Terminology processing methods, devices, equipment, storage media, and program products

Abstract

The present application provides a term processing method, apparatus, device, and computer-readable storage medium; the method includes: obtaining an input word in a specific field; constructing a first term operation tree corresponding to the input word; based on the first term operation tree and a second term operation tree pre-constructed for each standard word in the term standard table of the specific field, recalling the term standard table to obtain multiple candidate standard words corresponding to the input word; determining the tree similarity between the first term operation tree and the second term operation tree corresponding to each candidate standard word, sorting the multiple candidate standard words in descending order according to the tree similarity, and determining the candidate standard words at the head of the descending sorting result as the standard words to be checked; screening the composition of each standard word to be checked, and determining the standard words to be checked that meet the rationality index as the standard words that match the input word. Through the present application, the accuracy of matching standard words can be effectively improved.

Description

Technical Field

[0001] The present application relates to artificial intelligence technology, and in particular to a terminology processing method, device, electronic device, computer-readable storage medium, and computer program product. Background Art

[0002] Artificial Intelligence (AI) refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. AI also studies the design principles and implementation methods of various intelligent machines, enabling them to possess the capabilities of perception, reasoning, and decision-making.

[0003] Terminology standardization is an important application of artificial intelligence in natural language processing, and is used in fields such as medical care and advice.

[0004] In the related art, since the input words to be processed are highly random, the accuracy of the standard words determined to match the input words is low. There is no effective solution in the related art on how to improve the accuracy of matching standard words in the term standardization process. Summary of the Invention

[0005] The embodiments of the present application provide a term processing method, apparatus, electronic device, computer-readable storage medium, and computer program product, which can effectively improve the accuracy of matching standard words.

[0006] The technical solution of the embodiment of the present application is implemented as follows:

[0007] This embodiment of the present application provides a term processing method, including:

[0008] Get input words for a specific field;

[0009] constructing a first term operation tree corresponding to the input word;

[0010] Based on the first terminology operation tree and a second terminology operation tree pre-constructed for each standard word in the terminology standard table of the specific field, performing a recall process on the terminology standard table to obtain a plurality of candidate standard words corresponding to the input word;

[0011] determining tree similarities between the first term operation tree and a second term operation tree corresponding to each candidate standard word, sorting the plurality of candidate standard words in descending order according to the tree similarities, and determining some candidate standard words at the head of the descending sorting results as the standard words to be searched;

[0012] The composition of each of the to-be-checked standard words is screened, and the obtained to-be-checked standard words that meet the rationality index are determined as standard words that match the input word.

[0013] The present invention provides a terminology processing device, including:

[0014] The acquisition module is used to obtain input words in a specific field;

[0015] A construction module, configured to construct a first term operation tree corresponding to the input word;

[0016] a first recall module configured to perform a recall process on the terminology standard table based on the first terminology operation tree and a second terminology operation tree pre-constructed for each standard word in the terminology standard table of the specific field, to obtain a plurality of candidate standard words corresponding to the input word;

[0017] a first determining module configured to determine tree similarities between the first term operation tree and a second term operation tree corresponding to each candidate standard word, sort the plurality of candidate standard words in descending order according to the tree similarities, and determine some candidate standard words at the head of the descending sorting results as the standard words to be searched;

[0018] The second determination module is used to screen the composition of each of the to-be-checked standard words, and determine the obtained to-be-checked standard words that meet the rationality index as the standard words that match the input word.

[0019] An embodiment of the present application provides an electronic device, including:

[0020] a memory for storing executable instructions;

[0021] The processor is used to implement the terminology processing method provided in the embodiment of the present application when executing the executable instructions stored in the memory.

[0022] An embodiment of the present application provides a computer-readable storage medium storing executable instructions for causing a processor to execute and implement the terminology processing method provided in the embodiment of the present application.

[0023] The present invention provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the terminology processing method described above in the present invention.

[0024] The embodiments of the present application have the following beneficial effects:

[0025] By recalling the terminology standard table based on the first terminology operation tree corresponding to the input word and the second terminology operation tree corresponding to the standard word, a plurality of candidate standard words corresponding to the input word are obtained, thereby preliminarily determining the scope of the standard words matching the input word, and then determining the standard words to be checked among the plurality of candidate standard words based on the tree similarity between the first terminology operation tree and the second terminology operation tree, thereby further narrowing the scope of the standard words matching the input word, and finally, through screening processing, accurately determining the standard words matching the input word, in this way, through multi-layer screening, gradually narrowing the scope of the standard words, and then accurately determining the standard words matching the input word, and at the same time, in the screening process, since the terminology operation tree can adapt to the disorder and randomness of the input word expression in different scenarios, it can effectively improve the accuracy of matching the standard words. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a structural diagram of the terminology processing system architecture provided by an embodiment of the present application;

[0027] Figure 2 is a structural diagram of a terminology processing device provided in an embodiment of the present application;

[0028] Figures 3A to 3E Schematic diagram of the process of terminology processing provided in the embodiment of the present application;

[0029] Figures 4A to 4E Schematic diagram of the principle of the term processing method provided in the embodiment of the present application;

[0030] Figures 5A to 5G It is a schematic diagram of the principle of the term processing method provided in the embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0032] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0033] In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0035] Before further describing the embodiments of the present application in detail, the nouns and terms involved in the embodiments of the present application are explained. The nouns and terms involved in the embodiments of the present application are subject to the following interpretations.

[0036] 1) Terminology standardization: This is an indispensable task in medical statistics. In clinical practice, there are often hundreds or even thousands of different ways of expressing the same diagnosis. The problem that terminology standardization aims to solve is to find corresponding standard medical terminology expressions for various clinical expressions.

[0037] 2) International Classification of Diseases (ICD): It is a system developed by the World Health Organization (WHO) that classifies diseases according to certain characteristics and uses rules to represent them in a coded manner. It is the basis for determining global health trends and statistics. It contains approximately 55,000 unique codes related to injuries, diseases, and causes of death, enabling health practitioners to exchange health information around the world using a common language.

[0038] 3) Short text matching task: This refers to the task of using a natural language processing (NLP) model to predict the semantic relevance of two short texts. The matching is generally performed using a distance metric in a vector space.

[0039] 4) Calculate Tree: A data structure consisting of a hierarchical set of n (n ≥ 1) finite nodes. Each node in a calculation tree has zero or more child nodes. A node without a parent is called a root node. Each root node has exactly one parent node. Except for the root node, each child node can be divided into multiple disjoint subtrees.

[0040] 5) Standard terms: Terms stored in standard tables are standardized expressions for describing specific things. They are unified and standardized expressions for repetitive things. They are based on the combined results of science, technology, and practical experience, and are published in a specific form as a basis and standard for common compliance. For example, in medical scenarios, standard terms can be standardized expressions for medical terms such as diseases, injuries, and medications. For example, the World Health Organization (WHO) publishes unified and standardized expressions for medical terms such as diseases, injuries, and medications in the International Classification of Diseases.

[0041] 6) Non-standard words: These are words that are not yet included in the standard table and are not yet standardized. For example, in medical settings, non-standard words can be non-standardized expressions of medical terms related to diseases, injuries, and medications. For example, non-standard words can be the colloquial expressions of medical terms related to diseases, injuries, and medications used by doctors and patients.

[0042] During the implementation of the embodiments of this application, the applicant discovered that the related technology has the following problems:

[0043] In the related art, input words are usually standardized based on the machine learning model algorithm, but the premise that the machine learning model needs to meet the rationality is ignored. The modeling of the input words and standard words by the machine learning model is too simple, which makes the related art unable to accurately distinguish similar concepts. In the related art, the problem of incorrect matching due to the disorder of the input word order has not been solved. For example, the input word "chronic peritoneal hemorrhage with tumor" should be properly disassembled to obtain "chronic peritoneal hemorrhage with peritoneal tumor" and then the standardization process of the machine learning model should be carried out. The term operation tree in the embodiment of the present application is flexible enough to adapt to the disorder and arbitrariness of the input word expression in different scenarios. Thereby, the accuracy of matching standard words can be effectively improved.

[0044] Medical terminology standardization is a crucial technical capability in the healthcare informationization process and a cornerstone of medical artificial intelligence. Medical terminology standardization aims to map non-standard or non-standard diagnostic expressions to standardized or standard diagnostic expressions within the medical standards system. Related technologies for terminology standardization often employ machine learning or deep learning model algorithms, simplistically interpreting the standardization task as a short text matching task while ignoring the medical significance of terminology standardization. In other words, these technical solutions lack medical rationale. Furthermore, the explanations provided in these technologies are often black-box and unacceptable to doctors and other practitioners.

[0045] The terminology processing method provided in the embodiment of the present application decouples the medical part and the model algorithm part by fully analyzing and understanding the terminology standardization task, and at the same time uses the specially designed data structure of the terminology operation tree (i.e., the operation tree described above) to realize the interaction between the medical part and the model algorithm part, so that the medical part and the model algorithm part can truly perform their respective functions, further improving the performance and interpretability of the overall engine.

[0046] The embodiments of the present application provide a terminology processing method, apparatus, electronic device, computer-readable storage medium, and computer program product, which can effectively improve the accuracy of matching standard words. The following describes exemplary applications of the electronic device provided by the embodiments of the present application. The electronic device provided by the embodiments of the present application can be implemented as various types of user terminals such as laptop computers, tablet computers, desktop computers, set-top boxes, mobile devices (for example, mobile phones, portable music players, personal digital assistants, dedicated messaging devices, portable gaming devices), and can also be implemented as servers.

[0047] See also Figure 1 , Figure 1 This is an architectural diagram of a terminology processing system 100 provided in an embodiment of the present application. To implement application scenarios of terminology processing (for example, in the application scenario of medical informatization, medical terminology is standardized; in the application scenario of educational informatization, professional terminology is standardized; in the application scenario of network search, search keywords are standardized), a terminal (terminal 400 is shown as an example) is connected to a server 200 via a network 300. The network 300 can be a wide area network or a local area network, or a combination of the two.

[0048] The terminal 400 is used for the user to use the client 410, which is displayed on a graphical interface 410-1 (graphic interface 410-1 is shown as an example). The terminal 400 and the server 200 are connected to each other via a wired or wireless network.

[0049] In some embodiments, the server 200 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal 400 can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, smart voice interaction device, smart home appliance, car terminal, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiments of the present application.

[0050] In some embodiments, the client of terminal 400 receives an input word and sends it to server 200 via network 300. Server 200 determines a standard word that matches the input word and sends the matching standard word to graphical interface 410-1 in terminal 400 for display.

[0051] In some embodiments, the client of the terminal 400 receives an input word, determines a standard word that matches the input word, and displays the standard word that matches the input word in the graphical interface 410 - 1 in the terminal 400 .

[0052] See also Figure 2 , Figure 2 2 is a structural diagram of a server 200 of the terminology processing method provided in an embodiment of the present application. Figure 2 The server 200 shown includes: at least one processor 210, a memory 250, and at least one network interface 220. The various components in the server 200 are coupled together via a bus system 240. It is understood that the bus system 240 is used to achieve connection and communication between these components. In addition to the data bus, the bus system 240 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, the bus system 240 is not described in detail. Figure 2 Various buses are labeled as bus system 240 .

[0053] The processor 210 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0054] The memory 250 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard drives, optical drives, etc. The memory 250 may optionally include one or more storage devices that are physically remote from the processor 210.

[0055] The memory 250 includes volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be a read-only memory (ROM), and the volatile memory may be a random access memory (RAM). The memory 250 described in the embodiments of the present application is intended to include any suitable type of memory.

[0056] In some embodiments, the memory 250 can store data to support various operations, examples of which include programs, modules, and data structures, or a subset or superset thereof, as exemplified below.

[0057] The operating system 251 includes system programs for processing various basic system services and performing hardware-related tasks, such as a framework layer, a core library layer, a driver layer, etc., which are used to implement various basic services and process hardware-based tasks.

[0058] The network communication module 252 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 220. Exemplary network interfaces 220 include Bluetooth, Wireless LAN (WiFi), and Universal Serial Bus (USB).

[0059] In some embodiments, the terminology processing device provided in the embodiments of the present application can be implemented in software. Figure 2 A term processing device 255 stored in memory 250 is shown. This device can be software in the form of a program or plug-in, and includes the following software modules: an acquisition module 2551, a construction module 2552, a first recall module 2553, a first determination module 2554, and a second determination module 2555. These modules are logical and can be arbitrarily combined or further separated according to the functions they implement. The functions of each module will be described below.

[0060] In some embodiments, a terminal or server can implement the terminology processing methods provided in the embodiments of the present application by running a computer program. For example, the computer program can be a native program or software module in an operating system; it can be a native application (APP), that is, a program that needs to be installed in the operating system to run, such as a medical information application.

[0061] In other embodiments, the terminology processing device provided in the embodiments of the present application can be implemented in hardware. As an example, the terminology processing device provided in the embodiments of the present application can be a processor in the form of a hardware decoding processor, which is programmed to execute the terminology processing method provided in the embodiments of the present application. For example, the processor in the form of a hardware decoding processor can adopt one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0062] It is understandable that in the embodiments of the present application, when data related to input words is involved and is applied to specific products or technologies, user permission or consent must be obtained, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0063] The terminology processing method provided in the embodiment of the present application will be explained in combination with the exemplary application and implementation of the terminology processing server provided in the embodiment of the present application.

[0064] In some embodiments, Figure 4A This is a schematic diagram of the principle of the term processing method provided in the embodiment of this application. Figure 4A Based on the first term operation tree corresponding to the input word and the second term operation tree pre-constructed for each standard word in the term standard table for a specific field, the standard words in the term standard table are recalled, sorted, and screened to obtain standard words that match the input word, thereby effectively improving the accuracy of matching standard words.

[0065] See also Figure 3A , Figure 3A This is a flow chart of the term processing method provided in the embodiment of the present application, which will be combined with Figure 3A Steps 101 to 105 are shown for illustration, and the execution subject of the following steps 101 to 105 can be the aforementioned server or terminal.

[0066] In step 101, an input word of a specific field is obtained.

[0067] For example, specific fields may include medical, educational, and advice fields. For example, in the medical field, the input words obtained may be colloquial expressions of medical terms used by doctors and patients regarding illnesses, injuries, and medications. For example, in the advice field, the input words obtained may be colloquial expressions of certain standard terms.

[0068] In step 102, a first term operation tree corresponding to the input word is constructed.

[0069] In some embodiments, constructing a first term operation tree corresponding to an input word can be achieved by performing a split encoding process on the input word to obtain components of the input word, wherein the components include modifier components, positional components, root components, and logical components. Based on the modifier components, positional components, root components, and logical components, constructing the first term operation tree corresponding to the input word.

[0070] For example, see Figure 4B , Figure 4B This is a schematic diagram of the principle of the term processing method provided in the embodiment of the present application. When the input word is "progressive diabetic chronic bleeding / with / peritoneal tumor", the first term operation tree corresponding to the input word can be constructed as follows Figure 4B As shown, the root node of the first term operation tree is "companion", and the leaf nodes of the first term operation tree are "diabetic", "committed", "chronic", and "peritoneum", among which the parent node of the leaf nodes "diabetic", "committed", and "chronic" is "bleeding", and the parent node of the leaf node "peritoneum" is "tumor".

[0071] In step 103, based on the first terminology operation tree and the second terminology operation tree pre-constructed for each standard word in the terminology standard table for a specific field, the terminology standard table is recalled to obtain multiple candidate standard words corresponding to the input word.

[0072] As an example, in the medical field, a standard terminology table can be the International Classification of Diseases. The International Classification of Diseases is a system developed by the World Health Organization that classifies diseases according to certain characteristics and uses codes to represent them. See Table 1 below, which shows a partial standard terminology table provided in the examples of this application.

[0073] Table 1 Standard terminology

[0074]

[0075] As an example, based on the first term operation tree and the second term operation tree pre-constructed for each standard word in the term standard table for a specific field, the term standard table in Table 1 above is recalled to obtain multiple candidate standard words corresponding to the input word. For example, when the input word is "new cryptococcal encephalitis", the corresponding multiple candidate standard words can be "cerebral cryptococcosis", "cryptococcal meningitis", "new cryptococcal meningitis", and "cryptococcal encephalitis".

[0076] For example, see Figure 4C , the second term operation tree corresponding to the standard word "Cryptococcus neoformans meningitis" in the term standard table can be as follows Figure 4C In the second term operation tree shown, the root node of the second term operation tree is "meningitis", and the child nodes of the root node are "novel" and "cryptococcus".

[0077] In some embodiments, see Figure 3B , Figure 3B is a flowchart of the term processing method provided in the embodiment of the present application, Figure 3B Step 103 shown can be implemented through steps 1031 to 1032, which are described below respectively.

[0078] In step 1031 , node indices between the first terminology operation tree and a second terminology operation tree pre-constructed for each standard word in the terminology standard table for a specific field are determined.

[0079] The node index represents the consistency between the nodes of the first term operation tree and the nodes of the second term operation tree, and each node corresponds to a term component.

[0080] For example, see Figure 4D , Figure 4D This is a schematic diagram of the principle of the term processing method provided in an embodiment of the present application. The node index between the term component "companion" at the root node of the first term operation tree and the term component "meningitis" at the root node of the second term operation tree pre-constructed for each standard word in the term standard table for a specific field is determined.

[0081] In some embodiments, see Figure 3C , Figure 3C is a flowchart of the term processing method provided in the embodiment of the present application, Figure 3C Step 1031 shown can be implemented by executing steps 10311 to 10313 on the second term operation tree corresponding to any standard word in the term standard table of a specific field, which are respectively described below.

[0082] In step 10311, a root node index is determined according to the root node of the first term operation tree and the root node of the second term operation tree.

[0083] In some embodiments, see Figure 4D Based on the root node "companion" of the first term operation tree and the root node "meningitis" of the second term operation tree, a root node index is determined. The root node index represents the consistency between the root node of the first term operation tree and the root node of the second term operation tree.

[0084] In step 10312, leaf node indices are determined based on the part leaf nodes of the first term operation tree and the part leaf nodes of the second term operation tree.

[0085] Among them, the leaf nodes include part leaf nodes and modification leaf nodes. The part leaf nodes correspond to part components, and the modification leaf nodes correspond to modification components. The part leaf nodes of the first term operation tree correspond to the part components of the input word, and the part leaf nodes of the second term operation tree correspond to the part components of the standard word.

[0086] In some embodiments, see Figure 4D A leaf node index is determined based on the location leaf node "peritoneum" of the first term operation tree and the location leaf node "Cryptococcus" of the second term operation tree. The leaf node index represents the consistency between the leaf nodes of the first term operation tree and the leaf nodes of the second term operation tree.

[0087] In step 10313, the root node index and the leaf node index are determined as the node indexes between the first term operation tree and the second term operation tree.

[0088] In this way, by determining the root node index and the leaf node index as the node index between the first term operation tree and the second term operation tree, the determined node index not only represents the consistency between the leaf nodes of the first term operation tree and the leaf nodes of the second term operation tree, but also represents the consistency between the root node of the first term operation tree and the root node of the second term operation tree, so that the determined node index can accurately reflect the consistency between the nodes of the first term operation tree and the nodes of the second term operation tree.

[0089] In step 1032, the standard words corresponding to the node indicators that meet the recall condition are determined as a plurality of candidate standard words corresponding to the input word.

[0090] The recall condition includes that some nodes of the first term operation tree and the second term operation tree are the same, and the recall condition includes a root node indicator recall condition and a leaf node indicator recall condition.

[0091] As an example, referring to Table 1 above, when the node index of the standard word "cryptococcosis" meets the recall condition and the standard word "pulmonary cryptococcosis" does not meet the recall condition, the standard word "cryptococcosis" is determined as the candidate standard word corresponding to the input word.

[0092] In some embodiments, the above step 1031 can also be implemented by performing the following processing for each standard word in the terminology standard table: when the root node indicator corresponding to the standard word meets the root node indicator recall condition, and the leaf node indicator corresponding to the standard word meets the leaf node indicator recall condition, the standard word is determined as a candidate standard word corresponding to the input word.

[0093] Among them, the root node indicator recall condition includes that the root node of the first term operation tree is the same as the root node of the second term operation tree, and the leaf node indicator recall condition includes that the local leaf node of the first term operation tree is the same as the local leaf node of the second term operation tree.

[0094] As an example, when the root node indicator corresponding to the standard word "cryptococcosis" represents that the root node of the first term operation tree is the same as the root node of the second term operation tree corresponding to the standard word "cryptococcosis", the root node indicator corresponding to the standard word "cryptococcosis" meets the root node indicator recall condition. When the part leaf node indicator corresponding to the standard word "cryptococcosis" represents that the part leaf node of the first term operation tree is the same as the part leaf node of the second term operation tree corresponding to the standard word "cryptococcosis", the leaf node indicator corresponding to the standard word "cryptococcosis" meets the leaf node indicator recall condition. At this time, the root node indicator corresponding to the standard word "cryptococcosis" meets the root node indicator recall condition, and the leaf node indicator corresponding to the standard word "cryptococcosis" meets the leaf node indicator recall condition, then the standard word "cryptococcosis" is determined as the candidate standard word corresponding to the input word.

[0095] In this way, by judging whether the root node indicator corresponding to the standard word meets the root node indicator recall condition, and whether the leaf node indicator corresponding to the standard word meets the leaf node indicator recall condition, it is determined whether the standard word is determined as a candidate standard word corresponding to the input word, thereby recalling or screening the standard word from two aspects of the leaf node indicator recall condition and the root node indicator recall condition to obtain the corresponding candidate standard word, thereby significantly improving the accuracy of the obtained candidate standard word.

[0096] In step 104, the tree similarity between the first term operation tree and the second term operation tree corresponding to each candidate standard word is determined, the multiple candidate standard words are sorted in descending order according to the tree similarity, and the candidate standard words at the head of the descending sorting result are determined as the standard words to be checked.

[0097] In some embodiments, tree similarity represents the degree of similarity between the first term operation tree and the second term operation tree. Descending sorting may be performed by sorting the trees from largest to smallest according to the tree similarity values, resulting in a descending sort result. The candidate standard word at the head of the descending sort result may be the first candidate standard word, the second candidate standard word, and so on.

[0098] In some embodiments, see Figure 3B , Figure 3B is a flowchart of the term processing method provided in the embodiment of the present application, Figure 3B In step 104 shown, the tree similarity between the first term operation tree and the second term operation tree corresponding to each candidate standard word is determined. This can be achieved by executing steps 1041 to 1044 for the second term operation tree corresponding to each candidate standard word, which are explained below.

[0099] In step 1041 , a first edit distance between a root node of a first term operation tree and a root node of a second term operation tree is determined.

[0100] In some embodiments, the first edit distance represents the minimum number of single-character edit operations required to transition from the root node of the first term operation tree to the root node of the second term operation tree.

[0101] For example, see Figure 4E , a first edit distance dist(C1, C2) between the root node C2 of the first term operation tree and the root node C1 of the second term operation tree may be determined.

[0102] In step 1042 , a second edit distance between each intermediate node of the first term operation tree and each intermediate node of the second term operation tree is determined.

[0103] Among them, the intermediate node is the child node of the root node.

[0104] In some embodiments, the second edit distance represents the minimum number of single-character edit operations required to transition from an intermediate node of the first term operation tree to an intermediate node at the same position in the second term operation tree.

[0105] For example, see Figure 4E In the second term operation tree, intermediate nodes R1 and R2 are both children of root node C1. A second edit distance dist(R1, R3) can be determined between intermediate node R3 of the first term operation tree and intermediate node R1 of the second term operation tree. A second edit distance dist(R2, R4) can be determined between intermediate node R4 of the first term operation tree and intermediate node R2 of the second term operation tree.

[0106] In step 1043 , a third edit distance between the leaf nodes of the first term operation tree and the leaf nodes of the second term operation tree is determined.

[0107] Among them, the leaf node is the child node of the intermediate node.

[0108] In some embodiments, the third edit distance represents the minimum number of single-character edit operations required to convert from a leaf node of the first term operation tree to a leaf node at the same position in the second term operation tree.

[0109] For example, see Figure 4E In the second term operation tree, leaf nodes A1 and A2 are both children of intermediate node R1. A third edit distance dist(A1, A3) can be determined between leaf node A3 of the first term operation tree and leaf node A1 of the second term operation tree. A third edit distance dist(A2, A4) can be determined between leaf node A4 of the first term operation tree and leaf node A2 of the second term operation tree.

[0110] In step 1044 , the tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word is determined based on the first edit distance, the second edit distance, and the third edit distance.

[0111] In some embodiments, determining the tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word based on the first edit distance, the second edit distance, and the third edit distance in the above step 1044 can be achieved in the following manner: parameterizing the first edit distance to obtain a first tree similarity corresponding to the first edit distance; parameterizing the second edit distance to obtain a second tree similarity corresponding to the second edit distance; parameterizing the second edit distance and the third edit distance to obtain a third tree similarity corresponding to the third edit distance; and summing the first tree similarity, the second tree similarity, and the third tree similarity to obtain the tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word.

[0112] As an example, the first edit distance is parameterized, and the expression of the first tree similarity corresponding to the first edit distance can be:

[0113] SIM1(X, Y)=α(1-dist(C X , C Y )) (1)

[0114] Among them, dist(C x , C Y ) represents the first edit distance, SIM1(X, Y) represents the first tree similarity, and α is a hyperparameter for adjusting the first edit distance.

[0115] As an example, the second edit distance is parameterized, and the expression of the second tree similarity corresponding to the second edit distance can be:

[0116]

[0117] Among them, dist(R Xi , R Yj ) represents the second edit distance, β is the hyperparameter for adjusting the second edit distance, and SIM2(x, Y) represents the second tree similarity.

[0118] As an example, the second edit distance and the third edit distance are parameterized, and the expression of the third tree similarity corresponding to the third edit distance can be obtained as follows:

[0119]

[0120] Among them, SIM3(X, Y) represents the third tree similarity, dist(R Xi , R Yj ) represents the second edit distance, dist(A Xi , A Yj ) represents the third edit distance, dist(B Xi , B Yj )) represents the third edit distance, and γ is a hyperparameter for adjusting the third edit distance.

[0121] As an example, the first tree similarity, the second tree similarity, and the third tree similarity are summed up, and the expression of the tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word can be:

[0122]

[0123] Among them, dist(C X , C Y ) represents the first edit distance, dist(R xi , R Yj ) The second edit distance, dist(A Xi , A Yj ) represents the third edit distance, dist(B Xi , B Yj) represents the third edit distance, SIM(X, Y) represents the similarity between the first term operation tree X and the second term operation tree Y, α is the hyperparameter for adjusting the first edit distance, β is the hyperparameter for adjusting the second edit distance, γ is the hyperparameter for adjusting the third edit distance, SIM3(X, Y) represents the third tree similarity, SIM2(X, Y) represents the second tree similarity, and SIM1(X, Y) represents the first tree similarity.

[0124] In this way, by determining the tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word based on the first edit distance, the second edit distance and the third edit distance, the similarity between the first term operation tree and the second term operation tree can be accurately judged, thereby accurately determining the standard word to be checked among multiple candidate standard words.

[0125] In step 105, the composition of each to-be-checked standard word is screened, and the obtained to-be-checked standard word that meets the rationality index is determined as the standard word that matches the input word.

[0126] As an example, when the standard words to be checked are "male asymptomatic uterine cancer" and "symptomatic uterine cancer", the components of the standard words to be checked can be "male", "asymptomatic", "uterine cancer", and "symptomatic". The components of the standard word "male asymptomatic uterine cancer" to be checked, "male", "asymptomatic", and "uterine cancer", are screened. Specifically, the screening process may include a negation check, a medical rationality check, etc., wherein the negation check may be to delete the standard word to be checked when a negation is found in the components of the standard word to be checked. For example, if "asymptomatic" contains the negation "none", then the standard word to be checked "male asymptomatic uterine cancer" is deleted. The medical rationality check may be to delete the standard word to be checked when a component that does not conform to medical rationality is found in the components of the standard word to be checked. For example, since men do not have a uterus, "male" and "uterine cancer" cannot appear in the standard word to be checked at the same time. Then, through the medical rationality check, it is determined that the standard word to be checked "male asymptomatic uterine cancer" does not conform to medical rationality, and therefore the standard word to be checked "male asymptomatic uterine cancer" is deleted. Similarly, through screening processing, the standard word to be checked "symptomatic uterine cancer" can be determined as a standard word that matches the input word.

[0127] In this way, by screening the composition of each standard word to be checked, the standard word to be checked that meets the rationality index is determined as the standard word matching the input word, so that the rationality of the standard word matching the input word is higher.

[0128] In some embodiments, see Figure 3D , Figure 3D This is a flow chart of the term processing method provided in the embodiment of the present application. Figure 3D Before step 101 shown, steps 106 to 108 may be executed to construct a second term operation tree, which will be described below.

[0129] In step 106, a standard terminology table is obtained.

[0130] As an example, referring to the terminology standard table shown in Table 1 above, the standard word corresponding to code B45 is cryptococcosis, the standard word corresponding to B45.0 is pulmonary cryptococcosis, etc. The terminology standard table represents the mapping relationship between codes and standard words.

[0131] In step 107, each standard word in the terminology standard table is split and coded to obtain the components of each standard word.

[0132] Among them, components include modifying components, positional components, root components and logical components.

[0133] As an example, referring to Table 1 above, the standard word "Cryptococcus neoformans pneumonia" in the terminology standard table is split and coded, and the components of the obtained standard word "Cryptococcus neoformans pneumonia" can be "Cryptococcus", "new type", and "pneumonia", among which the modifying component is "new type", the root component is "pneumonia", and the location component is "Cryptococcus". The standard word "Cryptococcus neoformans pneumonia" has no logical components.

[0134] In step 108, a second term operation tree corresponding to the standard word is constructed based on the modifying components, the position components, the root components and the logical components.

[0135] In some embodiments, the construction of the second terminology operation tree corresponding to the standard word based on the modifying components, the positional components, the root components and the logical components in the above step 108 can be achieved in the following manner: determining the logical component as the root node of the second terminology operation tree; determining the root component as the intermediate node of the second terminology operation tree, wherein the intermediate node is the child node of the root node; determining the positional components and the modifying components as the leaf nodes of the second terminology operation tree, wherein the leaf node is the child node of the intermediate node; and constructing the second terminology operation tree corresponding to the standard word based on the root node, the intermediate node and the leaf node.

[0136] For example, see Figure 4E In the second terminology operation tree, the logical component C1 is determined as the root node of the second terminology operation tree; the root components R1 and R2 are determined as the intermediate nodes of the second terminology operation tree; the position components A1, A2 and the modifying components B1, B2 are determined as the leaf nodes of the second terminology operation tree; based on the root node C1, the intermediate nodes R1, R2 and the leaf nodes A1, A2, B1, B2, a second terminology operation tree corresponding to the standard word is constructed.

[0137] In some embodiments, see Figure 3D ,exist Figure 3D Before step 104 shown, candidate annotation words corresponding to the input word may be determined by executing at least one of steps 109 and 110 , which are respectively described below.

[0138] In step 109, the recall index of the input word relative to each standard word in the terminology standard table is determined, and the standard word with the largest recall index value is determined as the candidate standard word corresponding to the input word.

[0139] Among them, the recall index represents the similarity between the input word and the standard word.

[0140] In some embodiments, see Figure 3E , Figure 3E This is a flow chart of the term processing method provided in an embodiment of the present application. Determining the recall index of the input word relative to each standard word in the term standard table in the above step 109 can be achieved by executing steps 1091 to 1095.

[0141] In step 1091 , the edit distance between the input word and each standard word in the terminology standard table is determined, wherein the edit distance represents the degree of similarity between the literal features of the input word and the standard word.

[0142] As an example, the edit distance between input word A and standard word B in the terminology standard table can be dist(A, B). The edit distance between the input word and the standard word in the terminology standard table represents the minimum number of single-character editing operations required to convert the input word to the standard word.

[0143] In step 1092, a vector model is called based on the input word to perform vectorization processing to obtain an input word vector corresponding to the input word.

[0144] In step 1093, the vector model is called in turn based on each standard word in the terminology standard table to perform vectorization processing to obtain the standard word vector corresponding to the standard word.

[0145] In some embodiments, vectorization processing can be implemented by calling a vector model, where the vector model can be a BERT model, an ALBERT model, a word2vec model, a MedBERT model, etc.

[0146] In step 1094, the word vector similarity between the input word vector and the standard word vector corresponding to each standard word in the terminology standard table is determined. The word vector similarity represents the degree of similarity in semantic features between the input word and the standard word.

[0147] In some embodiments, the word vector similarity between the input word vector and the standard word vector can be the Hamming distance, Euclidean distance, cosine similarity, etc. between the input word vector and the standard word vector.

[0148] In step 1095 , the recall index of the input word relative to each standard word in the term standard table is determined based on the word vector similarity and the edit distance.

[0149] In some embodiments, in the above step 1095, the recall index of the input word relative to each standard word in the terminology standard table is determined based on the word vector similarity and the edit distance. The following processing can be performed for each standard word in the terminology standard table: the opposite of the edit distance between the input word and the standard word is determined as the literal feature, and the word vector similarity between the input word and the standard word is determined as the semantic feature; the literal feature and the semantic feature are weighted averaged, and the obtained weighted average processing result is determined as the recall index of the input word relative to each standard word in the terminology standard table.

[0150] As an example, the expression of the determined recall index may be:

[0151]

[0152] Among them, A represents the input word, B represents the standard word, score(A, B) represents the recall index, dist(A, B) represents the edit distance between the input word A and the standard word B, sim(A′, B′) represents the word vector similarity between the input word vector A′ and the standard word vector B′, A′ represents the input word vector, and B′ represents the standard word vector.

[0153] In step 110, a target root level matching the input word is determined among multiple root levels of the term standard table, and the standard words of the sublevels under the target root level and the standard words of the target root level are determined as candidate standard words corresponding to the input word.

[0154] Among them, the granularity of the standard words at the root level is greater than that of the standard words at the sub-levels under the root level, and the granularity of the standard words is negatively correlated with the degree of detail of the standard words.

[0155] As an example, referring to Table 1 above, the root level of the terminology standard table may be the level coded as B45, where the standard term "cryptococcosis" resides. The root level code shown in Table 1 above is 3 digits, and all other levels containing these 3 digits are sublevels of the root level. The granularity of the standard term "cryptococcosis" at the root level is greater than the standard term "pulmonary cryptococcosis" at the sublevel of the root level. The granularity of a standard term is negatively correlated with the level of detail of the standard term; the standard term "pulmonary cryptococcosis" is more detailed than the standard term "cryptococcosis."

[0156] In some embodiments, in the above step 110, determining the target root level that matches the input word among the multiple root levels of the terminology standard table can be achieved in the following way: determining the root level similarity between the input word and the standard word of each root level in the terminology standard table respectively; and determining the root level corresponding to the standard word whose root level similarity is greater than the similarity threshold as the target root level that matches the input word.

[0157] As an example, determine the root level similarity 5 between the input word and the standard word at the root level in the terminology standard table (the level where the standard word coded as B45 is located), and determine the root level similarity 6 between the input word and the standard word at the root level in the terminology standard table (the level where the standard word coded as B46 is located), where the root level similarity 5 is greater than the similarity threshold, and the root level similarity 6 is less than the similarity threshold, and the root level corresponding to the root level similarity 5 is determined as the target root level that matches the input word.

[0158] In this way, by recalling the term standard table based on the first term operation tree corresponding to the input word and the second term operation tree corresponding to the standard word, multiple candidate standard words corresponding to the input word are obtained, thereby preliminarily determining the scope of the standard words matching the input word, and then determining the standard words to be checked among the multiple candidate standard words based on the tree similarity between the first term operation tree and the second term operation tree, thereby further narrowing the scope of the standard words matching the input word, and finally through screening processing, accurately determining the standard words matching the input word. In this way, through multi-layer screening, the scope of the standard words is gradually narrowed, and then the standard words matching the input word are accurately determined. At the same time, in the screening process, since the term operation tree can adapt to the disorder and randomness of the input word expression in different scenarios, it can effectively improve the accuracy of matching standard words.

[0159] The following describes an exemplary application of the embodiment of the present application in an actual medical information application scenario.

[0160] Medical terminology standardization is a key cornerstone of medical artificial intelligence and plays a crucial role in numerous scenarios. For example, in the data center for health records, medical terminology standardization can significantly reduce the workload of medical record coders and enable hospitals to quickly and cost-effectively build a data center for information storage and query. Another example is that medical terminology standardization can also standardize and connect data from multiple hospitals of different levels and regions, thereby facilitating the construction of intelligent epidemic prevention and control dashboards and smart epidemic monitoring.

[0161] For example, see Figure 5A The clinical terminology standardization engine can help relevant departments to conduct intelligent assisted underwriting, help relevant departments to integrate data from all parties and provide a unified annotation diagnostic data interface. Figure 5AIn the display interface shown, the subcategories corresponding to gastric cancer may include pyloric gastric cancer and virus-related gastric cancer; the parent categories corresponding to gastric cancer may include cancer, primary malignant tumor of the stomach, morphological changes of gastric cancer may be cancer, and the site of occurrence of gastric cancer is the stomach, as well as a series of professional terminology expressions about gastric cancer.

[0162] See also Figure 5B , Figure 5B This is a schematic diagram of the overall architecture of the term processing method provided in the embodiment of this application. Figure 5B The overall architecture of the terminology processing method shown includes an engine input part, an offline calculation part, and an online processing part. The engine input part, the offline calculation part, and the online processing part are described below.

[0163] See also Figure 5B In engine input 1, the input consists of non-standard words (i.e., the input words described above) and a standard table of medical terms (i.e., the standard table described above). After rejecting, replacing, and segmenting the non-standard words, the engine performs principal component recognition based on the processing results, and constructs a term operation tree corresponding to the non-standard words based on the recognition results.

[0164] See also Figure 5B In offline computation 2, the medical terminology standard table is processed to obtain a medical terminology bridge table. The data processing includes block processing, indexing processing, and replacement processing. Principal component recognition is performed on the standard terms in the medical terminology bridge table, and a terminology operation tree is constructed based on the results of the principal component recognition processing.

[0165] See also Figure 5B Online computation 3 includes a multi-way recall model, a refined ranking model, and an output check model. Non-standard input words (i.e., the input words described above) undergo replacement, segmentation, and rejection processing before being processed by the multi-way recall model, refined ranking model, and output check model to produce standard output words.

[0166] The following is an explanation of the construction process of the term operation tree. Figure 5C , Figure 5C It is a flowchart of the term processing method provided in the embodiment of the present application.

[0167] First, the input word is split to obtain the key components in the term operation tree.

[0168] As an example, key components in the term operation tree may include: disease modifier components, disease anatomical site components, disease root components, and disease link logic components.

[0169] As an example, the input word can be split to first identify and locate the disease link logic component C of the input word. Specifically, the dictionary recognition function (lookup) can be used to identify the three types of splitting logics of "companion, and, or", and the splitting logic can be combined into the disease link logic component. Then, the disease modifying component, disease anatomical site component, and disease root component of the input word can be identified and located. Specifically, the dictionary recognition function (lookup) can be used for identification, or the sequence labeling model can be used for identification. The disease modifying component, disease anatomical site component, and disease root component of the input word can be identified and located separately and in parallel, so that the disease modifying component, disease anatomical site component, and disease root component obtained by identification and location will not be nested or partially overlapped.

[0170] As an example, for an input word: M = {m1, m2, m3, ..., m n}, where n is the length of the input word. The disease modifier, disease anatomical site, and disease root components of the input word are identified and located separately and in parallel in the following way:

[0171] Identify and locate the disease modifying component of the input word, and obtain the disease modifying component A of the input word = {m k , m k+1 , m k+2 ,…,m k+l}, where A={m k , m k+1 , m k+2 ,…,m k+l}∈M={m1, m2, m3,..., m n}, k represents the position of the disease modifier in the input word, l represents the length of the disease modifier, 1≤k≤n; 0≤l≤nk.

[0172] Identify and locate the disease anatomical site component of the input word, and obtain the disease anatomical site component B of the input word = {m q , m q+1 , m q+2 ,...,m q+w}, where B = {m q , m q+1 , m q+2 ,...,m q+l}∈M={m1, m2, m3,..., m n}, q represents the position of the anatomical site component of the disease in the input word, and w represents the length of the disease modification component, 1≤q≤n; 0≤w≤nq.

[0173] Identify and locate the disease root component of the input word, and obtain the disease root component R of the input word = {mr , m r+1 , m r+2 ,...,m r+x}, where R = {m r , m r+1 , m r+2 ,...,m r+x}∈M={m1, m2, m3,..., m n}, r represents the position of the anatomical site component of the disease in the input word, x represents the length of the disease modification component, 1≤r≤n; 0≤x≤nr.

[0174] Then, the disease modification component A of the input word, the disease anatomical site component B of the input word, and the disease root component R of the input word are merged to obtain a subtree structure of the term operation tree, wherein the subtree structure does not include the root node of the term operation tree.

[0175] See also Figure 5C After obtaining the disease modifier component A, the disease anatomical site component B, the disease root component R, and the disease linking logical component C of the input word, the disease modifier component A, the disease anatomical site component B, the disease root component R, and the disease linking logical component C of the input word are encoded. Then, based on the encoding result and the disease modifier component A, the disease anatomical site component B, the disease root component R, and the disease linking logical component C, a term operation tree is constructed to obtain a term operation tree. The term operation tree can also be decoded to obtain a corresponding decoding result.

[0176] See also Figure 5D , Figure 5D This is a schematic diagram of the principle of the term operation tree provided in the embodiment of this application. Figure 5D In the term operation tree 51, the disease link logic component C is taken as the root node of the term operation tree 51, the disease root component R (R1 and R2) of the input word is taken as the child node of the root node of the term operation tree 51, and the disease modification component A of the input word and the disease anatomical part component B of the input word are taken as the leaf nodes of the term operation tree 51, wherein the disease modification component A1 of the input word, the disease modification component A2 of the input word and the disease anatomical part component B1 of the input word are the child nodes of the disease root component R1 of the input word, and the disease anatomical part component B2 of the input word is the child node of the disease root component R2 of the input word.

[0177] In some embodiments, see Figure 5D The term operation tree 52 in Figure 5DThe term operation tree 51 and the term operation tree 52 shown are term operation trees with the same structure. The term operation tree 52 is an example of the term operation tree 51. In the term operation tree 52, it can be seen that under the disease root component "bleeding", there are the disease modifier components "diabetic", "progressive" and the disease root component "chronic", while under the disease root component "tumor", there is the disease anatomical site component "peritoneum".

[0178] See also Figure 5E , Figure 5E : This is a schematic diagram of the principle of the term operation tree provided in an embodiment of the present application. During the decoding process, an input word can be obtained based on the constructed term operation tree. The input word obtained by decoding does not consider the order between the components of the same level of the input word. For example, when decoding the term operation tree 51, the input word obtained can be A1D1 / C / R2, A1B1R1 / C / B2R2, A1A2B1R1 / C / B2R2. When decoding the term operation tree 52, the input word obtained can be diabetic / progressive / chronic / hemorrhage / with / peritoneum / tumor, diabetic / chronic / progressive / hemorrhage / with / peritoneum / tumor, progressive / diabetic / chronic / hemorrhage / with / peritoneum / tumor.

[0179] In this way, since the input word is obtained based on the constructed term operation tree, the input word obtained by decoding does not consider the order between the same-level components of the input word. Converting the input word from a short text into a term operation tree structure can effectively solve the problems of disorder and arbitrariness in the expression of input words in different scenarios.

[0180] In some embodiments, see Figure 5F In real-world scenarios, doctors or related medical workers have different writing standards, and natural language itself also has a lot of omissions. Therefore, after the above processing, the term operation tree can be expanded. For example, the term operation tree 51 is expanded to obtain the term operation tree 53. Figure 5F As shown, when the R2 component does not have a disease-modifying component, the leading R1 component can transfer the two disease-modifying components A1 / A2 to the trailing R2 component. For example, the example "diabetes / progressive / chronic / bleeding / with / peritoneum / tumor" can be expanded by transferring any one or more of "diabetes," "progressive," and "chronic" to "tumor," thereby obtaining "diabetes / progressive / chronic / bleeding / with / chronic / peritoneum / tumor."

[0181] In this way, by constructing a term operation tree, medical terms in the form of short text (such as diagnosis) can be structured in a more fine-grained manner, thereby constructing a term operation tree. The term operation tree needs to be strictly defined, and each part in the term operation tree corresponds to a different medical meaning. The flexibility and horizontal scalability of the term operation tree also ensure that the business needs of real scenarios are met to the greatest extent possible.

[0182] See also Figure 5B , the recall method of the BERT word vector similarity in the multi-way recall model is more dependent on the selected features, and different features often greatly affect the final recall effect. In order to effectively improve the overall recall rate of the multi-way recall model, a combination of literal features and semantic features can be used for recall. Among them, the literal feature is mainly the similarity distance (Levenshtein Distance), which is simply to obtain the edit distance between two entity words. The smaller the edit distance, the closer the two entity words are. The semantic feature refers to the similarity obtained by obtaining the semantic word vector of the input text. The word vector model provided in the embodiment of the present application can be compatible with but not limited to: word2vec, BERT, ALBERT, etc. In medical scenarios, the word vector model (Med BERT) can be used to obtain the word vector, wherein the expression of the word vector model (Med BERT) can be:

[0183] w=enc(w),enc∈{MedBERT,BERT,ALBERT} (6)

[0184] For input text w of different lengths, the word vector model can model it into word vectors of equal dimensions.

[0185] By merging the word vector of the input word and the word vector of the standard word, the cosine similarity between the input word and the standard word is calculated, and the entity word with the highest similarity is ranked as the entity chain result of the input entity word.

[0186]

[0187] Where A represents the input word, B represents the standard word, A′ represents the word vector corresponding to the input word, and B′ represents the word vector corresponding to the standard word. dist(A,B) represents the edit distance between the input word and the standard word, and sim(A′,B′) represents the cosine similarity between the word vector corresponding to the input word and the word vector corresponding to the standard word.

[0188] See also Figure 5B, three / six-digit code multi-level recall in the multi-channel recall model. For an input word A, it is necessary to recall all possible Bs in the standard system, which will lead to the problem of high computational complexity and may also introduce considerable noise. Therefore, the embodiment of the present application designs a data object based on the coding characteristics of ICD10 to accelerate the efficiency and effectiveness of recall. See Table 2 below, which is a standard terminology table provided in the embodiment of the present application.

[0189] Table 2 Standard terminology

[0190]

[0191] Referring to Table 2 above, we can see that the shortest code is a 3-digit code and the longest code is a 6-digit code + additional code. This terminology system ensures that terms with the same 3-digit code express similar concepts or concepts that belong to a certain large concept, while the differences between different 6-digit codes are mostly differences in finer-grained concepts, such as "B45.0 Pulmonary Cryptococcosis" and "B45.1 Cerebral Cryptococcosis". Therefore, during the recall process, only non-standard words and all 3-digit code terms can be processed, that is, only the similarity with "B45 Cryptococcosis" will be calculated. If the similarity is greater than a certain threshold, all terms under the 3-digit code concept (that is, all terms with the first three digits being "B45") will be recalled. This not only reduces the computational complexity, but also uses coarse-grained concepts to avoid some erroneous recalls, thereby ensuring the efficiency and effectiveness of the recall module.

[0192] See also Figure 5G , Figure 5G This is a schematic diagram of the principle of the term operation tree provided by the embodiment of the present application. In the multi-way recall model, the term operation tree constructed by the standard word and the original word (i.e., the input word described above) can help screen the candidate standard words for recall. Figure 5G As shown, the disease root component D and the anatomical component B in the term operation tree can be used as the basis for screening, and the recalled standard word can be limited to the original input word (i.e., the input word described above) with the same disease root component D and the same anatomical component B. If the original input word is "gastric bleeding", then after the term principal component screening, diseases whose disease root component D is not "stomach" will be excluded. The term principal component screening can take advantage of the term operation tree to screen the standard words in the recall stage, thereby further reducing the loss in the recall process and reducing the computational complexity of the fine ranking model. See Figure 5G ,The recall module is used to recall the disease root D and anatomical part B.

[0193] The embodiment of the present application fully utilizes the advantages of the data structure of the term operation tree while avoiding a significant increase in the engine's computational load. By adjusting the weights of the three hyperparameters alpha (α), beta (β), and gamma (γ), the importance of information at different levels can be distinguished, where alpha (α) represents the consistency of the connection component, beta (β) represents the consistency of the disease root component in two diagnostic terms, and gamma (γ) represents the consistency of the disease modifier component and the anatomical site component in the same disease root in two diagnostic terms.

[0194]

[0195]

[0196] Among them, dist(C X , C Y ) represents the first edit distance, dist(R Xi , R Yj ) The second edit distance, dist(A Xi , A Yj ) represents the third edit distance, dist(B Xi , B Yj ) represents the third edit distance, SIM(X,Y) represents the similarity between the first term operation tree X and the second term operation tree Y, α is the hyperparameter for adjusting the first edit distance, β is the hyperparameter for adjusting the second edit distance, γ is the hyperparameter for adjusting the third edit distance, SIM3(X,Y) represents the third tree similarity, SIM2(X,Y) represents the second tree similarity, and SIM1(X,Y) represents the first tree similarity.

[0197] In this way, the similarity between the recalled standard words and the input words is calculated by using the above-mentioned method of calculating the edit distance of the term operation tree, and then the final standard word ranking result is obtained by arranging them from large to small according to the similarity.

[0198] See also Figure 5B In the output check model, the sorted standard terms are primarily checked for medical plausibility, including negation checks. For example, if "no fever" is a negation, standard terms containing "no fever" will be directly excluded. In the medical plausibility check, if "male" and "uterus" appear simultaneously in a standard term, the standard term will be directly excluded. The output check model can be configured according to actual business needs and requires extensive medical knowledge to support it.

[0199] Continue to see Figure 5B , you can also match the input non-standard words with the business whitelist. When the input non-standard words match the business whitelist, the input non-standard words are directly determined as output standard words.

[0200] In order to verify the effectiveness of the terminology processing method provided in the embodiments of the present application, experimental parameters based on a large amount of real diagnostic data are provided below.

[0201] As an example, by annotating the diagnostic data, a total of 253 valid data were annotated, and another 400 data were not standardized data. The term processing method and related technology provided by the embodiment of the present application were evaluated on this data. By splitting the evaluation dimension into a separate diagnostic level (including precision, recall, score), that is, a one-to-many situation is regarded as multiple data for evaluation; and the input level (accuracy), that is, a one-to-many situation is correct only if all the standardization is correct.

[0202] Table 3 Experimental parameters

[0203] Accuracy Recall Fraction Accuracy Related technologies 0.513 0.544 0.528 0.496 This application 0.802 0.601 0.687 0.560

[0204] As shown in Table 3 above, the terminology processing method provided in this embodiment of the application achieves a nearly 30% improvement in normalization scores for real-world diagnostic data compared to related technologies. This demonstrates the effectiveness of the terminology processing method provided in this embodiment of the application. Furthermore, the defects in related technologies are addressed through the terminology operation tree, while also ensuring that the output results are medically reasonable. On this basis, terminology alignment and mapping can be performed on outpatient record data from multiple hospitals of different levels, thereby connecting heterogeneous data through the terminology system.

[0205] The following continues to describe the exemplary structure of the term processing device 255 provided in the embodiment of the present application as a software module. In some embodiments, such as Figure 2 As shown, the software modules in the terminology processing device 255 stored in the memory 250 may include: an acquisition module 2551, which is used to acquire an input word in a specific field; a construction module 2552, which is used to construct a first terminology operation tree corresponding to the input word; a first recall module 2553, which is used to recall the terminology standard table based on the first terminology operation tree and the second terminology operation tree pre-constructed for each standard word in the terminology standard table for the specific field, to obtain multiple candidate standard words corresponding to the input word; a first determination module 2554, which is used to determine the tree similarity between the first terminology operation tree and the second terminology operation tree corresponding to each candidate standard word, sort the multiple candidate standard words in descending order according to the tree similarity, and determine the part of the candidate standard words at the head of the descending sorting result as the standard words to be checked; a second determination module 2555, which is used to screen the composition of each standard word to be checked, and determine the standard words to be checked that meet the rationality index as the standard words matching the input word.

[0206] In some embodiments, the above-mentioned first recall module 2553 is also used to determine the node indicators between the first term operation tree and the second term operation tree pre-constructed for each standard word in the term standard table for a specific field, wherein the node indicators represent the consistency between the nodes of the first term operation tree and the nodes of the second term operation tree, and each node corresponds to a term component; the standard words corresponding to the node indicators that meet the recall conditions are determined as multiple candidate standard words corresponding to the input words, wherein the recall conditions include that some nodes of the first term operation tree and the second term operation tree are the same.

[0207] In some embodiments, the above-mentioned first recall module 2553 is also used to perform the following processing on the second term operation tree corresponding to any standard word in the term standard table of a specific field: determine the root node index based on the root node of the first term operation tree and the root node of the second term operation tree; determine the leaf node index based on the part leaf node of the first term operation tree and the part leaf node of the second term operation tree, wherein the leaf node includes the part leaf node and the modified leaf node; determine the root node index and the leaf node index as the node index between the first term operation tree and the second term operation tree.

[0208] In some embodiments, the above-mentioned first recall module 2553 is also used to perform the following processing for each standard word in the terminology standard table: when the root node indicator corresponding to the standard word meets the root node indicator recall condition, and the leaf node indicator corresponding to the standard word meets the leaf node indicator recall condition, the standard word is determined as a candidate standard word corresponding to the input word; wherein, the root node indicator recall condition includes that the root node of the first term operation tree and the root node of the second term operation tree are the same, and the leaf node indicator recall condition includes that the part leaf node of the first term operation tree and the part leaf node of the second term operation tree are the same.

[0209] In some embodiments, the first determination module 2554 is further configured to perform the following processing on the second terminology operation tree corresponding to each candidate standard word: determining a first edit distance between the root node of the first terminology operation tree and the root node of the second terminology operation tree; determining a second edit distance between each intermediate node of the first terminology operation tree and each intermediate node of the second terminology operation tree, wherein the intermediate node is a child node of the root node; determining a third edit distance between a leaf node of the first terminology operation tree and a leaf node of the second terminology operation tree, wherein the leaf node is a child node of the intermediate node; and determining the tree similarity between the first terminology operation tree and the second terminology operation tree corresponding to the candidate standard word based on the first edit distance, the second edit distance, and the third edit distance.

[0210] In some embodiments, the first determination module 2554 is further configured to parameterize the first edit distance to obtain a first tree similarity corresponding to the first edit distance; parameterize the second edit distance to obtain a second tree similarity corresponding to the second edit distance; parameterize the second edit distance and the third edit distance to obtain a third tree similarity corresponding to the third edit distance; and sum the first tree similarity, the second tree similarity, and the third tree similarity to obtain a tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word.

[0211] In some embodiments, the above-mentioned terminology processing device 255 also includes: a second acquisition module, used to obtain a terminology standard table; a splitting and encoding module, used to perform splitting and encoding processing on each standard word in the terminology standard table to obtain the components of each standard word, wherein the components include modifying components, part components, root components and logical components; a second construction module, used to construct a second terminology operation tree corresponding to the standard word based on the modifying components, part components, root components and logical components.

[0212] In some embodiments, the above-mentioned second construction module is further used to determine the logical component as the root node of the second term operation tree; determine the root component as the intermediate node of the second term operation tree, wherein the intermediate node is the child node of the root node; determine the part component and the modifying component as the leaf node of the second term operation tree, wherein the leaf node is the child node of the intermediate node; and construct the second term operation tree corresponding to the standard word based on the root node, the intermediate node and the leaf node.

[0213] In some embodiments, the term processing device 255 further includes: a second recall module for determining the recall index of the input word relative to each standard word in the term standard table, and determining the standard word with the largest recall index value as the candidate standard word corresponding to the input word, wherein the recall index represents the degree of similarity between the input word and the standard word. A third recall module for determining a target root level that matches the input word among multiple root levels of the term standard table, and determining the standard words of the sub-levels under the target root level and the standard words of the target root level as the candidate standard words corresponding to the input word, wherein the granularity of the standard words of the root level is greater than that of the standard words of the sub-levels under the root level, and the granularity of the standard words is negatively correlated with the degree of detail of the standard words.

[0214] In some embodiments, the above-mentioned second recall module is also used to determine the edit distance between the input word and each standard word in the terminology standard table, wherein the edit distance represents the degree of similarity in literal features between the input word and the standard word; based on the input word, the vector model is called to perform vectorization processing to obtain the input word vector corresponding to the input word; based on each standard word in the terminology standard table, the vector model is called in turn to perform vectorization processing to obtain the standard word vector corresponding to the standard word; the word vector similarity between the input word vector and the standard word vector corresponding to each standard word in the terminology standard table is determined, and the word vector similarity represents the degree of similarity in semantic features between the input word and the standard word; based on the word vector similarity and the edit distance, the recall index of the input word relative to each standard word in the terminology standard table is determined.

[0215] In some embodiments, the above-mentioned second recall module is also used to perform the following processing on each standard word in the terminology standard table: determining the opposite of the edit distance between the input word and the standard word as the literal feature, and determining the word vector similarity between the input word and the standard word as the semantic feature; performing weighted averaging processing on the literal feature and the semantic feature, and determining the obtained weighted average processing result as the recall index of the input word relative to each standard word in the terminology standard table.

[0216] In some embodiments, the above-mentioned third recall module is also used to determine the root-level similarity between the input word and each root-level standard word in the terminology standard table; and the root level corresponding to the standard word whose root-level similarity is greater than the similarity threshold is determined as the target root level that matches the input word.

[0217] The present invention provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the terminology processing method described above in the present invention.

[0218] The embodiment of the present application provides a computer-readable storage medium storing executable instructions, wherein the executable instructions are stored. When the executable instructions are executed by a processor, the processor will execute the term processing method provided by the embodiment of the present application, for example, Figure 3A The term processing method shown.

[0219] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface storage, optical disk, or CD-ROM; or various devices including one or any combination of the above memories.

[0220] In some embodiments, executable instructions may be in the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0221] As an example, executable instructions may, but need not, correspond to a file in a file system, may be stored as part of a file that stores other programs or data, such as in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinating files (e.g., files storing one or more modules, subroutines, or code portions).

[0222] By way of example, executable instructions may be deployed to be executed on one computing device, or on multiple computing devices at one site, or on multiple computing devices distributed across multiple sites and interconnected by a communication network.

[0223] In summary, the embodiments of the present application have the following beneficial effects:

[0224] (1) By recalling the terminology standard table based on the first terminology operation tree corresponding to the input word and the second terminology operation tree corresponding to the standard word, a plurality of candidate standard words corresponding to the input word are obtained, thereby preliminarily determining the scope of the standard words matching the input word, and then determining the standard words to be checked among the plurality of candidate standard words based on the tree similarity between the first terminology operation tree and the second terminology operation tree, thereby further narrowing the scope of the standard words matching the input word, and finally, through screening processing, accurately determining the standard words matching the input word. In this way, through multi-layer screening, the scope of the standard words is gradually narrowed, and then the standard words matching the input word are accurately determined. At the same time, in the screening process, since the terminology operation tree can adapt to the disorder and randomness of the input word expression in different scenarios, it can effectively improve the accuracy of matching the standard words.

[0225] (2) By determining the root node index and the leaf node index as the node index between the first term operation tree and the second term operation tree, the determined node index not only represents the consistency between the leaf nodes of the first term operation tree and the leaf nodes of the second term operation tree, but also represents the consistency between the root node of the first term operation tree and the root node of the second term operation tree, so that the determined node index can accurately reflect the consistency between the nodes of the first term operation tree and the nodes of the second term operation tree.

[0226] (3) By judging whether the root node indicator corresponding to the standard word meets the root node indicator recall condition, and whether the leaf node indicator corresponding to the standard word meets the leaf node indicator recall condition, it is determined whether the standard word is determined as a candidate standard word corresponding to the input word, thereby recalling or screening the standard word from two aspects of the leaf node indicator recall condition and the root node indicator recall condition to obtain the corresponding candidate standard word, thereby significantly improving the accuracy of the obtained candidate standard word.

[0227] (4) By determining the tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word based on the first edit distance, the second edit distance and the third edit distance, the similarity between the first term operation tree and the second term operation tree can be accurately judged, thereby accurately determining the standard word to be checked among the multiple candidate standard words.

[0228] (5) By screening the composition of each standard word to be checked, the standard word to be checked that meets the rationality index is determined as the standard word that matches the input word, so that the rationality of the standard word that matches the input word is higher.

[0229] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent replacements, and improvements made within the spirit and scope of the present application are included in the scope of protection of the present application.

Claims

1. A term processing method, characterized in that: The method comprises: Get input words for a specific field; Performing split encoding processing on the input word to obtain components of the input word; wherein the components include the modifying components of the input word, the position components of the input word, the root components of the input word, and the logical components of the input word; constructing a first term operation tree corresponding to the input word based on the modifying components of the input word, the position components of the input word, the root components of the input word, and the logical components of the input word; determining a node index between the first terminology operation tree and a second terminology operation tree pre-constructed for each standard word in the terminology standard table for the specific field, wherein the node index represents consistency between nodes of the first terminology operation tree and nodes of the second terminology operation tree, each of the nodes corresponding to a terminology component; determining the standard words corresponding to the node indicators that meet a recall condition as a plurality of candidate standard words corresponding to the input word, wherein the recall condition includes that some nodes of the first term operation tree and the second term operation tree are the same; determining tree similarities between the first term operation tree and a second term operation tree corresponding to each candidate standard word, sorting the plurality of candidate standard words in descending order according to the tree similarities, and determining some candidate standard words at the head of the descending sorting results as the standard words to be searched; The composition of each of the to-be-checked standard words is screened, and the obtained to-be-checked standard words that meet the rationality index are determined as standard words that match the input word.

2. The method according to claim 1, characterized in that Determining node indices between the first terminology operation tree and a second terminology operation tree pre-constructed for each standard word in the terminology standard table for the specific field includes: The following processing is performed on the second term operation tree corresponding to any one of the standard words in the term standard table of the specific field: determining a root node index according to the root node of the first term operation tree and the root node of the second term operation tree; determining leaf node indices according to the part leaf nodes of the first term operation tree and the part leaf nodes of the second term operation tree, wherein the leaf nodes include the part leaf nodes and the modified leaf nodes; The root node index and the leaf node index are determined as node indexes between the first term operation tree and the second term operation tree.

3. The method according to claim 1, characterized in that The recall conditions include root node indicator recall conditions and leaf node indicator recall conditions; The step of determining the standard words corresponding to the node indicators that meet the recall condition as the plurality of candidate standard words corresponding to the input word includes: The following processing is performed for each standard word in the terminology standard table: When the root node index corresponding to the standard word meets the root node index recall condition, and the leaf node index corresponding to the standard word meets the leaf node index recall condition, determining the standard word as the candidate standard word corresponding to the input word; The root node index recall condition includes that the root node of the first term operation tree is the same as the root node of the second term operation tree, and the leaf node index recall condition includes that the part leaf node of the first term operation tree is the same as the part leaf node of the second term operation tree.

4. The method according to claim 1, wherein The determining of the tree similarity between the first term operation tree and the second term operation tree corresponding to each candidate standard word includes: The following processing is performed on the second term operation tree corresponding to each candidate standard word: determining a first edit distance between a root node of the first term operation tree and a root node of the second term operation tree; determining a second edit distance between each intermediate node of the first term operation tree and each intermediate node of the second term operation tree, wherein the intermediate nodes are child nodes of the root node; determining a third edit distance between a leaf node of the first term operation tree and a leaf node of the second term operation tree, wherein the leaf node is a child node of the intermediate node; A tree similarity between the first term operation tree and a second term operation tree corresponding to the candidate standard word is determined based on the first edit distance, the second edit distance, and the third edit distance.

5. The method according to claim 4, characterized in that The determining, based on the first edit distance, the second edit distance, and the third edit distance, of a tree similarity between the first term operation tree and a second term operation tree corresponding to the candidate standard word includes: parameterizing the first edit distance to obtain a first tree similarity corresponding to the first edit distance; parameterizing the second edit distance to obtain a second tree similarity corresponding to the second edit distance; parameterizing the second edit distance and the third edit distance to obtain a third tree similarity corresponding to the third edit distance; The first tree similarity, the second tree similarity, and the third tree similarity are summed to obtain a tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word.

6. The method according to claim 1, characterized in that Before obtaining the input words in the specific field, the method further includes: Obtaining the standard table of terminology; Performing split encoding processing on each of the standard words in the terminology standard table to obtain components of each of the standard words, wherein the components include a modifying component of the standard word, a position component of the standard word, a root component of the standard word, and a logical component of the standard word; A second term operation tree corresponding to the standard word is constructed based on the modifying components of the standard word, the position components of the standard word, the root components of the standard word and the logical components of the standard word.

7. The method according to claim 6, characterized in that The step of constructing a second term operation tree corresponding to the standard word based on the modifying components of the standard word, the position components of the standard word, the root components of the standard word, and the logical components of the standard word includes: determining the logical component of the standard word as the root node of the second term operation tree; Determining the root component of the standard word as an intermediate node of the second term operation tree, wherein the intermediate node is a child node of the root node; Determining the position component of the standard word and the modifying component of the standard word as leaf nodes of the second term operation tree, wherein the leaf nodes are child nodes of the intermediate node; A second term operation tree corresponding to the standard word is constructed based on the root node, the intermediate nodes and the leaf nodes.

8. The method according to claim 1, characterized in that Before determining the tree similarity between the first term operation tree and the second term operation tree corresponding to each candidate standard word, the method further includes: Perform at least one of the following: Determining a recall index of the input word relative to each standard word in the terminology standard table, and determining the standard word with the largest recall index value as the candidate standard word corresponding to the input word, wherein the recall index represents the degree of similarity between the input word and the standard word; Among the multiple root levels of the terminology standard table, a target root level that matches the input word is determined, and the standard words of the sub-levels under the target root level and the standard words of the target root level are determined as candidate standard words corresponding to the input word, wherein the granularity of the standard words of the root level is greater than the standard words of the sub-levels under the root level, and the granularity of the standard words is negatively correlated with the degree of detail of the expression of the standard words.

9. The method according to claim 8, characterized in that Determining the recall index of the input word relative to each standard word in the terminology standard table, including: Determining an edit distance between the input word and each standard word in the terminology standard table, wherein the edit distance represents a degree of similarity between literal features of the input word and the standard word; Performing vectorization processing based on the input word by calling a vector model to obtain an input word vector corresponding to the input word; Based on each standard word in the terminology standard table, the vector model is called in sequence to perform vectorization processing to obtain a standard word vector corresponding to the standard word; Determining word vector similarities between the input word vector and the standard word vector corresponding to each standard word in the terminology standard table, wherein the word vector similarities represent the degree of similarity of semantic features between the input word and the standard word; Determine a recall index of the input word relative to each standard word in the term standard table according to the word vector similarity and the edit distance.

10. The method according to claim 9, characterized in that Determining the recall index of the input word relative to each standard word in the terminology standard table according to the word vector similarity and the edit distance includes: The following process is performed for each standard word in the terminology standard table: Determine the inverse of the edit distance between the input word and the standard word as a literal feature, and determine the word vector similarity between the input word and the standard word as a semantic feature; The literal features and the semantic features are weighted averaged, and the obtained weighted average processing result is determined as the recall index of the input word relative to each standard word in the terminology standard table.

11. The method according to claim 8, characterized in that The step of determining a target root level matching the input word from among the plurality of root levels in the terminology standard table comprises: Determining root-level similarities between the input word and each root-level standard word in the terminology standard table; The root level corresponding to the standard word whose root level similarity is greater than the similarity threshold is determined as the target root level that matches the input word.

12. A terminology processing device, characterized in that: The device comprises: The acquisition module is used to obtain input words in a specific field; a construction module configured to perform split encoding processing on the input word to obtain components of the input word; wherein the components include a modifying component of the input word, a positional component of the input word, a root component of the input word, and a logical component of the input word; and construct a first term operation tree corresponding to the input word based on the modifying component of the input word, the positional component of the input word, the root component of the input word, and the logical component of the input word; A first recall module is configured to determine node indices between the first terminology operation tree and a second terminology operation tree pre-constructed for each standard word in the terminology standard table for the specific field, wherein the node indices represent consistency between nodes of the first terminology operation tree and nodes of the second terminology operation tree, each of the nodes corresponding to a terminology component; and determine the standard words corresponding to the node indices that meet a recall condition as a plurality of candidate standard words corresponding to the input word, wherein the recall condition includes that some nodes of the first terminology operation tree and the second terminology operation tree are identical. a first determining module configured to determine tree similarities between the first term operation tree and a second term operation tree corresponding to each candidate standard word, sort the plurality of candidate standard words in descending order according to the tree similarities, and determine some candidate standard words at the head of the descending sorting results as the standard words to be searched; The second determination module is used to screen the composition of each of the to-be-checked standard words, and determine the obtained to-be-checked standard words that meet the rationality index as the standard words that match the input word.

13. The device according to claim 12, characterized in that The first recall module is further configured to: The following processing is performed on the second term operation tree corresponding to any one of the standard words in the term standard table of the specific field: determining a root node index according to the root node of the first term operation tree and the root node of the second term operation tree; determining leaf node indices according to the part leaf nodes of the first term operation tree and the part leaf nodes of the second term operation tree, wherein the leaf nodes include the part leaf nodes and the modified leaf nodes; The root node index and the leaf node index are determined as node indexes between the first term operation tree and the second term operation tree.

14. The device according to claim 12, characterized in that The recall conditions include root node indicator recall conditions and leaf node indicator recall conditions; The first recall module is further configured to: The following processing is performed for each standard word in the terminology standard table: When the root node index corresponding to the standard word meets the root node index recall condition, and the leaf node index corresponding to the standard word meets the leaf node index recall condition, determining the standard word as the candidate standard word corresponding to the input word; The root node index recall condition includes that the root node of the first term operation tree is the same as the root node of the second term operation tree, and the leaf node index recall condition includes that the part leaf node of the first term operation tree is the same as the part leaf node of the second term operation tree.

15. The device according to claim 12, characterized in that The first recall module is further configured to: The following processing is performed on the second term operation tree corresponding to each candidate standard word: determining a first edit distance between a root node of the first term operation tree and a root node of the second term operation tree; determining a second edit distance between each intermediate node of the first term operation tree and each intermediate node of the second term operation tree, wherein the intermediate nodes are child nodes of the root node; determining a third edit distance between a leaf node of the first term operation tree and a leaf node of the second term operation tree, wherein the leaf node is a child node of the intermediate node; A tree similarity between the first term operation tree and a second term operation tree corresponding to the candidate standard word is determined based on the first edit distance, the second edit distance, and the third edit distance.

16. The device according to claim 15, characterized in that The first recall module is further configured to: parameterizing the first edit distance to obtain a first tree similarity corresponding to the first edit distance; parameterizing the second edit distance to obtain a second tree similarity corresponding to the second edit distance; parameterizing the second edit distance and the third edit distance to obtain a third tree similarity corresponding to the third edit distance; The first tree similarity, the second tree similarity, and the third tree similarity are summed to obtain a tree similarity between the first term operation tree and the second term operation tree corresponding to the candidate standard word.

17. An electronic device, characterized in that: The electronic device comprises: a memory for storing executable instructions; A processor is configured to implement the terminology processing method according to any one of claims 1 to 11 when executing the executable instructions stored in the memory.

18. A computer-readable storage medium storing executable instructions, characterized in that: When the executable instructions are executed by a processor, the terminology processing method according to any one of claims 1 to 11 is implemented.

19. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the terminology processing method according to any one of claims 1 to 11 is implemented.

Images