight table corresponding to the historical question-answer dataset of each type of question based on the type of target question information; and converting the target question information into a first text vector based on the first target weight table, the second target weight table, and the vocabulary database.

[0085] Optionally, before calculating the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset to determine the highest similarity value, the method further includes: processing the standard question and answer dataset for each type of question and answer to obtain a set of text vectors corresponding to the standard question and answer dataset for each type of question and answer; and obtaining the second text vector from the set of text vectors corresponding to the standard question and answer dataset for each type of question and answer based on the type of target question information.

[0086] Optionally, after feeding back the target question-and-answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold, the method further includes: updating the standard question-and-answer dataset and the historical question-and-answer dataset based on the target question information and the target question-and-answer data. The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0087] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program with the following method steps: obtaining initial question information of a target object, and analyzing and processing the initial question information using a target network model to obtain target question information; converting the target question information into a first text vector based on a first target weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and a second target weight table corresponding to the historical question-and-answer dataset of the target object, wherein the first target weight table consists of words in the standard question-and-answer dataset and the weight values ​​corresponding to the words in the standard question-and-answer dataset, and the second target weight table consists of words in the historical question-and-answer dataset and the weight values ​​corresponding to the words in the historical question-and-answer dataset; calculating the similarity between the first text vector and the second text vector corresponding to each question-and-answer data in the standard question-and-answer dataset, and determining the highest similarity value; if the highest similarity value is greater than a preset threshold, then feeding back the target question-and-answer data corresponding to the highest similarity value to the target object, wherein the target question-and-answer data is the question-and-answer data in the standard question-and-answer dataset.

[0088] Optionally, before obtaining the initial question information of the target object and analyzing and processing the initial question information using the target network model to obtain the target question information, the method further includes: obtaining a standard question-and-answer dataset and cleaning the standard question-and-answer dataset to obtain a processed standard question-and-answer dataset; inputting the processed standard question-and-answer dataset into the target classifier for classification to obtain a standard question-and-answer dataset for each type of question and answer; and training the initial neural network model using the standard question-and-answer dataset for each type of question and answer to obtain the target network model.

[0089] Optionally, before converting the target question information into a first text vector based on the first target weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and the second target weight table corresponding to the historical question-and-answer dataset of the target object, the method further includes: processing the standard question-and-answer dataset of each type of question and answer using a first preset algorithm to obtain the weight table corresponding to the standard question-and-answer dataset of each type of question and answer and the vocabulary database corresponding to the standard question-and-answer dataset of each type of question and answer.

[0090] Optionally, before converting the target question information into a first text vector based on the first weight table corresponding to the standard question-and-answer dataset of the intelligent question-and-answer system and the second weight table corresponding to the historical question-and-answer dataset of the target object, the method further includes: obtaining a historical question-and-answer dataset and cleaning the historical question-and-answer dataset to obtain a processed historical question-and-answer dataset; classifying the processed historical question-and-answer dataset using a target classifier to obtain a historical question-and-answer dataset for each type of question; and processing the historical question-and-answer dataset for each type of question to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question.

[0091] Optionally, processing the historical question-and-answer dataset for each type of question-and-answer to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question-and-answer includes: processing the historical question-and-answer dataset for each type of question-and-answer using a second preset algorithm based on the weight table corresponding to the standard question-and-answer dataset for each type of question-and-answer and the vocabulary database corresponding to the standard question-and-answer dataset for each type of question-and-answer, to obtain the initial weight table for the historical question-and-answer data of each type of question-and-answer; and correcting the weight values ​​in the initial weight table using a Gaussian filter to obtain the latest weight table corresponding to the historical question-and-answer dataset for each type of question-and-answer.

[0092] Optionally, converting target question information into a first text vector based on a first target weight table corresponding to the standard question-answer dataset of the intelligent question-answering system and a second target weight table corresponding to the historical question-answer dataset of the target object includes: obtaining a first target weight table from the weight table corresponding to the standard question-answer dataset of each type of question based on the type of target question information; obtaining a second target weight table from the latest weight table corresponding to the historical question-answer dataset of each type of question based on the type of target question information; and converting the target question information into a first text vector based on the first target weight table, the second target weight table, and the vocabulary database.

[0093] Optionally, before calculating the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset to determine the highest similarity value, the method further includes: processing the standard question and answer dataset for each type of question and answer to obtain a set of text vectors corresponding to the standard question and answer dataset for each type of question and answer; and obtaining the second text vector from the set of text vectors corresponding to the standard question and answer dataset for each type of question and answer based on the type of target question information.

[0094] Optionally, after feeding back the target question-and-answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold, the method further includes: updating the standard question-and-answer dataset and the historical question-and-answer dataset based on the target question information and the target question-and-answer data.

[0095] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0100] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0101] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0102] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0103] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0104] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A processing method for an intelligent question-answering system, characterized in that, include: The initial problem information of the target object is obtained, and the target network model is used to analyze and process the initial problem information to obtain the target problem information; Based on the first target weight table corresponding to the standard question and answer dataset of the intelligent question answering system and the second target weight table corresponding to the historical question and answer dataset of the target object, the target question information is converted into a first text vector. The first target weight table consists of words in the standard question and answer dataset and the weight values ​​corresponding to the words in the standard question and answer dataset, and the second target weight table consists of words in the historical question and answer dataset and the weight values ​​corresponding to the words in the historical question and answer dataset. The first text vector is compared with the second text vector corresponding to each question and answer data in the standard question and answer dataset to determine the highest similarity value; If the highest similarity value is greater than a preset threshold, the target question-and-answer data corresponding to the highest similarity value is fed back to the target object, wherein the target question-and-answer data is the question-and-answer data in the standard question-and-answer dataset; Before converting the target question information into a first text vector based on the first target weight table corresponding to the standard question-answer dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answer dataset of the target object, the method further includes: using a target classifier to classify the standard question-answer dataset and the historical question-answer dataset respectively to obtain a standard question-answer dataset for each type of question and a historical question-answer dataset for each type of question. Based on the weight table corresponding to the standard question and answer dataset for each type of question and answer and the vocabulary database corresponding to the standard question and answer dataset for each type of question and answer, the historical question and answer dataset for each type of question and answer is processed by the second preset algorithm to obtain the initial weight table of the historical question and answer data for each type of question and answer; By correcting the weight values ​​in the initial weight table using a Gaussian filter, the latest weight table corresponding to the historical question-and-answer dataset for each type of question-and-answer is obtained.

2. The method according to claim 1, characterized in that, Before obtaining the initial problem information of the target object and analyzing and processing the initial problem information using a target network model to obtain the target problem information, the method further includes: Obtain the standard question-and-answer dataset and perform data cleaning on the standard question-and-answer dataset to obtain the processed standard question-and-answer dataset; The processed standard question-and-answer dataset is input into the target classifier for classification to obtain the standard question-and-answer dataset for each category of questions and answers. The target network model is obtained by learning and training the initial neural network model using the standard question-answer dataset for each type of question answer.

3. The method according to claim 2, characterized in that, Before converting the target question information into a first text vector based on the first target weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answering dataset of the target object, the method further includes: The standard question-and-answer dataset for each type of question and answer is processed by a first preset algorithm to obtain a weight table and a vocabulary database corresponding to the standard question-and-answer dataset for each type of question and answer.

4. The method according to claim 3, characterized in that, Before converting the target question information into a first text vector based on the first weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and the second weight table corresponding to the historical question-answering dataset of the target object, the method further includes: Obtain the historical question and answer dataset, and perform data cleaning on the historical question and answer dataset to obtain the processed historical question and answer dataset; The processed historical question-and-answer dataset is classified using the target classifier to obtain historical question-and-answer datasets for each category of questions and answers. The historical question-and-answer datasets for each question-and-answer category are processed to obtain the latest weight table corresponding to the historical question-and-answer datasets for each question-and-answer category.

5. The method according to claim 1, characterized in that, Based on the first target weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answering dataset of the target object, the conversion of the target question information into a first text vector includes: Based on the type of the target question information, the first target weight table is obtained from the weight table corresponding to the standard question-answer dataset for each type of question and answer; Based on the type of the target question information, the second target weight table is obtained from the latest weight table corresponding to the historical question and answer dataset for each type of question and answer; Based on the first target weight table, the second target weight table, and the word library, the target question information is converted into a first text vector.

6. The method according to claim 5, characterized in that, Before calculating the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question-answer dataset, and determining the highest similarity value, the method further includes: The standard question-and-answer dataset for each type of question-and-answer is processed to obtain the text vector set corresponding to the standard question-and-answer dataset for each type of question-and-answer. Based on the type of the target question information, the second text vector is obtained from the text vector set corresponding to the standard question and answer dataset for each type of question and answer.

7. The method according to claim 1, characterized in that, After feeding back the target question-and-answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold, the method further includes: Based on the target question information and the target question-and-answer data, the standard question-and-answer dataset and the historical question-and-answer dataset are updated.

8. A processing device for an intelligent question-and-answer system, characterized in that, include: The first acquisition unit is used to acquire initial problem information of the target object and to analyze and process the initial problem information using a target network model to obtain target problem information. The conversion unit is used to convert the target question information into a first text vector based on a first target weight table corresponding to the standard question and answer dataset of the intelligent question answering system and a second target weight table corresponding to the historical question and answer dataset of the target object. The first target weight table consists of words in the standard question and answer dataset and the weight values ​​corresponding to the words in the standard question and answer dataset, and the second target weight table consists of words in the historical question and answer dataset and the weight values ​​corresponding to the words in the historical question and answer dataset. The calculation unit is used to calculate the similarity between the first text vector and the second text vector corresponding to each question and answer data in the standard question and answer dataset, and to determine the highest similarity value. A feedback unit is configured to feed back the target question-and-answer data corresponding to the highest similarity value to the target object if the highest similarity value is greater than a preset threshold, wherein the target question-and-answer data is the question-and-answer data in the standard question-and-answer dataset; Before converting the target question information into a first text vector based on the first target weight table corresponding to the standard question-answering dataset of the intelligent question-answering system and the second target weight table corresponding to the historical question-answering dataset of the target object, the device further includes: a classification unit, used to classify the standard question-answering dataset and the historical question-answering dataset respectively using a target classifier to obtain a standard question-answering dataset and a historical question-answering dataset for each type of question-answering; The device further includes: a processing module, configured to process the historical question-and-answer dataset of each type of question and answer using a second preset algorithm based on the weight table corresponding to the standard question-and-answer dataset of each type of question and answer and the vocabulary database corresponding to the standard question-and-answer dataset of each type of question and answer, to obtain an initial weight table of the historical question-and-answer data of each type of question and answer; and a correction module, configured to correct the weight values ​​in the initial weight table using a Gaussian filter to obtain the latest weight table corresponding to the historical question-and-answer dataset of each type of question and answer.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the storage medium is located to perform the processing method of the intelligent question-answering system according to any one of claims 1 to 7.

10. A processor, characterized in that, The processor is used to run a program, wherein the program executes the processing method of the intelligent question-answering system according to any one of claims 1 to 7.

Images