Query result acquisition method and device, storage medium, electronic device and program product

Abstract

The invention discloses a query result obtaining method and device, a storage medium, an electronic device and a program product. The method comprises the following steps: acquiring query information from a client; based on the query information, determining a to-be-queried medical database; a knowledge graph corresponding to the medical database is called, the knowledge graph is constructed based on multiple entities in the medical database and the incidence relation among the multiple entities, the entities are obtained by extracting the medical database, and nodes in the knowledge graph are used for representing the entities; a query result associated with the query information is obtained from the knowledge graph, the association degree of the query result and the query information is higher than an association degree threshold value, and the query result comprises at least one entity; and displaying the query result in a display interface of the client. The technical problem of low medical resource recommendation accuracy is solved.

Description

Technical Field

[0001] The present application relates to the field of medical information, and more specifically, to a method, device, storage medium, electronic device and program product for obtaining query results. Background Art

[0002] Medical resources are usually extracted through optical character recognition (OCR). However, traditional OCR methods have limited processing capabilities for complex backgrounds, uneven lighting, font deformation, etc., and there is a technical problem of low accuracy in recommending medical resources.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present application provide a method, device, storage medium, electronic device and program product for obtaining query results, so as to at least solve the technical problem of low accuracy in recommending medical resources.

[0005] According to one aspect of an embodiment of the present application, a method for obtaining query results is provided, and the method may include: obtaining query information from a client; determining a medical database to be queried based on the query information; retrieving a knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationships between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; obtaining a query result associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than a correlation threshold, and the query result contains at least one entity; and displaying the query result on a display interface of the client.

[0006] Optionally, based on the query information, determining the medical database to be queried includes: in response to obtaining the query information, extracting a network token from the query information; verifying the network token to obtain a verification result; based on the verification result, determining the reading permission of the client; and determining the medical database that the reading permission allows access to as the medical database to be queried.

[0007] Optionally, query results associated with the query information are obtained from the knowledge graph, including: obtaining access indicators of the client, wherein the access indicators are used to characterize the client's satisfaction with different historical query results; using the access indicators to guide the conversation analysis model to generate query results based on the knowledge graph.

[0008] Optionally, displaying the query result on a display interface of the client includes: converting text in the query result to obtain a conversion result; and displaying the conversion result on the display interface.

[0009] Optionally, the method may also include: classifying the medical database to obtain a first sub-database and a second sub-database, wherein the first medical data in the first sub-database and the second medical data in the second sub-database are different, and the medical database includes the first medical data and the second medical data; extracting the first medical data using a convolutional neural network to obtain first extracted information; extracting the second medical data using a recurrent neural network to obtain second extracted information; and constructing a knowledge graph based on the first extracted information and the second extracted information.

[0010] Optionally, using a convolutional neural network to extract the first medical data to obtain first extracted information includes: encoding the first medical data using a convolutional neural network to obtain encoded data; and recognizing characters in the encoded data to obtain first extracted information.

[0011] Optionally, based on the first extraction information and the second extraction information, a knowledge graph is constructed, including: determining a storage format of the first extraction information and the second extraction information; storing the first extraction information and the second extraction information according to the storage format to obtain a knowledge graph.

[0012] Optionally, the first extracted information and the second extracted information are semi-structured information.

[0013] Optionally, the method may further include: extracting characteristic information of the query information; determining information to be recommended that is associated with the characteristic information in the medical database; and displaying the information to be recommended in a display interface.

[0014] According to another aspect of an embodiment of the present application, a device for obtaining query results is also provided, and the device may include: a first acquisition unit, used to obtain query information from a client; a determination unit, used to determine a medical database to be queried based on the query information; a calling unit, used to call a knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; the acquisition unit, used to obtain a query result associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than a correlation threshold, and the query result contains at least one entity; a display unit, used to display the query result on a display interface of the client.

[0015] According to another aspect of an embodiment of the present application, a computer-readable storage medium is also provided, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the method for obtaining query results of the embodiment of the present application.

[0016] According to another aspect of an embodiment of the present application, a processor is further provided, the processor being used to run a program, wherein the program, when run by the processor, executes the method for obtaining query results of an embodiment of the present application.

[0017] According to another aspect of an embodiment of the present application, a program product is further provided. The program product includes computer instructions, wherein the computer instructions, when executed by a processor, implement the method for obtaining query results of an embodiment of the present application.

[0018] In an embodiment of the present application, query information is obtained from a client; based on the query information, a medical database to be queried is determined; a knowledge graph corresponding to the medical database is retrieved, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entity is extracted from the medical database, and the nodes in the knowledge graph are used to represent the entity; from the knowledge graph, a query result associated with the query information is obtained, wherein the correlation between the query result and the query information is higher than the correlation threshold, and the query result contains at least one entity; the query result is displayed in the display interface of the client. That is to say, in an embodiment of the present application, a knowledge graph is constructed based on the medical database in advance, and when the query information is obtained, the medical database to be queried is determined based on the query information, and the query result associated with the query information is obtained from the knowledge graph corresponding to the medical database, and the query result is displayed in the display interface of the client, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy of medical resource recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 is a flowchart of a method for obtaining query results according to an embodiment of the present application;

[0021] Figure 2 is a schematic diagram of a query result acquisition device according to an embodiment of the present application;

[0022] Figure 3 is a structural block diagram of a computer terminal according to an embodiment of the present application;

[0023] Figure 4 It is a block diagram of an electronic device according to a method for obtaining query results according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] According to an embodiment of the present application, an embodiment of a method for obtaining query results is provided, and the steps shown in the flowchart of 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 can be executed in an order different from that shown here.

[0027] In this embodiment, a method for obtaining query results is proposed. A knowledge graph can be constructed in advance based on a medical database. When query information is obtained, the medical database to be queried is determined based on the query information, and query results associated with the query information are obtained from the knowledge graph corresponding to the medical database. The query results are displayed in the display interface of the client, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy of medical resource recommendations.

[0028] Figure 1 FIG. 1 is a flow chart of a method for obtaining query results according to an embodiment of the present application. Figure 1 As shown, the method may include the following steps:

[0029] Step S102, obtaining query information from the client.

[0030] In the technical solution provided in the above step S102 of the present application, the above client can be a mobile terminal or a computer terminal, for example, it can be a user's mobile phone, tablet or other terminal. It should be noted that this is only an example, and there is no specific restriction on the type of client. The above query information can be multimodal information uploaded by the client, and can be voice, text, image, graphic information, etc., and can be used to query the popularization of medical knowledge, disease query, ancient book search, difficult and complicated disease query, etc. It should be noted that this is only an example, and there is no specific restriction on the query content and expression form of the query information.

[0031] Optionally, this embodiment utilizes online medical reading to provide medical information reading and query services through network technology, and can obtain query information from the client.

[0032] Step S104, determining the medical database to be queried based on the query information.

[0033] In the technical solution provided in the above step S104 of the present application, the medical database to be queried can be determined based on the query information. The above medical database may include multimodal medical data, may include information such as images, text, and voice, may be used to store medical knowledge, may include multiple medical data (hereinafter referred to as data), may include publicly available medical literature, medical reports, clinical guidelines, and medical classics, etc. It should be noted that this is only for example, and there is no specific restriction on the type of medical database.

[0034] Optionally, in response to obtaining the query information, the medical database to be queried can be determined based on the network token in the query information, or the medical database to be queried can be determined by judging the query information. For example, if the query information is "determine the cause of heart disease", it can be determined that the medical database to be queried is a medical database related to heart disease. It should be noted that this is only an example, and there is no specific limitation on the method of determining the medical database to be queried.

[0035] Step S106, retrieve the knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities.

[0036] In the technical solution provided in step S106 of the present application, the above-mentioned entity can be a person's name, a place name, an organization name, etc., which can be obtained by extracting text information in the medical database. The knowledge graph can include nodes and edges. Nodes can be used to represent entities, and edges can be used to represent the relationship between multiple entities.

[0037] Optionally, a knowledge graph can be constructed in advance based on different medical databases. After the medical database to be queried is determined, the knowledge graph corresponding to the medical database can be retrieved, and the knowledge graph can be processed to determine the query result corresponding to the query information.

[0038] Optionally, this embodiment may associate the medical database and the knowledge graph through identification, or may associate the medical database and the knowledge graph through methods such as embedding points. It should be noted that this is only an example, and there is no specific limitation on the method of associating the medical database and the knowledge graph.

[0039] For example, a large amount of text data in a medical database can be processed, including word segmentation, part-of-speech tagging, syntactic analysis and other data, in order to extract useful information from the text. Entity recognition technology is used to identify entities in the text, such as people, places, organizations, etc. Relationship extraction technology is used to extract relationships between entities from the text. For example, the relationships between people mentioned in the text (such as work relationships, family relationships, etc.) are identified and represented as edges in the knowledge graph. The extracted entities and relationships are represented as nodes and edges in the knowledge graph to construct a knowledge graph. After the medical database to be queried is determined, the knowledge graph built based on the medical database can be retrieved.

[0040] Step S108, obtaining a query result associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than a correlation threshold, and the query result contains at least one entity.

[0041] In the technical solution provided in the above step S108 of the present application, the knowledge graph can be searched to determine the query results associated with the query information. The query results can be medical knowledge and ancient literature and works related to the query information, the correlation between the query results and the query information is higher than the correlation threshold, and the query results contain at least one entity.

[0042] For example, by obtaining the query information "causes of rheumatism" from the client, query results related to the causes of rheumatism can be obtained, such as genetic causes, climatic causes, etc. of rheumatism.

[0043] In this embodiment, the user can input query information (for example, relevant symptoms or disease names), and the corresponding query results (that is, medical consultation answers) can be generated based on the query information input by the user and the existing medical resource library (for example, medical knowledge and data), and the source of medical knowledge can be located to enable online reading, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy in medical resource recommendations.

[0044] Step S110, displaying the query result on the display interface of the client.

[0045] In the technical solution provided in the above step S110 of the present application, after the query result is obtained through the search, the query result can be displayed to the user to provide the user with online reading and browsing. The above display interface can be a display screen of the client.

[0046] Furthermore, the client may adjust, evaluate, regenerate, etc. the query results in the display interface, and may further improve the query results in response to instructions issued by the client.

[0047] Through the above steps S102 and S108 of the present application, query information from the client is obtained; based on the query information, the medical database to be queried is determined; the knowledge graph corresponding to the medical database is retrieved, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entity is extracted from the medical database, and the nodes in the knowledge graph are used to represent the entity; from the knowledge graph, the query result associated with the query information is obtained, wherein the correlation between the query result and the query information is higher than the correlation threshold, and the query result contains at least one entity; the query result is displayed in the display interface of the client. That is to say, in the embodiment of the present application, the knowledge graph is constructed based on the medical database in advance, and when the query information is obtained, the medical database to be queried is determined based on the query information, and the query result associated with the query information is obtained from the knowledge graph corresponding to the medical database, and the query result is displayed in the display interface of the client, thereby achieving the technical effect of improving the accuracy of medical resource recommendation and solving the technical problem of low accuracy of medical resource recommendation.

[0048] The above method of this embodiment is further introduced below.

[0049] As an optional implementation, step S104, based on the query information, determines the medical database to be queried, including: in response to obtaining the query information, extracting a network token from the query information; verifying the network token to obtain a verification result; based on the verification result, determining the reading permission of the client; and determining the medical database that the reading permission allows access to as the medical database to be queried.

[0050] In this embodiment, if query information from the client is obtained, a network token can be extracted from the query information. The network token can be (JSON Web Token, referred to as JWT Token), which can be used to verify the identity of the client, the access rights of the client, and other information. Further, the network token can be verified to obtain a verification result. Based on the verification result, the reading permission of the client can be determined. For example, it can be used to determine the medical database that the client is allowed to access. The medical database that the reading permission allows access to can be determined as the medical database to be queried.

[0051] Optionally, the use of knowledge graph data is managed through user identity identification and permission management.

[0052] Optionally, regarding front-end login authentication, the JWT verification method can be used. This verification method does not require the server to save the response information, and there will be no inconvenience in session extension. The load in JWT (for example, PayLoad) can store common information for information exchange. By using JWT, the purpose of reducing the number of times the server queries the database can be achieved. JWT can identify the user's identity and determine the user's reading authority based on the user's identity and authorization status. For example, it can be used to determine whether the user has the right to read e-books.

[0053] For example, the client can send a request containing a JWT to the server. After receiving the request, the server extracts the JWT and decodes it. The client can verify the decoded JWT to verify whether the signature of the JWT is valid. If the signature verification passes, the server can check the statements in the JWT (such as expiration time, issuer, etc.) to ensure that the JWT is within the validity period and is issued by a trusted issuer. Furthermore, if the JWT verification passes, the server allows the client to access the corresponding medical resource library. Otherwise, the server denies the client access to the corresponding medical resource library and returns a corresponding error message.

[0054] In this embodiment, when obtaining query information, the JWT network token in the query information can be extracted, and the JWT network token can be verified to determine the medical database that the client is allowed to access. Therefore, when determining the query result, it can be determined only in the medical database that the client is allowed to access, avoiding the search of a large range of data, thereby achieving the purpose of improving data processing efficiency.

[0055] As an optional implementation, step S108, obtaining query results associated with the query information from the knowledge graph, including: obtaining the client's access indicators, wherein the access indicators are used to characterize the client's satisfaction with different historical query results; using the access indicators to guide the conversation analysis model to generate query results based on the knowledge graph.

[0056] In this embodiment, the above-mentioned access indicators can be used to measure the quality of session access, and can be indicators of session access quality, for example, they can be indicators such as number of visits, average number of visits per person, total visit time, single visit time, single visit depth, number of bounces, bounce rate, number of exits, exit rate, average visit time per person, total page dwell time, average page dwell time, etc. It should be noted that this is only for example purposes, and there is no specific limitation on the type of access indicators.

[0057] Optionally, the access index of the client is obtained. The access index can be used to determine the client's satisfaction with different historical query results. According to the access index, the conversation analysis model can be guided to generate query results based on the knowledge graph. That is, through the access index, the conversation analysis model can be guided to generate query results with high user satisfaction. Among them, the above-mentioned session analysis model can be used to recommend query results to the client (that is, the user), such as web pages, reading materials, books, papers and other information related to the query information.

[0058] For example, a data recommendation system can be used to perform data analysis and recommendations. The data recommendation system may include content analysis, user tags, evaluation analysis, content security, etc. The data recommendation system collects users' reading behaviors, such as reading time, reading speed, reading mode, etc. A conversation analysis model may be used in the data recommendation system to recommend relevant reading materials to users. The conversation analysis model may include a variety of indicators for measuring session access quality, such as the number of visits, the average number of visits per person, the total visit time, the single visit time, the single visit depth, the number of bounces, the bounce rate, the number of exits, the exit rate, the average visit time per person, the total page dwell time, the average page dwell time, and other indicators.

[0059] As an optional implementation, step S110, displaying the query result on the display interface of the client, includes: converting the text in the query result to obtain a conversion result; and displaying the conversion result on the display interface.

[0060] In this embodiment, the text in the query results can be converted into images to obtain conversion results, and the conversion results can be displayed in the display interface. By converting the text in the query results into images, the problem of content in the display interface being selected or copied can be avoided, thereby ensuring the security of the online reading platform, preventing the leakage of users' personal information, and preventing the illegal copying and dissemination of e-book content.

[0061] Optionally, in addition to converting the text in the query results, you can also use Cascading Style Sheets (CSS) style sheet control and scripting language (JavaScript) technology to prohibit operations such as right-click copying to prevent page content from being selected and copied. There is no specific restriction on the protection methods and measures here.

[0062] As an optional implementation, the method may further include: classifying the medical database to obtain a first sub-database and a second sub-database, wherein the first medical data in the first sub-database and the second medical data in the second sub-database are different, and the medical database includes the first medical data and the second medical data; extracting the first medical data using a convolutional neural network to obtain first extracted information; extracting the second medical data using a recurrent neural network to obtain second extracted information; and constructing a knowledge graph based on the first extracted information and the second extracted information.

[0063] In this embodiment, the first sub-database may include images, texts, videos and other contents, and may be small-scale (e.g., small feature information volume) medical data, such as data with only a small amount of text information, independent text information without contextual connection, or graphic information with obvious features. The second sub-database may also include images, texts, videos and other contents, and may be medical data containing text line information, serialized data, and contextual connection or fuzzy graphic information. The types of medical data in the first sub-database and the second sub-database are different. It should be noted that this is only for example, and there is no specific limitation on the type of the first sub-database and the type of the second sub-database.

[0064] Optionally, the above-mentioned convolutional neural network can be used to construct an optical character recognition based on convolutional neural networks (Convolut ionalNeural Networks, referred to as CNNs), which may include convolutional layers, pooling layers and fully connected layers, and can be used to learn the spatial hierarchical features of the first medical data (i.e., input data), which can be trained based on the handwritten digit recognition (MNIST) data set. The above-mentioned MNIST data set may contain images of handwritten digits. The above-mentioned recurrent neural network can be used to construct an optical character recognition based on a recurrent neural network (Recurrent Neural Network-based OpticalCharacter Recognit ion, referred to as RNN), and the above-mentioned first medical data / second medical data can be images, texts and other materials, and can be text materials related to medicine. The above-mentioned first extracted information can be used to characterize the content and information in the first medical data, and can include entities in the first medical data, and the associations between entities. The above-mentioned second extracted information can be used to characterize the content and information in the second medical data, and can include entities in the second medical data, and the associations between entities.

[0065] In order to improve the accuracy of information extraction from medical data, in this embodiment, the medical data in the medical database are classified to obtain different types of first sub-databases and second sub-databases. For different types of sub-databases, optical character recognition based on convolutional neural networks or optical character recognition based on recurrent neural networks can be used to extract text features from the medical data.

[0066] In this embodiment, for data search, browsing and positioning, it is necessary to collect basic information such as search keywords or search features. On this basis, for the collection of graphic information, the features in the image can be automatically extracted based on a convolutional neural network (CNN) or a recurrent neural network (RNN), thereby achieving the purpose of extracting features in the image without manually designing a feature extraction algorithm.

[0067] Optionally, for medical data with small scale or small feature information, such as only a small amount of text information, independent text information without context, and graphic information with obvious features, the convolutional neural network-based optical character recognition technology (CNN-based OCR) can be used to extract features. For text line information, serialized data, and feature information with vague context or graphic information, the recurrent neural network-based optical character recognition technology (RNN-based OCR) can be used to recognize text features.

[0068] For example, for the construction of convolutional neural networks, it is assumed that CNNs can be constructed through convolutional layers, pooling layers, and fully connected layers. When constructing CNNs, CNNs can be constructed using high-level programming languages ​​(Python) and open source deep learning libraries (Keras). Assuming that the final CNNs are constructed using the MNIST dataset, the dataset may contain images of handwritten digits. Further, the convolutional neural network can be constructed by the following steps: installing tensorflow and keras libraries, setting the corresponding hyperparameters, inputting the image size after the settings are completed, the image size needs to be made dynamic, and according to the changes in the feature information uploaded by the user, the received image information is loaded and preprocessed as data, and the labels are converted to one-hot encoding after processing. After the encoding conversion is successful, the convolutional neural network can be constructed, and then the convolutional neural network can be trained and evaluated using the MNIST dataset to obtain the final convolutional neural network.

[0069] In this embodiment, a CNN is created that includes two convolutional layers, a maximum pooling layer, two hidden layers (Dropout) for preventing overfitting, and two fully connected layers. Among them, the convolutional layer can be used to extract image features, the pooling layer can be used to reduce the dimension of the feature map, the Dropout layer can randomly discard the output of a part of the neurons, and the fully connected layer can be used for classification tasks. It should be noted that this is only an example, and the method of constructing CNNs, the structure of CNNs, and the process of training CNNs can adjust the structure and parameters of the model according to actual conditions, and no specific restrictions are made here. After training the CNNs, the trained CNNs can be used to extract text features in medical data.

[0070] In this embodiment, since RNN can process sequence data, RNN can be used to identify text lines. In this embodiment, RNN-based OCR can be used to process sequence data (e.g., text lines). At the same time, the RNN model can memorize contextual information, so it can also process texts in some connected and cursive situations. For example, it can search for offline doctor's handwritten prescription information, or analyze data query and prescription and medication. Among them, the above-mentioned second sub-database may include medical data containing text lines, or may include offline doctor's handwritten prescription information, prescription and medication information, etc. It should be noted that this is only an example, and there is no specific restriction on the type of the second medical data in the second sub-database.

[0071] Optionally, RNN-based OCR can be trained in advance. During the training of RNN-based OCR, the connectionist temporal classification (CTC) loss function and the network layer (Time Distributed) in the RNN that processes sequence data can be used to generate an output for each time step. Since the CTC loss function is used, there is no need to align the sequences during training, and the RNN can learn how to map the input sequence to the output label sequence by itself. At the same time, the placeholders in the code can be replaced by a data generator (your_data_generator), which can generate appropriate input data and labels, as well as the length of the input and labels. Furthermore, the trained RNN-based OCR can be used to extract the second medical data in the second sub-database to obtain the second extracted information.

[0072] It should be noted that, in this embodiment, in addition to using RNN-based OCR and CNNs-based OCR to identify medical data, manual verification of input information can also be used instead of OCR technology to achieve the purpose of combining input information with artificial intelligence.

[0073] As an optional implementation, a convolutional neural network is used to extract the first medical data to obtain first extracted information, including: encoding the first medical data using a convolutional neural network to obtain encoded data; and recognizing characters in the encoded data to obtain first extracted information.

[0074] In this embodiment, extracting the first medical data using a convolutional neural network may include the following steps: encoding the first medical data using a neural network to obtain encoded data, and using a classifier to recognize characters in the encoded data to obtain first extracted information. Among them, the above-mentioned classifier can be a text processing layer (IncepText), which can be used to implement OCR methods for large-scale, aspect ratio and direction change problems. The above-mentioned classifier uses convolution kernels of different sizes to detect texts of different sizes and aspect ratios. At the same time, deformable convolution layer operations and special pooling layers (deformable PSROI pooling) are introduced into the convolutional neural network to improve the detection effect of texts in any direction. In this embodiment, by constructing a CNN-based OCR, the purpose of automatically extracting features in the image and improving the processing speed is achieved.

[0075] For example, when the first medical data is an image, a convolutional neural network can be used to encode the image, and then a classifier can be used for character recognition. Among them, the classifier can be a text processing layer (IncepText), which can be used for OCR methods for large scale, aspect ratio and direction change problems. The inception module in GoogLeNet is used to solve these problems. The inception structure uses convolution kernels of different sizes to detect text of different sizes and aspect ratios, and introduces deformable convolution layer operations and deformable PSROI pooling layers (special pooling layers) to improve the detection effect of text in any direction. The advantage of the CNN-based OCR method is that it can automatically extract features from images and has a fast processing speed.

[0076] As an optional implementation, a knowledge graph is constructed based on the first extraction information and the second extraction information, including: determining the storage format of the first extraction information and the second extraction information; storing the first extraction information and the second extraction information according to the storage format to obtain a knowledge graph.

[0077] In this embodiment, the first medical data is extracted using a convolutional neural network to obtain first extracted information; the second medical data is extracted using a recurrent neural network to obtain second extracted information. Further, the first extracted information and the second extracted information can be processed to construct a knowledge graph for subsequent artificial intelligence (AI) retrieval. The above storage format may include: planning data type, field length, listening address, port, authentication mechanism, table structure, index, view, stored procedure, etc. It should be noted that this is only an example, and there is no specific restriction on the type of storage format.

[0078] Optionally, the data storage format of the database can be set in advance, wherein the storage format may include table structure, index, view, stored procedure and other formats. It should be noted that this is only an example and there is no specific restriction on the type of storage format. After the first extraction information and the second extraction information are obtained, the storage formats of the first extraction information and the second extraction information can be determined respectively, and the first extraction information and the second extraction information can be stored according to the pre-set storage format to obtain a knowledge graph.

[0079] For example, the information to be searched (ie, medical data) can be stored for repeated retrieval during subsequent search queries. At the same time, when it is necessary to search the collected information, a knowledge base can be established, which can be used to store the knowledge graph.

[0080] In this embodiment, the first extracted information and the second extracted information can be extracted from medical databases such as publicly available medical literature, medical reports, clinical guidelines, and medical classics, and the first extracted information and the second extracted information can be processed to construct a knowledge graph for subsequent AI retrieval.

[0081] Optionally, the collected data (i.e., the first extracted information and the second extracted information) are analyzed by data processing and analysis technology (e.g., Python, NoSQL, GrowingIO, etc.). For storage, an intelligent base (IBase for short) database can be used, which is fully compatible with various large, medium and small databases, and can provide import and link support capabilities for traditional relational databases (e.g., Oracle, Sybase, SQLServer, DB2, Informix).

[0082] In this embodiment, for the construction of the IBase database, the IBase installation program (for example, an ISO image or a set of installation files) can be obtained, the installation program can be run on the server, and the installation process can be completed according to the prompts, which may include selecting the installation type (such as client, server, etc.), specifying the installation directory, setting system parameters, etc. After the installation is complete, configure the database instance, which includes creating a database, setting the listening address, port, authentication mechanism, etc. Design the database schema, including table structure, indexes, views, stored procedures, etc. Plan data types and field lengths to meet data storage requirements. After the first extraction information and the second extraction information are obtained, the storage formats corresponding to the first extraction information and the second extraction information can be determined respectively, and the first extraction information and the second extraction information can be stored according to the storage format to obtain a knowledge graph, and the knowledge graph is stored in a pre-built database.

[0083] Optionally, after building the IBase database, you can use Structured Query Language (SQL) statements or graphical interface tools to create and fill data tables, and perform data backup and recovery operations to ensure data security. Monitor database performance, analyze slow queries, and optimize SQL statements. Adjust database parameters, such as cache size, concurrency control, etc., to improve performance. Set user permissions to limit access to sensitive data. Implement security policies such as auditing, encryption, and firewalls. Perform maintenance tasks regularly, such as rebuilding indexes and updating statistics. Use database monitoring tools to monitor database status in real time and prevent potential problems.

[0084] In this embodiment, after the knowledge graph is constructed, the following steps can be performed to supplement and improve the knowledge graph: events in the text, such as acquisitions, elections, disasters, etc., can be identified through dependency syntactic analysis, semantic role labeling (SRL) or a special relationship extraction model, and the participants and roles of the event can be determined. The identified entities are linked to corresponding entities in the existing knowledge base, such as Wikipedia entries. The problem of entities with the same name is solved to ensure the correct identification of the real-world reference of the entity. Information from different sources is integrated to resolve conflicts and enrich entity descriptions. Implicit knowledge is inferred through inference rules or machine learning models to increase the integrity and coherence of the graph, define the hierarchy of concepts and relationships, create an ontology model, and construct the extracted entities, relationships and attributes into a graph data structure, such as using triples (e.g., RDF), store the knowledge graph, and provide a query interface or query language (e.g., SPARQL) so that users or applications can retrieve and use the information in the knowledge graph. Information is extracted from new data sources regularly to update the knowledge graph. Implement quality assurance measures, such as manual review, to ensure the accuracy and timeliness of the maps.

[0085] As an optional implementation manner, the first extracted information and the second extracted information are semi-structured information.

[0086] In this embodiment, in order to facilitate query statistics, the data storage management method can adopt semi-structured data. Among them, semi-structured data is data first and then structure, which can be data between completely structured data (such as data in relational databases and object-oriented databases) and completely unstructured data (such as sound, image files, etc.), can be in the form of Hypertext Markup Language (HTML) documents, can be self-describing, and the structure and content of the data are mixed together without obvious distinction, and can be expressed in the form of trees or graphs.

[0087] Optionally, semi-structured data is used to facilitate the construction of the knowledge graph.

[0088] As an optional implementation, the method may further include: extracting characteristic information of the query information; determining the information to be recommended associated with the characteristic information in the medical database; and displaying the information to be recommended in a display interface.

[0089] In this embodiment, characteristic information of the query information can be extracted, and based on the characteristic information, information to be recommended that is associated with the characteristic information can be retrieved from the medical database; and the information to be recommended is displayed in the display interface, wherein the characteristic information can be used to characterize the content of the query information, and can be a binary code, a word, or other code, and the form of expression of the characteristic information is not specifically limited here. The information to be recommended can be information associated with the query information, and can be information that the client may be interested in. For example, if the query information is "the cause of rheumatism", the information to be recommended can be "the treatment method of rheumatism", "the symptoms of rheumatism", etc.

[0090] Optionally, feature information of the query information is extracted, an algorithm is adopted according to the provided search features, and the extracted feature information is used to determine the information to be recommended. The information to be recommended can be displayed in a display interface for selection by the client.

[0091] Optionally, in order to adapt to the small Internet medical clinic system and reduce the size of the artificial intelligence dialogue model, an application library (for example, Gradio application) can be used, and the front-end and back-end separation method can be used to facilitate data processing, display and use. The web front-end page and the back-end database are used to build an easy-to-use and feature-rich online medical reading service. This module can be embedded in the existing medical service system, and the module size is reduced through page routing control. According to the graphic information uploaded by the user, relevant medical knowledge and ancient literature and works are obtained. At the same time, the algorithm is used to recommend information that the user may be interested in based on the provided search features.

[0092] In an embodiment of the present application, a knowledge graph is constructed in advance based on a medical database. When query information is obtained, the medical database to be queried is determined based on the query information, and query results associated with the query information are obtained from the knowledge graph corresponding to the medical database. The query results are displayed in the display interface of the client, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy in medical resource recommendations.

[0093] The technical solution of the embodiment of the present application is illustrated below in conjunction with preferred implementation modes.

[0094] At present, traditional OCR methods are usually based on image processing and statistical machine learning technology. However, traditional OCR methods have limited processing capabilities for complex backgrounds, uneven lighting, font deformation, etc., the recognition rate needs to be improved, and there is a technical problem of low accuracy in recommending medical resources.

[0095] To solve the above problems, in this embodiment, a method for realizing online medical reading of an Internet clinic system based on big data and artificial intelligence technology is proposed. The method adopts big data processing, artificial intelligence technology and deep learning, such as convolutional neural network or recurrent neural network. The features in the image are automatically extracted through the above two neural networks without the need to manually design feature extraction algorithms, thereby greatly improving the recognition rate and matching speed of images, texts, etc. At the same time, the information recognized by the above two neural networks is combined with natural language processing (Natural Language Processing, referred to as NLP) technology through AI algorithm, and the search features are further accurately extracted using AI algorithm, thereby expanding the search breadth and depth of medical knowledge, thereby achieving the purpose of improving the accuracy of search information.

[0096] Optionally, the above-mentioned online medical reading mainly refers to providing medical information reading and query services through network technology, including popularization of medical knowledge, disease query, search for prescriptions in ancient books, query of difficult and complicated diseases, etc. The above-mentioned query services can be based on big data and artificial intelligence technology, and can provide personalized medical knowledge consultation and medical history development, greatly improving the efficiency and quality of medical services.

[0097] In this embodiment, the user can enter relevant symptoms or disease names. The AI ​​algorithm model can generate corresponding medical consultation answers based on the information input by the user and existing medical knowledge and data, and locate the source of medical knowledge to enable online reading, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy in medical resource recommendations.

[0098] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments; the size of the numbers in the embodiments is only for illustration and is not specifically limited here. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application and are not specifically limited here.

[0099] As an optional embodiment, features in the image are extracted using a convolutional neural network or a recurrent neural network.

[0100] In this embodiment, for data search, browsing and positioning, it is necessary to collect basic information such as search keywords or search features. On this basis, for the collection of graphic information, the features in the image can be automatically extracted based on a convolutional neural network or a recurrent neural network, thereby achieving the purpose of extracting features in the image without manually designing a feature extraction algorithm.

[0101] In this embodiment, for small-scale medical data, that is, medical data with a small amount of feature information, such as only a small amount of text information, independent text information without context, graphic information with obvious features, etc., a CNN-based OCR method can be used to extract features.

[0102] Optionally, CNN-based OCR may include the following steps to extract features from an image: a convolutional neural network may be used to encode the image, and then a classifier may be used to perform character recognition. The classifier may be a text processing layer, which may be used to implement an OCR method for large-scale, aspect ratio, and directional change problems. The classifier uses convolution kernels of different sizes to detect texts of different sizes and aspect ratios. At the same time, a deformable convolution layer operation and a special pooling layer are introduced into the convolutional neural network to enhance the detection effect of texts in any direction. In this embodiment, by constructing a CNN-based OCR, the purpose of automatically extracting features from an image and improving the processing speed is achieved.

[0103] For example, for the construction of convolutional neural networks, it is assumed that CNNs can be constructed through convolutional layers, pooling layers, and fully connected layers. When constructing CNNs, CNNs can be constructed using high-level programming languages ​​and open source deep learning libraries. Assuming that the final CNNs are constructed using the MNIST dataset, the dataset may contain images of handwritten digits. Further, the convolutional neural network can be constructed by the following steps: install tensorflow and keras libraries, set the corresponding hyperparameters, and enter the image size after setting. The image size needs to be made dynamic. According to the changes in the feature information uploaded by the user, the received image information is loaded and preprocessed as data. After processing, the label is converted to a one-hot encoding. After the encoding conversion is successful, the convolutional neural network can be constructed. After that, the convolutional neural network can be trained and evaluated using the MNIST dataset to obtain the final convolutional neural network.

[0104] In this embodiment, a CNN is created that includes two convolutional layers, a maximum pooling layer, two hidden layers for preventing overfitting, and two fully connected layers. Among them, the convolutional layer can be used to extract image features, the pooling layer can be used to reduce the dimension of the feature map, the Dropout layer can randomly discard the output of a part of the neurons, and the fully connected layer can be used for classification tasks. It should be noted that this is only an example, and the method of constructing CNNs, the structure of CNNs, and the process of training CNNs can adjust the structure and parameters of the model according to actual conditions, and no specific restrictions are made here. After training the CNNs, the trained CNNs can be used to extract text features in medical data.

[0105] Optionally, when constructing the knowledge graph corresponding to the medical database, the trained convolutional neural network can be used to process the first medical data in the first sub-database in the medical database to obtain the first extracted information.

[0106] In this embodiment, for text line information, serialized data, and contextual information or ambiguous feature information of text and image information, an RNN-based OCR method may be used, which uses a recurrent neural network to recognize text lines.

[0107] Since RNN can process sequence data, it can be used to recognize text lines. In this embodiment, RNN-based OCR is used to process sequence data (e.g., text lines). At the same time, the RNN model can remember context information, so it can also process texts in some cases such as cursive writing, etc. For example, it can retrieve offline doctor's handwritten prescription information, or analyze data query and prescription and medication.

[0108] In this embodiment, during the training of RNN-based, the connection time classification loss function and the network layer in the RNN that processes sequence data can be used to generate an output for each time step. Since the CTC loss function is used, there is no need to align the sequence during training, and the RNN can learn how to map the input sequence to the output label sequence by itself. At the same time, the placeholders in the code can be replaced by a data generator that can generate appropriate input data and labels, as well as the length of the input and label.

[0109] As an optional embodiment, after processing the graphic information, the information to be searched can be stored for repeated retrieval during subsequent search queries. At the same time, a knowledge base needs to be established to search and query the collected information.

[0110] In this embodiment, data can be extracted from publicly available medical literature, medical reports, clinical guidelines, and medical books, and file and data formatting can be achieved to complete the collection of knowledge data. Furthermore, the collected data can be processed to construct a knowledge graph for subsequent AI retrieval.

[0111] Optionally, the collected data (i.e., the first extracted information and the second extracted information) are analyzed by data processing and analysis technology. For storage, an intelligent basic database can be used, which is fully compatible with various large, medium and small databases and can provide import and link support capabilities for traditional relational databases (e.g., Oracle, Sybase, SQLServer, DB2, Informix).

[0112] In this embodiment, for the construction of the IBase database, the IBase installation program (e.g., ISO image or a set of installation files) can be obtained, the installation program can be run on the server, and the installation process can be completed according to the prompts, which may include selecting the installation type, specifying the installation directory, setting system parameters, etc. After the installation is completed, the database instance is configured, which includes creating a database, setting the listening address, port, authentication mechanism, etc. The database schema is designed, including table structure, index, view, stored procedure, etc. The data type and field length are planned to meet the needs of data storage.

[0113] Optionally, after building the IBase database, you can use structured query language statements or graphical interface tools to create and fill data tables, and perform data backup and recovery operations to ensure data security. Monitor database performance, analyze slow queries, and optimize SQL statements. Adjust database parameters, such as cache size, concurrency control, etc., to improve performance. Set user permissions to limit access to sensitive data. Implement security policies such as auditing, encryption, firewalls, etc. Perform maintenance tasks regularly, such as rebuilding indexes, updating statistics, etc. Use database monitoring tools to monitor database status in real time and prevent potential problems.

[0114] As an optional embodiment, after the data is collected and processed, an AI retrieval algorithm can be reconstructed based on the knowledge graph.

[0115] In this embodiment, in order to facilitate query statistics, the data storage management method can adopt semi-structured data. Among them, semi-structured data is data first and then structure, which can be data between completely structured data (such as data in relational databases and object-oriented databases) and completely unstructured data (such as sound and image files, etc.), can be in the form of hypertext markup language documents, can be self-describing, and the structure and content of the data are mixed together without obvious distinction, and can be expressed in the form of trees or graphs.

[0116] Optionally, using more semi-structured data can help build the knowledge graph.

[0117] As an optional embodiment, after the data collection and processing is completed, the AI ​​retrieval algorithm can be reconstructed based on the knowledge graph.

[0118] In this embodiment, natural language processing technology can be used to achieve accurate translation between computer and human language. The data in the database can be "translated" into multi-dimensional, deep structured and semi-structured data for processing to form a searchable and understandable knowledge graph.

[0119] Optionally, the process of building a knowledge graph may include text preprocessing, text representation, analysis, and modeling.

[0120] Optionally, obtain a medical resource library (for example, public medical literature, medical reports, clinical guidelines, medical books and other materials), clean the data in the medical resource library, remove noise, and process the standardized text (for example, such as word segmentation, stop word removal, stemming or word form restoration, etc.). Machine learning models, such as conditional random fields, RNNs, CNNs and other models, can be used to extract entities in the text in the medical resource library, and determine the relationship between multiple entities (for example, "works at", "located in", "belongs to", etc.), thereby obtaining first extracted information and second extracted information. The first extracted information and the second extracted information may respectively include multiple entities, and the relationship between entities.

[0121] For example, events in text, such as acquisitions, elections, disasters, etc., can be identified through dependency parsing, semantic role labeling (SRL), or specialized relationship extraction models, and the participants and roles of the events can be determined. Link the identified entities to corresponding entities in existing knowledge bases, such as Wikipedia entries. Solve the problem of entities with the same name and ensure that the real-world references of entities are correctly identified. Integrate information from different sources, resolve conflicts, and enrich entity descriptions. Infer implicit knowledge through inference rules or machine learning models, increase the completeness and coherence of the graph, define the hierarchy of concepts and relationships, create an ontology model, and construct the extracted entities, relationships, and attributes into a graph data structure, such as using triples (for example, RDF), store the knowledge graph, and provide a query interface or query language (for example, SPARQL) so that users or applications can retrieve and use the information in the knowledge graph. Extract information from new data sources regularly and update the knowledge graph. Implement quality assurance measures, such as manual review, to ensure the accuracy and timeliness of the graph.

[0122] Optionally, the process of constructing a knowledge graph using NLP technology may include steps such as text processing, entity recognition, relationship extraction, and knowledge representation. The following is an example process: a large amount of text data is processed using NLP technology, including word segmentation, part-of-speech tagging, syntactic analysis, etc., in order to extract useful information from the text. Entity recognition technology (e.g., RNN, CNNs, etc. models) is used to identify entities in the text, such as people, places, organizations, etc. This step helps to identify important information in the text and provide entity nodes for the construction of the knowledge graph. Relationship extraction technology is used to extract relationships between entities from the text. For example, the relationships between people mentioned in the text (such as work relationships, family relationships, etc.) are identified and represented as edges in the graph. The extracted entities and relationships are represented as nodes and edges in the graph to construct a knowledge graph. Among them, the nodes in the knowledge graph represent entities, and the edges represent the relationships between entities. By converting text information into a graph structure, the association relationship between entities and the knowledge organization can be more intuitively displayed.

[0123] For example, suppose you need to build a knowledge graph in the medical field. First, you can use RNN-based OCR or CNN-based OCR to extract a large amount of text data from medical literature, and use NLP technology to extract entities such as diseases, drugs, symptoms, and identify the relationships between them. For example, you can identify that "virus" is a disease and is associated with "fever". Then, these entities and relationships are represented as nodes and edges in the knowledge graph to construct a medical knowledge graph.

[0124] As an optional embodiment, the use of knowledge graph data can be managed through user identity identification and permission management.

[0125] In this embodiment, regarding the front-end login authentication, the JWT verification method can be used. This verification method does not need to save the reply information on the server side, and there will be no inconvenience in Session expansion. The load in JWT (for example, PayLoad) can store commonly used information for information exchange.

[0126] In this embodiment, by using JWT, the purpose of reducing the number of times the server queries the database is achieved. JWT can identify the identity of the user and determine the user's reading authority based on the user's identity and authorization status. For example, it can be used to determine whether the user has the right to read e-books.

[0127] As an optional embodiment, after the database and search method are established, an AI visual interaction method can be used to feed back the search results to the client.

[0128] In this embodiment, in order to adapt to the small Internet medical clinic system and reduce the size of the artificial intelligence dialogue model, an application library (for example, Gradio application) can be used, and the front-end and back-end separation method can be used to facilitate data processing, display and use. The front-end page of the web page is used in conjunction with the back-end database to build an easy-to-use and feature-rich online medical reading service. This module can be embedded in the existing medical service system, and the module size is reduced through page routing control. According to the graphic information uploaded by the user, the relevant medical knowledge and literature and ancient books are obtained. At the same time, according to the provided search features, an algorithm is used to recommend information that the user may be interested in.

[0129] Optionally, after the search results are obtained, they can be displayed to the user and provided for online reading and browsing.

[0130] For example, a data recommendation system can be used to perform data analysis and recommendations. The data recommendation system can include content analysis, user tags, evaluation analysis, content security, etc. The data recommendation system collects users' reading behaviors, such as reading time, reading speed, reading mode, etc., and uses a conversation analysis model to recommend relevant reading materials to users. The conversation analysis model can include a variety of indicators for measuring session access quality, such as the number of visits, the average number of visits per person, the total visit time, the single visit time, the single visit depth, the number of bounces, the bounce rate, the number of exits, the exit rate, the average visit time per person, the total page dwell time, the average page dwell time, and other indicators.

[0131] As an optional embodiment, a personal security protection module is used to ensure the security of the online reading platform, prevent the leakage of users' personal information, and prevent the illegal copying and dissemination of e-book content.

[0132] In this embodiment, the text in the query results can be converted into images to obtain conversion results, and the conversion results can be displayed in the display interface. By converting the text in the query results into images, the problem of content in the display interface being selected or copied can be avoided.

[0133] Optionally, in addition to converting the text in the query results, you can also use cascading styles, table style sheet control, and scripting language technology to prohibit operations such as right-click copying to prevent page content from being selected and copied. There is no specific restriction on the protection methods and measures here.

[0134] In this embodiment, deep learning is combined with big data algorithms to solve the limited processing capabilities of traditional OCR methods for complex backgrounds, uneven lighting, font deformation, etc., thereby improving the recognition rate and retrieval accuracy. And based on the knowledge graph, its own AI retrieval algorithm is reconstructed to "translate" the data in the database into multi-dimensional, deep structured and semi-structured data processing to form a searchable and understandable knowledge graph. By combining deep learning with traditional OCR, building a related knowledge database and forming a knowledge graph, the medical literature knowledge field combines modern medical knowledge with traditional ancient books to form an online medical knowledge reading platform with both depth and breadth that is installed on the Internet clinic system.

[0135] In an embodiment of the present application, a knowledge graph is constructed in advance based on a medical database. When query information is obtained, the medical database to be queried is determined based on the query information, and query results associated with the query information are obtained from the knowledge graph corresponding to the medical database. The query results are displayed in the display interface of the client, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy in medical resource recommendations.

[0136] According to an embodiment of the present application, a query result acquisition device is also provided. It should be noted that the query result acquisition device of this embodiment can be used to execute the query result acquisition method of embodiment 1 of the present application.

[0137] Figure 2 is a schematic diagram of a query result acquisition device according to an embodiment of the present application. Figure 2 As shown, the query result acquisition device 20 may include: a first acquisition unit 202 , a determination unit 204 , a calling unit 206 , an acquisition unit 208 and a display unit 210 .

[0138] The first acquisition unit 202 is used to acquire query information from the client.

[0139] The determination unit 204 is used to determine the medical database to be queried based on the query information.

[0140] The retrieving unit 206 is used to retrieve the knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities.

[0141] The acquisition unit 208 is used to acquire a query result associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than a correlation threshold, and the query result includes at least one entity.

[0142] The display unit 210 is used to display the query result on the display interface of the client.

[0143] The query result acquisition device of this embodiment acquires query information from a client through a first acquisition unit; determines a medical database to be queried based on the query information through a determination unit; retrieves a knowledge graph corresponding to the medical database through a retrieval unit, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; acquires query results associated with the query information from the knowledge graph through the acquisition unit, wherein the correlation between the query result and the query information is higher than a correlation threshold, and the query result contains at least one entity; displays the query results on a display interface of the client through a display unit, thereby achieving a technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy of medical resource recommendations.

[0144] The embodiment of the present application may provide a computer terminal, which may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal may also be replaced by a terminal device such as a mobile terminal.

[0145] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of the computer network.

[0146] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the method for obtaining query results: obtaining query information from the client; determining the medical database to be queried based on the query information; retrieving the knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; obtaining query results associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than the correlation threshold, and the query result contains at least one entity; and displaying the query results on the display interface of the client.

[0147] Optionally, Figure 3 is a structural block diagram of a computer terminal according to an embodiment of the present application, such as Figure 3 As shown, the computer terminal 308 may include: one or more (only one is shown in the figure) processors 302 , a memory 304 and a transmission device 306 .

[0148] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the query result acquisition method and device in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, the above-mentioned query result acquisition method is realized. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories can be connected to the computer terminal 308 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0149] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtain query information from the client; determine the medical database to be queried based on the query information; retrieve the knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; obtain query results associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than the correlation threshold, and the query result contains at least one entity; and display the query results on the display interface of the client.

[0150] It can be understood by those skilled in the art that Figure 3 The structure shown is for illustration only, and the computer terminal 308 may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (MID for short), a PAD, or other terminal devices. Figure 3 The structure of the computer terminal 308 is not limited. For example, the computer terminal 308 may also include Figure 3 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 3 Different configurations shown.

[0151] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0152] According to an embodiment of the present application, a computer-readable storage medium is further provided, the computer-readable storage medium including a stored program, wherein the program executes the method for obtaining the query result in Embodiment 1.

[0153] Optionally, in this embodiment, the computer-readable storage medium may be located in any one of the computer terminals in a computer terminal group in a computer network, or in any one of the mobile terminals in a mobile terminal group.

[0154] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for executing the following steps: obtaining query information from a client; determining a medical database to be queried based on the query information; retrieving a knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationships between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; obtaining query results associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than a correlation threshold, and the query result contains at least one entity; and displaying the query results on a display interface of the client.

[0155] Optionally, the computer-readable storage medium may also execute program code for the following steps: in response to obtaining query information, extracting a network token from the query information; verifying the network token to obtain a verification result; based on the verification result, determining the reading permission of the client; and determining the medical database that is allowed to be accessed by the reading permission as the medical database to be queried.

[0156] Optionally, the computer-readable storage medium may also execute program code for the following steps: obtaining access indicators of the client, wherein the access indicators are used to characterize the client's satisfaction with different historical query results; and using the access indicators to guide the conversation analysis model to generate query results based on the knowledge graph.

[0157] Optionally, the computer-readable storage medium may also execute program code for the following steps: converting text in the query result to obtain a conversion result; and displaying the conversion result in a display interface.

[0158] Optionally, the computer-readable storage medium may also execute program codes for the following steps: classifying the medical database to obtain a first sub-database and a second sub-database, wherein the first medical data in the first sub-database and the second medical data in the second sub-database are different, and the medical database includes the first medical data and the second medical data; extracting the first medical data using a convolutional neural network to obtain first extracted information; extracting the second medical data using a recurrent neural network to obtain second extracted information; and constructing a knowledge graph based on the first extracted information and the second extracted information.

[0159] Optionally, the computer-readable storage medium may also execute program codes for the following steps: encoding the first medical data using a convolutional neural network to obtain encoded data; and recognizing characters in the encoded data to obtain first extracted information.

[0160] Optionally, the above-mentioned computer-readable storage medium can also execute the program code of the following steps: determine the storage format of the first extraction information and the second extraction information; store the first extraction information and the second extraction information according to the storage format to obtain a knowledge graph.

[0161] In this embodiment, a knowledge graph is constructed in advance based on a medical database. When query information is obtained, the medical database to be queried is determined based on the query information, and query results associated with the query information are obtained from the knowledge graph corresponding to the medical database, and the query results are displayed in the display interface of the client, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy in medical resource recommendations.

[0162] According to an embodiment of the present application, a processor is also provided, which is used to run a program, wherein the method for obtaining the query result in Example 1 is executed when the program is run by the processor.

[0163] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of the computer network.

[0164] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the multilingual translation method: obtaining query information from the client; determining the medical database to be queried based on the query information; retrieving the knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the association relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; obtaining query results associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than the correlation threshold, and the query result contains at least one entity; and displaying the query result on the display interface of the client.

[0165] Wherein, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the multilingual translation method and device in the embodiment of the present application, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned multilingual translation method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to a computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0166] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtain query information from the client; determine the medical database to be queried based on the query information; retrieve the knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on multiple entities in the medical database and the relationship between the multiple entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; obtain query results associated with the query information from the knowledge graph, wherein the correlation between the query result and the query information is higher than the correlation threshold, and the query result contains at least one entity; and display the query results on the display interface of the client.

[0167] Optionally, the processor may also execute the program code of the following steps: in response to obtaining query information, extracting a network token from the query information; verifying the network token to obtain a verification result; determining the reading permission of the client based on the verification result; and determining the medical database that the reading permission allows access to as the medical database to be queried.

[0168] Optionally, the processor may also execute the program code of the following steps: obtaining the access index of the client, wherein the access index is used to characterize the satisfaction of the client with different historical query results; using the access index to guide the conversation analysis model to generate query results based on the knowledge graph.

[0169] Optionally, the processor may also execute program codes of the following steps: converting text in the query result to obtain a conversion result; and displaying the conversion result in a display interface.

[0170] Optionally, the processor may also execute program codes of the following steps: classifying the medical database to obtain a first sub-database and a second sub-database, wherein the first medical data in the first sub-database and the second medical data in the second sub-database are different, and the medical database includes the first medical data and the second medical data; extracting the first medical data using a convolutional neural network to obtain first extracted information; extracting the second medical data using a recurrent neural network to obtain second extracted information; and constructing a knowledge graph based on the first extracted information and the second extracted information.

[0171] Optionally, the processor may also execute program codes of the following steps: encoding the first medical data using a convolutional neural network to obtain encoded data; and recognizing characters in the encoded data to obtain first extracted information.

[0172] Optionally, the processor may also execute the program code of the following steps: determining the storage format of the first extracted information and the second extracted information; storing the first extracted information and the second extracted information according to the storage format to obtain a knowledge graph.

[0173] Optionally, the processor may also execute program codes of the following steps: extracting characteristic information of the query information; determining information to be recommended that is associated with the characteristic information in the medical database; and displaying the information to be recommended on a display interface.

[0174] By adopting the embodiment of the present application, a knowledge graph is pre-constructed based on a medical database. When query information is obtained, the medical database to be queried is determined based on the query information, and query results associated with the query information are obtained from the knowledge graph corresponding to the medical database. The query results are displayed in the display interface of the client, thereby achieving the technical effect of improving the accuracy of medical resource recommendations and solving the technical problem of low accuracy of medical resource recommendations.

[0175] According to an embodiment of the present application, a computer program product is also provided, which includes computer instructions, wherein when the computer instructions are executed by a processor, the method for obtaining the query results in Example 1 is implemented.

[0176] An embodiment of the present application may provide an electronic device, which may include a memory and a processor.

[0177] Figure 4 It is a block diagram of an electronic device according to a method for obtaining query results of an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0178] like Figure 4 As shown, the device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 to a random access memory (RAM) 403. In RAM 403, various programs and data required for the operation of the device 400 can also be stored. The computing unit 401, ROM 402, and RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0179] A number of components in the device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0180] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as a verification method for data. For example, in some embodiments, the verification method for data may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the verification method for data described above may be executed. Alternatively, in other embodiments, the computing unit 401 may be configured to execute the data verification method in any other appropriate manner (for example, by means of firmware).

[0181] According to an embodiment of the present application, a method for obtaining query results is provided. It should be noted that the steps shown in the flowchart of 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 can be executed in an order different from that shown here.

[0182] Various embodiments of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that are executable and / or interpreted on a programmable system including at least one programmable processor that may be a dedicated or general purpose programmable processor that may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0183] The program code for implementing the method of the present application can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0184] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer 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 of the foregoing.

[0185] To provide interaction with a user, the systems and techniques described herein may be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or an LCD (liquid crystal display, monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a path ball) through which the user can provide input to the computer. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0186] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a Local Area Network (LAN), a Wide Area Network (WAN), and the Internet.

[0187] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0188] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0189] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0190] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0191] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0192] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0193] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0194] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for obtaining query results, characterized in that: include: Get query information from the client; Based on the query information, determining a medical database to be queried; Retrieving a knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on a plurality of entities in the medical database and associations between the plurality of entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; Acquire a query result associated with the query information from the knowledge graph, wherein the degree of association between the query result and the query information is higher than a degree of association threshold, and the query result includes at least one of the entities; The query result is displayed on the display interface of the client.

2. The method according to claim 1, characterized in that: The step of determining a medical database to be queried based on the query information includes: In response to obtaining the query information, extracting a network token from the query information; Verifying the network token to obtain a verification result; Based on the verification result, determining the reading authority of the client; The medical database that the reading authority allows access to is determined as the medical database to be queried.

3. The method according to claim 1, characterized in that The obtaining, from the knowledge graph, the query result associated with the query information includes: Acquire the access index of the client, wherein the access index is used to represent the satisfaction degree of the client to different historical query results; The access indicator is used to guide the conversation analysis model to generate the query result based on the knowledge graph.

4. The method according to claim 1, characterized in that: The step of displaying the query result on a display interface of the client includes: Convert the text in the query result to obtain a conversion result; The conversion result is displayed in the display interface.

5. The method according to claim 1, characterized in that The method further comprises: Classifying the medical database to obtain a first sub-database and a second sub-database, wherein the first medical data in the first sub-database and the second medical data in the second sub-database are different, and the medical database includes the first medical data and the second medical data; Extracting the first medical data using a convolutional neural network to obtain first extracted information; extracting the second medical data using a recurrent neural network to obtain second extracted information; Based on the first extracted information and the second extracted information, the knowledge graph is constructed.

6. The method according to claim 5, characterized in that The step of extracting the first medical data using a convolutional neural network to obtain first extracted information includes: Encoding the first medical data using the convolutional neural network to obtain encoded data; Characters in the encoded data are recognized to obtain first extraction information.

7. The method according to claim 5, characterized in that The constructing the knowledge graph based on the first extracted information and the second extracted information includes: determining a storage format of the first extracted information and the second extracted information; The first extracted information and the second extracted information are stored according to the storage format to obtain the knowledge graph.

8. The method according to claim 7, characterized in that The first extracted information and the second extracted information are semi-structured information.

9. The method according to any one of claims 1 to 8, characterized in that: The method further comprises: Extracting characteristic information of the query information; Determining the information to be recommended in the medical database that is associated with the characteristic information; The information to be recommended is displayed on the display interface.

10. A device for obtaining query results, characterized in that: include: A first acquisition unit, used to acquire query information from a client; A determination unit, configured to determine a medical database to be queried based on the query information; A retrieving unit, used to retrieve a knowledge graph corresponding to the medical database, wherein the knowledge graph is constructed based on a plurality of entities in the medical database and associations between the plurality of entities, the entities are extracted from the medical database, and the nodes in the knowledge graph are used to represent the entities; an acquisition unit, configured to acquire, from the knowledge graph, a query result associated with the query information, wherein the degree of association between the query result and the query information is higher than a degree of association threshold, and the query result includes at least one of the entities; A display unit is used to display the query result on the display interface of the client.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 9 when executed by a processor.

12. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 9.

13. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the method as claimed in any one of claims 1 to 9.

Images