Method, device and medium for constructing knowledge base for children's neurodevelopmental issues

Abstract

The present invention relates to the technical field of database construction, and in particular to a knowledge base construction method, device and medium for children's neurodevelopmental problems. The method comprises: generating initial text data of children's neurodevelopmental problems based on an existing database, cleaning and preprocessing the initial text data to obtain word segmentation results; converting vocabulary vectors according to the word segmentation results to obtain candidate vector sequences and identify subject semantic vocabulary vectors; analyzing the distribution of subject semantic vocabulary vectors and obtaining corresponding description levels; obtaining and marking vocabulary corresponding to each description level in the initial text data to form text ranges corresponding to each description level, obtaining storage priorities of subject semantic vocabulary vectors in combination with level truncation evaluation indicators, marking corresponding storage priorities based on text ranges, and completing the construction of a knowledge base; preventing the occurrence of similar symptoms leading to confusion in classification results, and enhancing the accuracy of knowledge data in the corresponding field of the knowledge base.

Description

Technical Field

[0001] The present invention relates to the technical field of database construction, and in particular to a method, device and medium for constructing a knowledge base for children's neurodevelopmental issues. Background Art

[0002] Children's neurodevelopmental problems refer to diseases based on abnormal development of the nervous system, which may interfere with children's acquisition and perception of the external environment, usually manifesting as obstacles in attention, memory, comprehension, language ability, problem-solving ability and social communication. If these obstacles are mild, they can be controlled to a certain extent through behavioral and educational interventions; therefore, in order to conduct more comprehensive diagnosis and treatment, it is necessary to provide professional diagnostic and treatment personnel with a convenient and efficient knowledge query channel.

[0003] In the prior art, a knowledge base is usually constructed based on the physical problems that need to be queried. However, when the text data is automatically classified and labeled through the natural language processing (NLP) method, there are some problems that need to be solved. Since some unclassified texts represent some similar disease problems, the existing distinction methods are not able to distinguish similar disease problems in these fields. This leads to poor classification results when text splitting and classification are performed when constructing the knowledge base, which affects the accuracy of the knowledge data in the corresponding field of the knowledge base. Summary of the invention

[0004] In order to solve the technical problem that the existing methods are not able to distinguish between similar fields of keywords, thus affecting the accuracy of knowledge base knowledge data, the purpose of the present invention is to provide a method for constructing a knowledge base for children's neurodevelopmental issues. The technical solution adopted is as follows:

[0005] Step S1: Generate initial text data of children's neurodevelopmental problems based on the existing database, and clean and preprocess the initial text data to obtain word segmentation results;

[0006] Step S2: converting the vocabulary vector according to the word segmentation result to obtain a candidate vector sequence, judging the semantic conformity index for each candidate vector in the candidate vector sequence, and obtaining the main semantic vocabulary vector through the semantic conformity index;

[0007] Step S3: Analyze the distribution of the subject semantic vocabulary vectors, calculate the level cutoff evaluation index, and obtain the description levels of each layer corresponding to all the subject semantic vocabulary vectors;

[0008] Step S4: Acquire and mark the vocabulary corresponding to each description level in the initial text data to form the text range corresponding to each description level, obtain the storage priority of the main semantic vocabulary vector in combination with the level truncation evaluation index, mark the corresponding storage priority based on the text range, and complete the construction of the knowledge base.

[0009] Furthermore, the initial text data includes professional literature, research materials and clinical case records.

[0010] Further, step S1 includes:

[0011] Based on the existing database, professionals screened the initial text data covering children's neurodevelopmental issues;

[0012] Cleaning the initial text data by regular expressions and / or building a normal vocabulary dictionary, removing noise words to obtain second-generation text data;

[0013] Split the second-generation text data to obtain the source word segmentation results and the part-of-speech tags of the corresponding word segmentations;

[0014] The source word segmentation result is obtained by removing punctuation marks and special symbols.

[0015] Furthermore, in step S2, the vocabulary vector is converted according to the word segmentation result to obtain a candidate vector sequence, including:

[0016] Step S21: converting each word in the word segmentation result into a corresponding vocabulary vector to form a vector space, clustering the vector space to obtain a plurality of clusters and unclassified vocabulary vectors, wherein the unclassified vocabulary vector is a vocabulary vector that deviates from children's neurodevelopmental problems;

[0017] Step S22: according to the preset center range of the cluster center, based on any cluster, the vocabulary vectors in the center range are recorded as candidate vectors, and the distance between each candidate vector and the cluster center is obtained respectively, and the candidate vectors are sorted in order from small to large according to the distance to generate a candidate vector sequence;

[0018] Repeat step S22 to obtain candidate vector sequences of all clusters.

[0019] Furthermore, in step S2, the semantic conformity index is determined for each candidate vector in the candidate vector sequence, and the corresponding calculation formula is:

[0020]

[0021] in, Represents the current candidate vector The semantics of the index are consistent with the index; Represents the current candidate vector The maximum distance between the vectors in the cluster and other candidate vectors; Represents the current candidate vector The minimum distance between the vectors in the cluster and other candidate vectors; express The derivative of the function, used for normalization.

[0022] Furthermore, in step S2, the subject semantic vocabulary vector is obtained by using the semantic matching index, including:

[0023] The semantic conformity indexes corresponding to all candidate vectors are obtained according to the calculation formula of the semantic conformity index, and a semantic conformity index sequence is generated based on the candidate vector sequence;

[0024] Find the peak value of the semantically consistent indicator sequence, and record the vocabulary vector corresponding to the candidate vector corresponding to the peak value as the main semantic vocabulary vector;

[0025] Traverse all clusters and obtain the main semantic vocabulary vector corresponding to each cluster.

[0026] Furthermore, in step S3, the distribution of the subject semantic vocabulary vector is analyzed, including:

[0027] According to the single text data with the smallest distance in the cluster corresponding to any subject semantic vocabulary vector, the range of the current subject semantic vocabulary vector is taken;

[0028] The range is gradually expanded with a step size of 1, and the corresponding range of each main semantic vocabulary vector in the current cluster is synchronously obtained to observe the changes in the semantic conformity indicators within the range.

[0029] Furthermore, in step S3, the hierarchical truncation evaluation index is calculated, and the description hierarchies corresponding to all the subject semantic vocabulary vectors are obtained, including:

[0030] Calculate the level cutoff evaluation index, and the corresponding calculation formula is:

[0031]

[0032] in, Indicates the hierarchical truncation evaluation index; Represents the current subject semantic vocabulary vector The corresponding range; Indicates the current step length; Representation and main semantic vocabulary vector The nearest vocabulary vectors; Represents the current subject vocabulary vector The corresponding semantics meet the indicators; The current step size range The semantics in the standard deviation of the indicator;

[0033] Based on the current subject semantic vocabulary vector Gradually increasing range Calculate the corresponding level cutoff evaluation indicators respectively;

[0034] When the step size is increased to When, range Corresponding evaluation , the vocabulary vector between the starting step and the current step is recorded as the current subject semantic vocabulary vector The corresponding Qth layer describes the level;

[0035] For step length , and the step length is not included The data points are from step size The corresponding main semantic vocabulary vector Start calculating the hierarchical truncation evaluation index and repeat the steps until the current subject semantic vocabulary vector is After selecting vocabulary vectors within the range of other main semantic vocabulary vectors from the current cluster class, the description levels corresponding to all the main semantic vocabulary vectors in the current cluster are generated.

[0036] The present invention also provides a knowledge base construction device for children's neurodevelopmental problems, which is used to run the knowledge base constructed by the knowledge base construction method for children's neurodevelopmental problems as described above, and includes the following modules:

[0037] A server module, a data storage module and a network service module are used to complete the construction of a knowledge base for children's neurodevelopmental issues, and the network service module is connected to the server module and the data storage module respectively;

[0038] Terminal display module, used to realize the input and output of knowledge base;

[0039] The power supply module is used for the power operation of any module including the server module, the data storage module, the network service module and the terminal display module.

[0040] The present invention also provides a medium storing program data, which, when executed, implements the method for constructing a knowledge base for children's neurodevelopmental issues as described in any of the above items.

[0041] The present invention has the following beneficial effects:

[0042] 1. By screening the words that accurately describe children's neurodevelopmental problems in the existing database and performing word segmentation, we can obtain the description vocabulary range of different levels for the current main text. By distinguishing the knowledge retrieval scope through words of different description levels, we can divide the specific fields corresponding to the target keywords entered by the user in children's neurodevelopmental problems, increase the distinction between text data in the knowledge base, and the accuracy obtained during data retrieval, prevent the situation where similar symptoms lead to confusion in classification results, so as to improve the classification results and enhance the accuracy of knowledge data in the corresponding field of the knowledge base.

[0043] 2. The knowledge base construction device and medium for children's neurodevelopmental issues provided by the present invention have the same beneficial effects as the knowledge base construction method for children's neurodevelopmental issues provided by the present invention, and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 A schematic diagram of the steps of a method for constructing a knowledge base for children's neurodevelopmental issues provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the knowledge base construction method, device and medium for children's neurodevelopmental problems proposed by the present invention, its specific implementation method, structure, characteristics and effects in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0047] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0048] The following is a detailed description of a method, device and medium for constructing a knowledge base for children's neurodevelopmental issues provided by the present invention in conjunction with the accompanying drawings.

[0049] In the existing knowledge text splitting process, most of the similar words shared by similar diseases lead to confusion in the classification results of two similar diseases, and the deficiency of not being able to accurately retrieve the knowledge corresponding to the current disease when searching the disease knowledge in the knowledge base; in one embodiment of the present invention, a knowledge base construction method for children's neurodevelopmental problems is provided, which screens the deviation of the description of similar diseases in the problem knowledge text, and then selects the adaptive self-attention range of the words with narrative deviation when allocating words, so as to achieve the purpose of accurately classifying knowledge data; in order to realize a knowledge base construction method for children's neurodevelopmental problems, a knowledge base construction device and medium for children's neurodevelopmental problems are provided, which are essentially a software system, which is composed of modules that realize corresponding functions, and the specific steps in the method are now introduced in detail.

[0050] See also Figure 1 , which shows a schematic diagram of the steps of a method for constructing a knowledge base for children's neurodevelopmental issues provided by an embodiment of the present invention, the method comprising:

[0051] Step S1: Generate initial text data of children's neurodevelopmental problems based on the existing database, and clean and preprocess the initial text data to obtain word segmentation results;

[0052] Step S2: converting the vocabulary vector according to the word segmentation result to obtain a candidate vector sequence, judging the semantic conformity index for each candidate vector in the candidate vector sequence, and obtaining the main semantic vocabulary vector through the semantic conformity index;

[0053] Step S3: Analyze the distribution of the subject semantic vocabulary vectors, calculate the level cutoff evaluation index, and obtain the description levels of each layer corresponding to all the subject semantic vocabulary vectors;

[0054] Step S4: Acquire and mark the vocabulary corresponding to each description level in the initial text data to form the text range corresponding to each description level, obtain the storage priority of the main semantic vocabulary vector in combination with the level truncation evaluation index, mark the corresponding storage priority based on the text range, and complete the construction of the knowledge base.

[0055] It should be noted that since two similar diseases share a large number of common words in their clinical manifestations and symptom descriptions, confusion often occurs when classifying the diseases. For example, some common symptom words such as "fever", "headache", "fatigue", etc. may appear in both diseases, making it difficult for doctors or diagnostic systems to accurately distinguish them. Therefore, the similarity in vocabulary not only increases the difficulty of diagnosis, but may also lead to misdiagnosis or missed diagnosis. Therefore, in order to improve the accuracy of the distinction, a knowledge base construction method for children's neurodevelopmental problems is proposed in this embodiment.

[0056] Children's neurodevelopmental problems include attention deficit hyperactivity disorder (ADHD), autism spectrum disorder (ASD), learning disabilities, language delay, etc., which will affect children's learning ability, behavioral performance and emotional management during their growth. Therefore, a knowledge base is built for children's neurodevelopmental problems so that parents and educators can understand such diseases in a timely manner and take appropriate intervention measures to help these children better cope with challenges and realize their maximum potential.

[0057] Preferably, the initial text data includes professional literature, research materials and clinical case records.

[0058] Preferably, step S1 comprises:

[0059] Based on the existing database, professionals screened the initial text data covering children's neurodevelopmental issues;

[0060] Cleaning the initial text data by regular expressions and / or building a normal vocabulary dictionary, removing noise words to obtain second-generation text data;

[0061] Split the second-generation text data to obtain the source word segmentation results and the part-of-speech tags of the corresponding word segmentations;

[0062] The source word segmentation result is obtained by removing punctuation marks and special symbols.

[0063] It should be noted that the existing database refers to a collection of various documents, research papers, clinical reports, medical guidelines and other relevant materials, which cover the latest research results, diagnostic standards, treatment methods and related statistical data in the field of child neurodevelopment; through these materials, it can be ensured that the generated initial text data is highly scientific and authoritative, laying the foundation for subsequent data analysis and model training.

[0064] Professionals refer to experts in the field of child neurodevelopment, such as pediatricians, neurologists, psychologists and researchers in related fields. By having them carefully screen and analyze the existing database, it can be ensured that the information used in the research and analysis of child neurodevelopmental issues is comprehensive and accurate. Professionals have richer knowledge, and screening efficiency and accuracy can be guaranteed.

[0065] Regular expressions are a powerful text processing tool that allows users to match and extract information from strings through specific patterns. They are used in various scenarios, such as data validation, text search, string replacement, etc. By using various special characters and symbols, regular expressions can flexibly define complex matching rules to achieve precise control of text; and building a normal vocabulary dictionary means creating a database containing a large number of words and their related information to greatly improve the efficiency and accuracy of text processing and analysis; in the process of removing noise words, any of the above methods can be adopted, or the two methods can be combined to complement each other, that is, regular expressions are used to quickly match and extract words in the dictionary, and vocabulary dictionaries can provide rich vocabulary resources for regular expressions.

[0066] It can be explained that the initial text data in this embodiment includes professional literature, research materials and clinical case records, which can ensure the classicity and representativeness of the screening data, ensure the accuracy and effectiveness of the analysis, avoid the negative impact of data redundancy, and ensure the high quality and reliability of the data.

[0067] Specifically, firstly, initial text data for children's neurodevelopmental issues are generated through professional screening based on the existing database; then, the initial text data is cleaned in different ways to remove noise words to ensure the quality and availability of the second-generation text data, wherein noise words refer to words in the initial text data that are not related to children's neurodevelopmental issues, such as mobile phone numbers, author introductions and other information; then, the second-generation text data is split to obtain the source word segmentation results and the part-of-speech tags of the corresponding word segmentations. Preferably, the splitting is performed through the precise mode of JIEBA word segmentation, that is, through precise algorithms and dictionary matching, a section of Chinese text can be segmented into multiple meaningful words, and JIEBA (JointIncremental Chinese Word Segmentation Based on Word Frequency and Dictionary) word segmentation will try to avoid erroneous word segmentation results to ensure that each word segmentation is accurate; finally, the source word segmentation result is removed from punctuation marks and special symbols to obtain the word segmentation result, ensuring that the word segmentation result is pure and clean, and the availability of the word segmentation result is guaranteed.

[0068] For better explanation, a specific implementation method is given based on the word segmentation result. The sentence that needs word segmentation is:

[0069] NDDS all have multiple occasions of developmental defects. Clinical manifestations can lead to impairment of the individual's social, academic or vocational social adaptability, but the range of defects can vary from specific learning limitations to comprehensive impairment of social skills or intelligence.

[0070] Through JIEBA word segmentation, and removing punctuation and special symbols, the word segmentation result is:

[0071] NDDS / all / exist / developmental / deficient / clinical / management / in / multiple / occasions / that / may / lead / to / impairment / in / the / individual’s / social / academic / or / occupational / social / adaptive / ability /

[0072] / But / the / deficiencies / can / range / from / specific / learning / limitations / to / global / impairment / of / social / skills / or / intelligence.

[0073] Understandably, children's neurodevelopmental problems involve many aspects, and different words are usually used in the text to describe specific symptoms. The evaluation of these symptoms is achieved by identifying the main semantic words. However, there are deviations in the narrative of each descriptive word, so in order to solve the deviation problem, it is necessary to identify the main semantic words for analysis.

[0074] Preferably, in step S2, converting the vocabulary vector according to the word segmentation result to obtain a candidate vector sequence includes:

[0075] Step S21: converting each word in the word segmentation result into a corresponding vocabulary vector to form a vector space, clustering the vector space to obtain multiple clusters and unclassified vocabulary vectors, where the unclassified vocabulary vector is a vocabulary vector that deviates from children's neurodevelopmental problems;

[0076] Step S22: according to the preset center range of the cluster center, based on any cluster, the vocabulary vectors in the center range are recorded as candidate vectors, and the distance between each candidate vector and the cluster center is obtained respectively, and the candidate vectors are sorted in order from small to large according to the distance to generate a candidate vector sequence;

[0077] Repeat step S22 to obtain candidate vector sequences of all clusters.

[0078] To clarify, a candidate vector refers to a set of possible vectors that are preset when performing calculations or analyses, which facilitates optimization and adjustment so as to select the best vector to meet specific conditions or goals.

[0079] Specifically, in step S21, each word in the word segmentation result is converted into a corresponding vocabulary vector through the word2vec (Word to Vector) algorithm, wherein the word2vec algorithm is a natural language processing technology based on a neural network, which can map words to a high-dimensional space so that semantically similar words also have a high similarity in the vector space, that is, converting text data into numerical form to facilitate subsequent calculations and analysis; then based on the vocabulary vector and according to the detailed description and narration of the problems related to the children's neurodevelopment, a special vector space is constructed, which contains various vocabulary vectors related to the children's neurodevelopment, that is, does not include related words such as "because", "so", "but", etc., so as to better understand and analyze these problems.

[0080] Then, the current vector space is clustered using the OPTICS (Ordering Points To Identify the Clustering Structure) algorithm to obtain multiple clusters and unclassified vocabulary vectors. The OPTICS algorithm refers to an algorithm for data clustering that can identify high-density and low-density areas in the data, thereby effectively discovering the clustering structure in the data. It does not require the number of clusters to be determined in advance, can handle clusters of various densities, and can identify noise points. Unclassified vocabulary vectors are vocabulary vectors that deviate from children's neurodevelopmental problems, that is, they have a weaker correlation with children's neurodevelopmental problems, or they appear less frequently in the data set, resulting in the algorithm being unable to effectively classify them into any meaningful clusters.

[0081] In step S22, a central range is preset according to the cluster center, that is, the central range is defined to represent the range of vocabulary vectors that are close to the expression of children's neurodevelopmental problems, that is, candidate vectors are obtained; the distance between each candidate vector and the corresponding cluster center is calculated, and these distances are arranged in order from small to large to generate an ordered sequence of candidate vectors; it can be explained that the distances may be the same, in which case the candidate vectors exist at the same level and in parallel, and a judgment is made based on the order of text vocabulary, preferably the vocabulary closest to the cluster center in the text vocabulary order interval; finally, step S22 is repeated to obtain the candidate vector sequence of all clusters in step S21.

[0082] In this embodiment, the scale of vocabulary vector data in the clusters is relatively large. In the specific description, the description is mostly for the data in a single cluster, while in actual application, each cluster is analyzed.

[0083] Preferably, in step S2, the semantic conformity index is determined for each candidate vector in the candidate vector sequence, and the corresponding calculation formula is:

[0084]

[0085] in, Represents the current candidate vector The semantics of the index are consistent with the index; Represents the current candidate vector The maximum distance between the vectors in the cluster and other candidate vectors; Represents the current candidate vector The minimum distance between the vectors in the cluster and other candidate vectors; express The derivative of the function, used for normalization.

[0086] To explain, the semantic conformity index means that in the language expression, the words, sentences and paragraphs used can accurately convey the meaning that the speaker or writer wants to express. It not only requires the text to be clear and unambiguous, but also requires its content to be logically reasonable and coherent, and can be correctly understood by the reader or listener.

[0087] express The derivative of the function is used for normalization. When the independent variable in the function approaches 0, the entire function value approaches the maximum value 1, indicating that the candidate vector currently being analyzed is The concentration is worse than that of other vocabulary vectors in the cluster, and there is a part of vocabulary that may be offset, affecting the subsequent accuracy evaluation of question splitting.

[0088] Preferably, obtaining the subject semantic vocabulary vector through the semantic conformity index in step S2 includes:

[0089] The semantic conformity indexes corresponding to all candidate vectors are obtained according to the calculation formula of the semantic conformity index, and a semantic conformity index sequence is generated based on the candidate vector sequence;

[0090] Find the peak value of the semantically consistent indicator sequence, and record the vocabulary vector corresponding to the candidate vector corresponding to the peak value as the main semantic vocabulary vector;

[0091] Traverse all clusters and obtain the main semantic vocabulary vector corresponding to each cluster.

[0092] It should be noted that the main semantic vocabulary refers to the vocabulary that best reflects children's neurodevelopmental problems and carries the main information and meaning.

[0093] Specifically, first, the semantic conformity indices corresponding to all candidate vectors are obtained according to the calculation formula of the semantic conformity index, and the semantic conformity index values ​​are placed at corresponding positions in the order of the candidate vector sequence to obtain the semantic conformity index sequence; then, the peak value of the current semantic conformity index sequence is found through AMPD (Adaptive Multi-scale Peak Detection, multi-scale peak search algorithm), and the vocabulary vector corresponding to the candidate vector corresponding to the peak value is recorded as the main semantic vocabulary vector. Among them, the multi-scale peak search algorithm is a technology for identifying and locating peaks in complex data sets. By searching at different scales, it can effectively detect local maximum points in the data, that is, by gradually refining the search range and combining feature information at different scales, the accurate identification and positioning of the peak value is ensured; finally, all clusters are traversed to obtain the main semantic vocabulary vectors corresponding to each cluster.

[0094] Preferably, analyzing the distribution of subject semantic vocabulary vectors in step S3 includes:

[0095] According to the single text data with the smallest distance in the cluster corresponding to any subject semantic vocabulary vector, the range of the current subject semantic vocabulary vector is taken;

[0096] The range is gradually expanded with a step size of 1, and the corresponding range of each main semantic vocabulary vector in the current cluster is synchronously obtained to observe the changes in the semantic conformity indicators within the range.

[0097] Specifically, according to any subject semantic vocabulary vector, find the text data closest to the subject semantic vocabulary vector in the cluster, and use this distance as the initial range of the current subject semantic vocabulary vector; then gradually expand the initial range starting with a step size of 1. At the same time, synchronize operations are performed on each subject semantic vocabulary vector in the current cluster to ensure that each subject semantic vocabulary vector can obtain its corresponding range.

[0098] As the scope gradually expands, the amount of text data corresponding to other descriptive words included in the scope will gradually increase; when the scope increases to a certain extent, there will be a more significant change in the semantic conformity index, which means that the current scope has covered a part of the words that can describe the subject semantic words; in this way, the descriptive words related to the subject semantic words are effectively obtained to enrich and expand the understanding and expression of the subject semantics.

[0099] Preferably, in step S3, the hierarchical truncation evaluation index is calculated, and the description hierarchies of each layer corresponding to all subject semantic vocabulary vectors are obtained, including:

[0100] Calculate the level cutoff evaluation index, and the corresponding calculation formula is:

[0101]

[0102] in, Indicates the hierarchical truncation evaluation index; Represents the current subject semantic vocabulary vector The corresponding range; Indicates the current step length; Representation and main semantic vocabulary vector The nearest vocabulary vectors; Represents the current subject vocabulary vector The corresponding semantics meet the indicators; The current step size range The semantics in the standard deviation of the indicator;

[0103] Based on the current subject semantic vocabulary vector Gradually increasing range Calculate the corresponding level cutoff evaluation indicators respectively;

[0104] When the step size is increased to When, range Corresponding evaluation , the vocabulary vector between the starting step and the current step is recorded as the current subject semantic vocabulary vector The corresponding Qth layer describes the level;

[0105] For step length , and the step length is not included The data points are from step size The corresponding main semantic vocabulary vector Start calculating the hierarchical truncation evaluation index and repeat the steps until the current subject semantic vocabulary vector is After selecting vocabulary vectors within the range of other main semantic vocabulary vectors from the current cluster class, the description levels corresponding to all the main semantic vocabulary vectors in the current cluster are generated.

[0106] It is explained that the hierarchical truncation evaluation index is a measure of the characteristics of the surrounding vocabulary vectors in describing the main semantic vocabulary vector at the current step size, that is, by analyzing the relationship between the surrounding vocabulary vectors and the main semantic vocabulary vectors, it is judged to what extent these words can accurately describe the main semantic vocabulary vector; when the hierarchical truncation evaluation index changes significantly, it means that there is a deviation in the description of the main semantic vocabulary vector, and it is necessary to obtain a part of vocabulary with more accurate description through hierarchical truncation to ensure that the description of the main semantic vocabulary vector is more precise and reliable.

[0107] Specifically, based on the current subject semantic vocabulary vector Gradually increasing range Calculate the corresponding level cutoff evaluation index respectively, that is, the gradually increasing range Indicates that there is diversity in step length, based on the current subject semantic vocabulary vector Each step length of Calculate the corresponding level truncation evaluation index; when the step length is increased to When, range Corresponding evaluation , the vocabulary vector between the starting step and the current step is recorded as the current subject semantic vocabulary vector The corresponding Qth layer describes the level, where the starting step is 0, that is, when the increase step is When the range The corresponding level cut-off evaluation index satisfies When this judgment condition is met, the vocabulary vector between the starting step size and the current step size is recorded as the current subject semantic vocabulary vector The corresponding Qth layer describes the level, which refers to the different attributes or characteristics used to distinguish and organize data; and the step size is , and the step length is not included The data points are from step size The corresponding main semantic vocabulary vector Start calculating the level truncation evaluation index, that is, for the step size In range, but not in step length Text data in the range from and step size The corresponding subject semantic vocabulary vector Start calculating the hierarchical truncation evaluation index; then repeat the above steps, based on the current subject semantic vocabulary vector , select all other subject semantic vocabulary vectors that exist in the current cluster class; in this way, generate the description levels corresponding to all the subject semantic vocabulary vectors in the current cluster. This setting ensures that every important subject semantic vocabulary vector can be fully covered and evaluated during the entire analysis process, thereby providing a solid foundation for subsequent analysis and processing.

[0108] It can be explained that in step S4, the vocabulary corresponding to each description level is obtained in the initial text data and marked to form the text range corresponding to each description level, and the storage priority of the main semantic vocabulary vector is obtained in combination with the level truncation evaluation index, and the corresponding storage priority is marked based on the text range to complete the construction of the knowledge base.

[0109] To clarify, storage priority refers to the classification and sorting of data of different types and importance during the data storage and management process to ensure that critical data has priority access to storage resources and processing capabilities. Its setting helps to improve the efficiency and security of data processing and ensure the smooth operation of critical businesses.

[0110] A knowledge base refers to a collection of systematic, organized storage and management information. In this embodiment, the knowledge base refers to the integration of all text data on children's neurodevelopmental issues. By efficiently organizing and retrieving information, the knowledge base can help users quickly find the knowledge they need, thereby improving work efficiency and decision-making quality.

[0111] Specifically, based on the description level in the aforementioned step S3, words related to the description of the corresponding level are extracted, and these words are marked to obtain their specific positions in the text to form a clear text range, that is, to obtain a detailed annotated text data set; then, the level truncation evaluation index D within the step interval of each description level corresponding to the current main semantic vocabulary vector is averaged, and the value is normalized by sigmoid as the storage priority of the current main semantic vocabulary vector, that is, if the step interval of the currently analyzed text data is 50 to 210, the mean of the level truncation evaluation index D of each description level within this interval is calculated; finally, the corresponding storage priority is marked based on the text range to reflect the actual situation of the corresponding level storage priority, which helps to optimize the allocation of storage resources, and can also ensure that the most important data can be saved and accessed quickly within the limited storage space, and the marked text data is systematically organized and archived to complete the construction of the knowledge base.

[0112] The purpose of tagging storage priorities based on text ranges is to ensure that the actual storage priorities at each level are accurately reflected. By carefully analyzing and evaluating factors such as the importance, frequency of use, and update frequency of text content, we can assign a suitable storage priority to each text paragraph or data set. This priority tagging not only helps optimize the allocation of storage resources, but also ensures that the most important data is saved first and accessed quickly within limited storage space.

[0113] It can be understood that by screening the words in the existing database that accurately describe children's neurodevelopmental problems and performing word segmentation, we can obtain the descriptive vocabulary ranges of different levels for the current main text. By distinguishing the knowledge retrieval scope through the words of different description levels, we can divide the specific fields corresponding to the target keywords entered by the user in children's neurodevelopmental problems, increase the distinction between text data in the knowledge base, and the accuracy obtained during data retrieval, prevent the situation where similar symptoms lead to confusion in classification results, so as to improve the classification results and enhance the accuracy of knowledge data in the corresponding fields of the knowledge base.

[0114] Specifically, in actual application, after the construction of the knowledge base is completed, when the user needs to search for children's neurodevelopmental problems, that is, when the user enters the search term, the current search term is first converted into a vocabulary vector through the word2vec algorithm; then the cosine similarity between the current vocabulary vector and the subject semantic vocabulary vector stored in the knowledge base is judged, and the text data corresponding to the hierarchical storage priority of each subject semantic vocabulary vector corresponding to the level less than the current cosine similarity is extracted from the knowledge base, and the text data is displayed on the user's visualization device, and the search is completed.

[0115] To explain, cosine similarity is a method to measure the similarity between two vectors, which is achieved by calculating the cosine value of the angle between the two vectors; if the calculated cosine similarity value is between -1 and 1, when the value is 1, it means that the two vectors are exactly the same; when the value is -1, it means that the two vectors are completely opposite; when the value is 0, it means that the two vectors are orthogonal, that is, there is no correlation between the two vectors.

[0116] An embodiment of the present invention provides a knowledge base construction device for children's neurodevelopmental issues, which is used to run a knowledge base constructed by a knowledge base construction method for children's neurodevelopmental issues provided in the above embodiment, and includes the following modules:

[0117] The server module, the data storage module and the network service module are used to complete the construction of a knowledge base for children's neurodevelopmental issues, and the network service module is connected to the server module and the data storage module respectively;

[0118] Terminal display module, used to realize the input and output of knowledge base;

[0119] The power supply module is used for the power operation of any module including the server module, the data storage module, the network service module and the terminal display module.

[0120] It is clarified that the server module refers to the key component in the computer network system that is responsible for processing various requests, data storage and management of network resources. It is used to provide clients with data processing, information retrieval, file storage and sharing, application running and other services to ensure the efficient, stable and secure operation of the entire knowledge base; the server module contains a high-performance processor (CPU, Central Processing Unit) that can quickly process a large number of computing tasks; dynamic random access memory (DRAM, Dynamic Random Access Memory) is used to store temporary data and program code; combining the two, the server module can efficiently run various applications and services to meet the needs of high concurrency and big data processing.

[0121] The data storage module is responsible for saving and managing various types of data information, including hard disk drives (HDD), which store data through magnetic media and can provide higher storage capacity; solid state drives (SSD), which use flash memory technology and have the advantages of high-speed reading and writing and low power consumption; disk arrays (RAID, Redundant Array of Independent Disks) provide data redundancy and improve performance by combining multiple hard disk drives together; network attached storage (NAS, Network Attached Storage) is a storage device connected through a network, allowing users to access and share data through the network.

[0122] The network service module refers to the software component in the computer network system that is responsible for providing various network service functions; it includes switches, routers, firewalls, etc. Among them, the switch is used to connect the server and data storage module, the router is responsible for the transmission of data between networks, and the firewall is used to protect the knowledge base from external attacks.

[0123] The terminal display module refers to the module that enables users to easily obtain and process various information, including visual display devices, such as monitors, etc., which realize the output of the knowledge base and can convert the text, images and video signals output by computers or other devices into visual images so that users can intuitively see various information and data; input devices, including keyboards, mice, etc., which realize knowledge base input modules, are the main tools for users to interact with the knowledge base.

[0124] The power supply module is used to ensure the normal operation of each module to avoid data loss and equipment damage; it includes an uninterruptible power supply (UPS) that can provide short-term power when power is interrupted to ensure the normal operation of the knowledge base; and a power distribution unit (PDU) that is responsible for distributing power to servers and other devices.

[0125] It can be understood that when a module of a knowledge base construction device for children's neurodevelopmental problems is in operation, it is necessary to utilize a knowledge base construction method for children's neurodevelopmental problems provided in the aforementioned embodiment. Therefore, whether the server module, data storage module, network service module, terminal display module and power supply module are integrated or different hardware is configured to produce functions similar to the effects achieved by the present invention, they all fall within the scope of protection of the present invention.

[0126] An embodiment of the present invention further proposes a medium, which stores program data. When the program data is executed, it implements the knowledge base construction method for children's neurodevelopmental problems as described in the aforementioned embodiment; the medium has the same beneficial effects as the aforementioned knowledge base construction method for children's neurodevelopmental problems provided above, which will not be repeated here.

[0127] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0128] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for constructing a knowledge base for children's neurodevelopmental issues, characterized in that: The method comprises: Step S1: Generate initial text data of children's neurodevelopmental problems based on the existing database, and clean and preprocess the initial text data to obtain word segmentation results; Step S2: converting the vocabulary vector according to the word segmentation result to obtain a candidate vector sequence, judging the semantic conformity index for each candidate vector in the candidate vector sequence, and obtaining the main semantic vocabulary vector through the semantic conformity index; Step S3: Analyze the distribution of the subject semantic vocabulary vectors, calculate the level cutoff evaluation index, and obtain the description levels of each layer corresponding to all the subject semantic vocabulary vectors; Step S4: Acquire and mark the words corresponding to each description level in the initial text data to form the text range corresponding to each description level, obtain the storage priority of the subject semantic vocabulary vector in combination with the level cutoff evaluation index, mark the corresponding storage priority based on the text range, and complete the construction of the knowledge base; In step S2, the vocabulary vector is converted according to the word segmentation result to obtain a candidate vector sequence, including: Step S21: converting each word in the word segmentation result into a corresponding vocabulary vector to form a vector space, clustering the vector space to obtain a plurality of clusters and unclassified vocabulary vectors, wherein the unclassified vocabulary vector is a vocabulary vector that deviates from children's neurodevelopmental problems; Step S22: according to the preset center range of the cluster center, based on any cluster, the vocabulary vectors in the center range are recorded as candidate vectors, and the distance between each candidate vector and the cluster center is obtained respectively, and the candidate vectors are sorted in order from small to large according to the distance to generate a candidate vector sequence; Repeat step S22 to obtain candidate vector sequences of all clusters; In step S2, the semantic conformity index is determined for each candidate vector in the candidate vector sequence, and the corresponding calculation formula is: in, Represents the current candidate vector The semantics of the index conform to the standard; Represents the current candidate vector The maximum distance between the vectors in the cluster and other candidate vectors; Represents the current candidate vector The minimum distance between the vectors in the cluster and other candidate vectors; express The derivative of the function is used for normalization; In step S2, the subject semantic vocabulary vector is obtained by using the semantic matching index, including: The semantic conformity indexes corresponding to all candidate vectors are obtained according to the calculation formula of the semantic conformity index, and a semantic conformity index sequence is generated based on the candidate vector sequence; Find the peak value of the semantically consistent indicator sequence, and record the vocabulary vector corresponding to the candidate vector corresponding to the peak value as the main semantic vocabulary vector; Traverse all clusters and obtain the main semantic vocabulary vector corresponding to each cluster; In step S3, the distribution of the subject semantic vocabulary vector is analyzed, including: According to the single text data with the smallest distance in the cluster corresponding to any subject semantic vocabulary vector, the range of the current subject semantic vocabulary vector is taken; The range is gradually expanded with a step size of 1, and the corresponding range is synchronously obtained for each subject semantic vocabulary vector in the current cluster, and the semantic conformity index changes that appear in the range are observed; In step S3, the hierarchical truncation evaluation index is calculated, and the description hierarchies corresponding to all the subject semantic vocabulary vectors are obtained, including: Calculate the level cutoff evaluation index, and the corresponding calculation formula is: in, Indicates the hierarchical truncation evaluation index; Represents the current subject semantic vocabulary vector The corresponding range; Indicates the current step length; Representation and main semantic vocabulary vector The nearest vocabulary vectors; Represents the current subject vocabulary vector The corresponding semantics meet the indicators; The current step size range The semantics in the standard deviation of the indicator; Based on the current subject semantic vocabulary vector Gradually increasing range Calculate the corresponding level cutoff evaluation indicators respectively; When the step size is increased to When, range Corresponding level cutoff evaluation index , the vocabulary vector between the starting step and the current step is recorded as the current subject semantic vocabulary vector The corresponding Qth layer describes the level; For step length , and the step length is not included The data points are from step size The corresponding main semantic vocabulary vector Start calculating the hierarchical truncation evaluation index and repeat the steps until the current subject semantic vocabulary vector is Select vocabulary vectors within the range of other main semantic vocabulary vectors from the current cluster class to generate description levels corresponding to all main semantic vocabulary vectors in the current cluster.

2. A method for constructing a knowledge base for children's neurodevelopmental issues as claimed in claim 1, characterized in that: The initial text data includes professional literature, research materials and clinical case records.

3. A method for constructing a knowledge base for children's neurodevelopmental issues as claimed in claim 2, characterized in that: Step S1 includes: Based on the existing database, professionals screened the initial text data covering children's neurodevelopmental issues; Cleaning the initial text data by regular expressions and / or building a normal vocabulary dictionary, removing noise words to obtain second-generation text data; Split the second-generation text data to obtain the source word segmentation results and the part-of-speech tags of the corresponding word segmentations; The source word segmentation result is obtained by removing punctuation marks and special symbols.

4. A knowledge base construction device for children's neurodevelopmental issues, characterized in that: The knowledge base constructed by the method for constructing a knowledge base for children's neurodevelopmental problems according to claim 1 comprises the following modules: A server module, a data storage module and a network service module are used to complete the construction of a knowledge base for children's neurodevelopmental issues, and the network service module is connected to the server module and the data storage module respectively; Terminal display module, used to realize the input and output of knowledge base; The power supply module is used for the power operation of any module including the server module, the data storage module, the network service module and the terminal display module.

5. A medium, characterized in that The medium stores program data, and when the program data is executed, the method for constructing a knowledge base for children's neurodevelopmental issues as described in any one of claims 1 to 3 is implemented.

Images