An intelligent retrieval system for a digital medical knowledge base

Abstract

The present invention discloses an intelligent retrieval system for a digital medical knowledge base, which relates to the field of intelligent retrieval. The present invention comprehensively analyzes the authoritative information and user usage information of the search interface, thereby obtaining the recommendation values of each search interface. When the user conducts a retrieval, the search interfaces are arranged from top to bottom according to the size of the recommendation values, improving the accuracy of the search and increasing user stickiness; by performing timed deletion and addition operations on the database, the database memory is released, and at the same time, it is ensured that users can retrieve the latest medical achievements; through the comprehensive analysis of customer service, the present invention can reasonably allocate customer service to users, and finally gives the choice to the user in the final selection link, improving user satisfaction.

Description

Technical Field

[0001] The present invention relates to the field of intelligent retrieval systems, and particularly to an intelligent retrieval system for a digital medical knowledge base. Background Art

[0002] With the wide application of information technology in the medical field, medical data has shown explosive growth. Various types of data, including electronic medical records, medical images, inspection reports, clinical guidelines, medical literature, etc., have been continuously accumulated. These data contain rich medical knowledge and clinical experience. However, due to the large amount of data and its dispersion in different systems and departments, the retrieval quality of traditional retrieval methods cannot be guaranteed, and users cannot distinguish the quality, which may lead to users receiving incorrect medical information and threatening the lives of users.

[0003] Moreover, the existing retrieval systems have a low retrieval accuracy rate, which easily causes users to repeatedly retrieve multiple times, resulting in a decline in user satisfaction.

[0004] Therefore, an intelligent retrieval system for a digital medical knowledge base is introduced. Summary of the Invention

[0005] In view of this, the present invention provides an intelligent retrieval system for a digital medical knowledge base to solve the problems raised in the above background art.

[0006] The object of the present invention can be achieved through the following technical solutions: An intelligent retrieval system for a digital medical knowledge base includes:

[0007] A data acquisition module: acquiring authoritative information and user usage information of the search interface; the authoritative information includes: the domain name weight of the search interface, the cited authors and the number of their literatures, and the number of times the search interface is cited; the user usage information includes: the user rating of the retrieval interface, the browsing duration of the user at each level, and the number of times the user forwards; and sending the authoritative information and user usage information of the search interface to the search interface evaluation module.

[0008] A search interface evaluation module: respectively obtaining the search interface authority value and the user usage value through the analysis of the authoritative information and user usage information of the interface; and then obtaining the recommended value of the search interface through the comprehensive analysis of the search interface authority value and the user usage value.

[0009] A database update module: setting an update time interval during the operation of the intelligent retrieval system, and deleting and adding search interfaces in the medical knowledge base when the update time interval point is reached.

[0010] A user interaction module: users can send retrieval instructions to the intelligent retrieval system through various interaction methods to obtain the information required by the users.

[0011] Interactive interface setting module: Based on the recommended values of each search interface, when the user issues a retrieval instruction, the matched search interfaces are arranged in descending order of the corresponding recommended values from top to bottom.

[0012] Artificial customer service module: Triggered when the user makes the same retrieval instruction multiple times within a short period, and the user can freely choose whether to connect to the artificial customer service.

[0013] In some embodiments, the authoritative value of the search interface is obtained as follows:

[0014] S1: Obtain the authoritative scores of the authors of the search interface as follows:

[0015] S1-1: Obtain the authors of each cited document in the search interface, obtain the number of papers published by each author in each journal, and obtain the impact factor of each journal from the Journal Citation Reports, denoted as , i = 1, 2...x, where x is the total number of journals; denote the number of papers published by each author in each journal as , j = 1, 2...n, where n is the total number of cited authors in the current search interface, and substitute the obtained impact factors of each journal and the number of papers published by each author in the corresponding journal into the formula: Thus, the journal scores of each author are obtained ;

[0016] S1-2: Obtain the award types of each author in the search interface and the corresponding number of awards. The award types are divided into: international, national, and provincial awards; preset the weight impact factors corresponding to each award type, multiply the number of awards of each type by the weight factor of the corresponding award type, and then sum them up to obtain the award scores of each author ;

[0017] S1-3: Obtain the H-index of each author in the search interface, denoted as the impact score , preset the journal scores of each author , award scores and the impact scores corresponding weight impact factors, multiply the journal scores , award scores and the impact scores of each author by their respective weight impact factors and sum them up. The result obtained is used as the authoritative index of each author ;

[0018] S2: Multiply the authoritative index of each author in the search interface by the number of works of the corresponding author, and then sum them up to obtain the literature comprehensive score WXP of the search interface;

[0019] S3: Preset the weight influence factors corresponding to the domain name weight of the search interface, the number of citations of the search interface, and the comprehensive literature score WXP of the search interface. Multiply the domain name weight of the search interface, the number of citations of the search interface, and the comprehensive literature score WXP of the search interface by their respective weight influence factors and then sum them up. The result obtained is used as the authority value ZHP of the search interface.

[0020] In some embodiments, the user usage value of the search interface is obtained as follows:

[0021] Taking a single user's single use of the intelligent retrieval system as the analysis interval, statistically analyze the browsing duration of the single user at each level within the search interface. Preset each group of duration intervals corresponding to the browsing duration, and based on the different levels, preset the scores corresponding to each duration interval of each level; match the duration intervals corresponding to the browsing duration of the user at each level, so as to obtain the scores corresponding to the browsing duration of the user at each level, and sum up the scores corresponding to the browsing duration of the user at each level, so as to obtain the duration usage score SC of the user;

[0022] Obtain the rating values and rating times of each user for the search interface, calculate the difference between the rating time of each user and the current time, so as to obtain the rating interval time of each user. Preset each group of rating interval time intervals corresponding to each group of credibility indicators, match the rating interval time intervals corresponding to the rating interval time of each user, so as to obtain the credibility indicators corresponding to the rating interval time of each user, multiply the rating values of each user for the search interface and the corresponding credibility indicators, so as to obtain the adjusted rating values of the search interface for each user, and take the average value of the adjusted rating values of the search interface of all users as the adjusted rating JQ;

[0023] Mark the user forwarding times as ZF, normalize the obtained duration usage score SC, adjusted rating JQ, and user forwarding times ZF and substitute them into the formula: Thus, the user usage value YHY is obtained, where z1, z2, and z3 are the weight influence factors corresponding to the duration usage score SC, weighted rating JQ, and user forwarding times ZF respectively.

[0024] In some embodiments, the recommendation value of the search interface is obtained as follows:

[0025] Normalize the obtained authority value ZHP and user usage value YHY of the search interface. Preset the weight influence factors corresponding to the authority value ZHP and user usage value YHY of the search interface. Multiply the authority value ZHP and user usage value YHY of the search interface by their respective preset weight influence factors and then sum them up. The result obtained is used as the recommendation value TJZ of the search interface.

[0026] In some embodiments, a deletion operation is performed on the search interface in the medical database. Specifically:

[0027] The search interfaces in the medical database are classified according to disease types, which include but are not limited to: cardiovascular and cerebrovascular diseases, pneumonia, colds, etc. During the operation of the medical database, a maintenance node is set. When reaching the maintenance node, the m search interfaces with the smallest recommended values TJZ under various disease types are extracted, where m > 5. The most recent usage score and the most recent usage time of each obtained search interface are acquired. The difference between the most recent usage time of each obtained search interface and the current maintenance node time is calculated to obtain the stop usage duration SJ of each search interface. The most recent usage score of the user is marked as PF. The stop usage duration SJ, the interface recommended value TJZ, and the most recent usage score PF of each search interface are normalized and then substituted into the formula: Thereby obtaining the deletion judgment index SCP of each search interface, where c1, c2, and c3 are the weight influence factors corresponding to the interface recommended value TJZ, the most recent usage score mark PF, and the stop usage duration SJ of the interface respectively. The n search interfaces with the lowest deletion judgment index SCP under each disease type are selected for deletion operation, where 2 < n < m.

[0028] In some embodiments, an addition operation is performed on the medical database. Specifically:

[0029] When reaching the maintenance node of the database, search interfaces related to medicine that do not exist in the current medical knowledge base are obtained from the external resource library. A time update value is set, and the time difference between the obtained search interfaces and the current maintenance node is calculated. When the time difference is less than the time update value, the corresponding search interfaces are directly added to the current medical knowledge base; when the time difference is greater than the time update value, a search interface recommended value threshold is preset, and the recommended value of the search interfaces with a time difference greater than the time update value is calculated. The recommended value of the search interfaces with a time difference greater than the time update value is compared with the preset search interface recommended value threshold. When the recommended value of the search interfaces with a time difference greater than the time update value is greater than the preset search interface recommended value threshold, the corresponding search interfaces are added to the current medical knowledge base.

[0030] In some embodiments, a suitable customer service staff is selected for the user. Specifically:

[0031] Set time nodes within a preset time interval, obtain the number of customers received, problem-solving rate, and each customer rating of each customer service staff at each time node. Take the average rating of each customer at each time node as the sub-average value corresponding to that time node. Preset the weight influence factors corresponding to the sub-average value, the number of customers received, and the problem-solving rate. Multiply the sub-average value, the number of customers received, and the problem-solving rate at each time node by the preset weight influence factors respectively and then sum them up. The final result obtained is used as the comprehensive customer rating KFP at the corresponding time node.

[0032] Extract four values from the customer service comprehensive rating KFP at each time node in the time interval, namely the value corresponding to the first time node, the value corresponding to the last time node, the maximum value, and the minimum value. Use the result obtained by subtracting the value corresponding to the first time node from the minimum value as the first analysis value W1. Use the result obtained by subtracting the maximum value from the value corresponding to the last time node as the second analysis value W2. Use the result obtained by subtracting the value corresponding to the first time node from the value corresponding to the last time node as the third analysis value W3. Substitute the obtained first analysis value W1, second analysis value W2, and third analysis value W3 into the formula: Thus, obtain the rating correction value corresponding to each customer service staff, where s1, s2, and s3 are the weight influence factors corresponding to the first analysis value W1, the second analysis value W2, and the third analysis value W3 respectively.

[0033] Take the average value of the customer service comprehensive rating KFP at each time node as the service rating of the corresponding customer service staff. Add the service rating of the customer service staff to the rating correction value of the corresponding customer service staff, thereby obtaining the corrected service rating T2 of each customer service staff.

[0034] When the user performs multiple identical search commands within a short period of time, trigger customer service intervention. The user can freely choose whether to connect to the artificial customer service. When the user chooses to connect to the artificial customer service, obtain the number of pending tasks T1, the corrected service rating T2, and the cumulative online duration T3 of each currently online customer service staff on the same day. Use the formula: Obtain the customer service reception index KFJ of each currently online customer service staff. Extract the top three customer service staff in terms of the customer service reception index KFJ among the currently online customer service staff. Obtain the age, working years, number of pending customers, avatar, and positive rating information of the corresponding customer service staff and package them and send them to the user's interaction interface for the user to independently choose which customer service staff to connect to.

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] The present invention comprehensively analyzes from two aspects of the authority and user usability of the search interface, thereby obtaining the recommended value of each search interface. When the user uses the retrieval system to perform a search, arrange the search interfaces in descending order according to the recommended value, avoiding the time loss and health loss brought to the user by receiving some messy information and information unrelated to diseases, and increasing the reliability of the retrieval effect.

[0037] By periodically deleting and adding search interfaces in the database, the present invention can ensure that users can retrieve the latest research results, improving the timeliness of data update; by analyzing the deletion judgment index of search interfaces in the database, some useless search interfaces are reasonably deleted, thereby periodically releasing the database memory and avoiding the bad experience brought by users retrieving such irrelevant information;

[0038] Through the comprehensive analysis of customer service, the present invention obtains the customer service reception index corresponding to each customer service, selects the top three customer services in terms of the customer service reception index, and packages and sends the relevant information to the user's interaction interface, allowing the user to independently choose which customer service to connect to, improving user satisfaction and customer stickiness. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In the following description of exemplary embodiments in conjunction with the drawings, more details, features and advantages of the present application are disclosed. In the drawings:

[0040] Figure 1 is the principle block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] The following will describe several embodiments of the present application in more detail with reference to the drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and purposes and should not be limited to the embodiments described herein. These embodiments are provided to make the present application comprehensive and complete and to fully convey the scope of the present application to those skilled in the art. The embodiments do not limit the present application.

[0042] Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the relevant art and / or the context of this specification, and will not be interpreted in an idealized or overly formal sense unless clearly defined herein.

[0043] Please refer to Figure 1 as shown, an intelligent retrieval system for a digital medical knowledge base includes: a data acquisition module, a search interface evaluation module, a database update module, a user interaction module, an interaction interface setting module, and an artificial customer service module;

[0044] Data acquisition module: Obtain the authoritative information and user usage information of the search interface; the authoritative information includes: the domain name weight of the search interface, the cited authors and the number of their literatures, and the number of times the search interface is cited; the user usage information includes: the user rating of the retrieval interface, the browsing duration of the user at each level, and the number of times the user forwards; and send the authoritative information and user usage information of the search interface to the search interface evaluation module;

[0045] Search interface evaluation module: Obtain the search interface authority value and user usage value respectively through the analysis of the authoritative information and user usage information of the interface; then obtain the recommendation value of the search interface through the comprehensive analysis of the search interface authority value and user usage value;

[0046] S1: Obtain the authoritative scores of the authors of the search interface, specifically as follows:

[0047] S1-1: Obtain the authors of each cited literature of the search interface, obtain the number of papers published by each author in each journal, and obtain the impact factor of each journal from the Journal Citation Reports, denoted as , i = 1, 2...x, x is the total number of journals; denote the number of papers published by each author in each journal as , j = 1, 2...n, n is the total number of cited authors of the current search interface, and substitute the obtained impact factors of each journal and the number of papers published by each author in the corresponding journal into the formula: Thus, obtain the journal scores of each author ;

[0048] S1-2: Obtain the award types and the corresponding number of awards of the authors of the search interface. The award types are divided into: international, national, and provincial awards; preset the weight impact factors corresponding to each award type, multiply the number of awards of each type by the weight factor of the corresponding award type, and then add them up to obtain the award scores of each author ;

[0049] Set different weight impact factors for different award types, taking into account the different gold contents of different types of awards, so that the award scores can better reflect the award situation of the corresponding authors;

[0050] S1-3: Obtain the H-index of the authors of the search interface, denoted as the impact score , preset the weight impact factors corresponding to the journal scores 、award scores and impact scores of each author, and substitute the journal scores 、award scores and impact scores Sum them up after multiplying by their respective weight influence factors, and use the resulting value as the authority index of each author. ;

[0051] By calculating the authority index of each author , it is possible to evaluate the authority of each author in the medical field within the search interface. The higher the authority index , the higher the authority of the corresponding author in the medical field, and the higher the gold content of their works;

[0052] S2: Multiply the authority index of each author in the search interface by the number of works of the corresponding author, and then sum them up to obtain the comprehensive literature score WXP of the search interface;

[0053] S3: Preset the weight influence factors corresponding to the domain name weight of the search interface, the number of citations of the search interface, and the comprehensive literature score WXP of the search interface. Multiply the domain name weight of the search interface, the number of citations of the search interface, and the comprehensive literature score WXP of the search interface by their respective weight influence factors and then sum them up. The resulting value is used as the authority value ZHP of the search interface;

[0054] Comprehensively judge the authority value of a search interface from three different dimensions, making the result more reliable and avoiding the contingency caused by a single factor;

[0055] Taking a single user's single use of the intelligent retrieval system as the analysis interval, statistically analyze the browsing duration of the single user at each level within the search interface. Preset the respective duration intervals corresponding to the browsing duration, and based on the different levels, preset the scores corresponding to each duration interval at each level; for example, for a browsing duration interval of 3 - 5 minutes, the score set by the user at the first level is 5, the score set at the second level is 8, and the score set at the third level is 13;

[0056] Match the duration interval corresponding to the browsing duration of the user at each level to obtain the score corresponding to the browsing duration of the user at each level, and sum up the scores corresponding to the browsing duration of the user at each level to obtain the user's duration usage score SC;

[0057] The higher the level the user browses and the longer the time the user stays on the interface with a higher level, it indicates that the information provided by the search interface is more accurate. Different scores are set at different levels for the same duration interval, considering the influence of the level factor on the search interface, making the analysis of the browsing duration more accurate;

[0058] Obtain the rating values and rating times of each user for the search interface. Subtract the rating time of each user from the current time to obtain the rating interval time of each user. Preset that each group of rating interval time ranges corresponds to each group of credibility indicators. Match the rating interval time range corresponding to the rating interval time of each user to obtain the credibility indicator corresponding to the rating interval time of each user. Multiply the rating value of each user for the search interface by the corresponding credibility indicator to obtain the adjusted rating value of the search interface for each user. Take the average of the adjusted rating values of all users as the adjusted rating JQ;

[0059] The search interface is constantly improving, and previous ratings cannot fully represent the present. Setting different credibility indicators for each group of rating interval time ranges takes into account the influence of time factors on the rating values, making the adjusted rating better reflect the rating situation of users for the interface;

[0060] Mark the number of user forwards as ZF, and substitute the obtained duration into the formula after normalizing the usage rating SC, adjusted rating JQ, and the number of user forwards ZF: Thus, obtain the user usage value YHY, where z1, z2, and z3 are the weight influence factors corresponding to the usage rating SC, weighted rating JQ, and the number of user forwards ZF respectively;

[0061] Normalize the obtained authority value ZHP of the search interface and the user usage value YHY. Preset the weight influence factors corresponding to the authority value ZHP and the user usage value YHY of the search interface. Multiply the authority value ZHP and the user usage value YHY of the search interface by their respective preset weight influence factors and then sum them. The result obtained is used as the recommended value TJZ of the search interface;

[0062] The higher the authority of the search interface is not necessarily the better. The user's perception of the search interface is also an important consideration in evaluating the search interface. Considering both the authority and the user usage value of the search interface and obtaining the recommended value of the search interface by setting weight influence factors can take care of more users;

[0063] Database update module: Set an update time interval during the operation of the intelligent retrieval system. When the update time interval point is reached, delete and add the search interface in the medical knowledge base;

[0064] Classify the search interfaces in the medical database according to disease types, which include but are not limited to: cardiovascular and cerebrovascular diseases, pneumonia, colds, etc. Set up a maintenance node during the operation of the medical database. When reaching the maintenance node, extract the m search interfaces with the smallest recommended values TJZ under various disease types, where m > 5. Obtain the most recent usage score and the most recent usage time of each obtained search interface. Calculate the difference between the most recent usage time of each obtained search interface and the current maintenance node time to obtain the discontinuation duration SJ of each search interface. Mark the user's most recent usage score as PF. Normalize the discontinuation duration SJ, interface recommended value TJZ, and the most recent usage score PF of each search interface and substitute them into the formula: Thus, obtain the deletion judgment index SCP of each search interface, where c1, c2, and c3 are the weight influence factors corresponding to the interface recommended value TJZ, the most recent usage score mark PF, and the discontinuation duration SJ of the interface respectively. Select the n search interfaces with the lowest deletion judgment index SCP under each disease type for deletion operation, where 2 < n < m;

[0065] By regularly deleting useless search interfaces in the database, it is possible to release the database memory and avoid database explosion;

[0066] When reaching the maintenance node of the database, obtain the medical-related search interfaces that do not exist in the current medical knowledge base in the external resource library. Set a time update value. Calculate the time difference between the obtained search interfaces and the current maintenance node. When the time difference is less than the time update value, directly add the corresponding search interfaces to the current medical knowledge base; when the time difference is greater than the time update value, preset a search interface recommended value threshold. Calculate the recommended value of the search interfaces with a time difference greater than the time update value. Compare the recommended value of the search interfaces with a time difference greater than the time update value with the preset search interface recommended value threshold. When the recommended value of the search interfaces with a time difference greater than the time update value is greater than the preset search interface recommended value threshold, add the corresponding search interfaces to the current medical knowledge base;

[0067] By increasing the content of the medical knowledge base, it is possible to enable users to receive the latest medical knowledge and improve the timeliness of information update;

[0068] User interaction module: Users can issue retrieval instructions to the intelligent retrieval system through various interaction methods to obtain the information they need;

[0069] Interaction interface setting module: Based on the recommended values of each search interface, when a user issues a retrieval instruction, arrange the matching search interfaces in descending order of the corresponding recommended values from top to bottom;

[0070] Artificial customer service module: Triggered when a user makes the same retrieval instruction multiple times within a short period. The user can freely choose whether to connect to the artificial customer service;

[0071] Set time nodes on a preset time interval, obtain the number of customers received, problem-solving rate, and each customer's score of each customer service staff at each time node. Take the average score of each customer at each time node as the sub-average value at the corresponding time node. Preset the weight influence factors corresponding to the sub-average value, the number of customers received, and the problem-solving rate. Multiply the sub-average value, the number of customers received, and the problem-solving rate at each time node by the preset weight influence factors respectively and then sum them up. The final result obtained is used as the comprehensive customer score KFP at the corresponding time node;

[0072] Judge the excellence of the customer service from multiple dimensions. The larger the comprehensive customer service score KFP, the higher the service ability of the customer service;

[0073] Extract four values from the comprehensive customer service scores KFP of each time node in the time interval, namely the value corresponding to the first time node, the value corresponding to the last time node, the maximum value, and the minimum value. Use the result obtained by subtracting the value corresponding to the first time node from the minimum value as the first analysis value W1; use the result obtained by subtracting the maximum value from the value corresponding to the last time node as the second analysis value W2; use the result obtained by subtracting the value corresponding to the first time node from the value corresponding to the last time node as the third analysis value W3. Substitute the obtained first analysis value W1, second analysis value W2, and third analysis value W3 into the formula: Thus, the score correction value corresponding to each customer service is obtained, where s1, s2, and s3 are the weight influence factors corresponding to the first analysis value W1, the second analysis value W2, and the third analysis value W3 respectively;

[0074] Take the average value of the comprehensive customer service scores KFP at each time node as the service score of the corresponding customer service. Add the service score of the customer service to the score correction value of the corresponding customer service to obtain the corrected service score T2 of each customer service;

[0075] When a user makes multiple identical retrieval instructions within a short period, the customer service is triggered to intervene. The user can freely choose whether to connect to the artificial customer service; when the user chooses to connect to the artificial customer service, obtain the number of pending tasks T1, the corrected service score T2, and the cumulative online duration T3 of each current online customer service. Use the formula: Obtain the customer service reception index KFJ of each current online customer service. Extract the top three customer services in terms of the customer service reception index KFJ among the current online customer services. Obtain the age, working years, number of pending customers, avatar, and positive review rate information of the corresponding customer services and package them and send them to the user's interaction interface for the user to independently choose which customer service to connect to;

[0076] Through the comprehensive analysis of the number of people T1 waiting to be processed, the corrected service score T2, and the cumulative online duration T3 of each customer service on the same day, it is possible to reasonably allocate customer services to the users accessing the customer service channel at the current time, avoid excessive waiting time for users, and give the right to choose which customer service to the users ultimately, which can improve user satisfaction;

[0077] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only the specific embodiments. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. An intelligent retrieval system for a digital medical knowledge base, characterized in that: include: Data acquisition module: obtain authoritative information and user usage information of the search interface; Authoritative information includes: search interface domain name weight, cited authors and the number of their documents, and the number of citations of the search interface; user usage information includes: user ratings of the search interface, user browsing time at each level, and user forwarding times; and the authoritative information and user usage information of the search interface are sent to the search interface evaluation module; Search interface evaluation module: obtains the search interface authority value and user usage value through analysis of the interface authority information and user usage information respectively; then obtains the search interface recommendation value through comprehensive analysis of the search interface authority value and user usage value; Database update module: set the update time interval during the operation of the intelligent retrieval system, and delete and add the search interface in the medical knowledge base when the update time interval is reached; User interaction module: Users can issue search instructions to the intelligent search system through a variety of interactive methods to obtain the information they need; Interactive interface setting module: based on the recommended values ​​of each search interface, when the user issues a search command, the matched search interfaces are arranged in order from top to bottom according to the size of the corresponding recommended values; Manual customer service module: It is triggered when a user performs the same search command multiple times in a short period of time. The user can freely choose whether to access manual customer service.

2. According to claim 1, the intelligent retrieval system for a digital medical knowledge base is characterized in that: Get the authority value of the search interface, specifically: S1: Get the authoritative score of each author in the search interface, as follows: S1-1: Get the authors of each cited document in the search interface, the number of papers published by each author in each journal, and the impact factor of each journal, recorded as , i=1,2...x, x is the total number of journals; the number of papers published by each author in each journal is , j=1, 2...n, n is the total number of authors cited in the current search interface, and the impact factor of each journal is obtained and the number of papers published by each author in the corresponding journal Substituting into the formula: Thus, we can get the journal score of each author. ; S1-2: Get the award type and corresponding number of awards of each author in the search interface. The award types are divided into: international, national and provincial awards; preset the weight impact factor corresponding to each award type, multiply the number of awards of each type by the weight factor of the corresponding award type, and then add them up to get the award score of each author ; S1-3: Get the H index of each author in the search interface and record it as the impact score , preset journal scores for each author , winning score And the impact score The corresponding weighted impact factor is the journal score of each author. , winning score And the impact score The author's authority index is obtained by multiplying the author's weighted impact factor and summing them up. ; S2: The authority index of each author in the search interface Multiply it by the number of works of the corresponding author, and then add them up to get the comprehensive literature score WXP on the search interface; S3: Preset the weight impact factors corresponding to the search interface domain name weight, the number of citations of the search interface and the comprehensive literature score WXP of the search interface. Multiply the search interface domain name weight, the number of citations of the search interface and the comprehensive literature score WXP of the search interface with their respective corresponding weight impact factors and sum them up. The result is used as the authority value ZHP of the search interface.

3. The intelligent retrieval system for a digital medical knowledge base according to claim 2, characterized in that: Get the user usage value of the search interface, specifically: Taking a single user's single use of the intelligent retrieval system as the analysis interval, statistically analyzing the browsing time of a single user at each level in the search interface within the interval, presetting each group of time intervals corresponding to the browsing time, and presetting each group of scores corresponding to each time interval of each level based on the different levels; matching the time intervals corresponding to the browsing time of the user at each level, thereby obtaining the scores corresponding to the browsing time of the user at each level, and accumulating the scores corresponding to the browsing time of the user at each level, thereby obtaining the user's time usage score SC; Obtain each user's rating value and rating time for the search interface, subtract each user's rating time from the current time to obtain each user's rating interval, preset each group of rating interval time intervals corresponding to each group of credibility indicators, match each user's rating interval time interval corresponding to each user's rating interval time, thereby obtaining each user's credibility indicator corresponding to the rating interval time, multiply each user's rating value for the search interface by the corresponding credibility indicator, thereby obtaining each user's search interface adjusted rating value, and take the average of all users' search interface adjusted rating values ​​as the adjusted rating JQ; The number of times the user forwards is marked as ZF, and the obtained duration usage score SC, adjustment score JQ and the number of times the user forwards ZF are normalized and entered into the formula: Thus, the user usage value YHY is obtained, where z1, z2 and z3 are the weight influence factors corresponding to the duration usage score SC, the weighted score JQ and the number of user forwarding times ZF respectively.

4. The intelligent retrieval system for a digital medical knowledge base according to claim 3, characterized in that: Get the recommended values ​​for the search interface, specifically: The obtained authority value ZHP and user usage value YHY of the search interface are normalized, and the corresponding weight influence factors of the authority value ZHP and user usage value YHY of the search interface are preset. The authority value ZHP and user usage value YHY of the search interface are multiplied by their respective preset weight influence factors and then summed up, and the result is used as the recommended value TJZ of the search interface.

5. The intelligent retrieval system for a digital medical knowledge base according to claim 4, characterized in that: Delete the search interface in the medical database, specifically: The search interfaces in the medical database are classified according to the disease type. When the update time interval of the database is reached, the m search interfaces with the smallest search interface recommendation value TJZ under various disease types are extracted, where m>5. The most recent use score and the most recent use time of each search interface are obtained. The most recent use time of each search interface is subtracted from the current maintenance node time to obtain the stop use time SJ of each search interface. The user's most recent use score is marked as PF. The stop use time SJ of each search interface, the interface recommendation value TJZ, and the most recent use score PF are normalized and entered into the formula: Thus, the deletion judgment index SCP of each search interface is obtained, where c1, c2 and c3 are the weight influence factors corresponding to the interface recommendation value TJZ, the most recent use score mark PF and the interface stop use time SJ respectively. The n search interfaces with the lowest deletion judgment index SCP under each disease type are selected for deletion operation. <n<m。 6. The intelligent retrieval system for a digital medical knowledge base according to claim 5, characterized in that: Perform additional operations on the medical database, specifically: When the update time interval of the database is reached, a medical-related search interface that does not exist in the current medical knowledge base is obtained from the external resource library, a time update value is set, and the time difference between the obtained search interface and the current maintenance node is calculated. When the time difference is less than the time update value, the corresponding search interface is directly added to the current medical knowledge base; When the time difference is greater than the time update value, a search interface recommendation value threshold is preset, the recommended value of the search interface where the time difference is greater than the time update value is calculated, and the recommended value of the search interface corresponding to the time difference being greater than the time update value is compared with the preset search interface recommendation value threshold; when the recommended value of the search interface corresponding to the time difference being greater than the time update value is greater than the preset search interface recommendation value threshold, the corresponding search interface is added to the current medical knowledge base.

7. The intelligent retrieval system for a digital medical knowledge base according to claim 6, characterized in that: Select appropriate customer service personnel for users, specifically: Set time nodes in the preset time interval, obtain the number of customers received, problem solving rate and customer scores of each customer service staff at each time node, take the average score of each customer at each time node as the average score at the corresponding time node, preset the weight influence factors corresponding to the average score, number of customers received and problem solving rate, multiply the average score, number of customers received and problem solving rate at each time node by the preset weight influence factors respectively, and then sum them up, and the final result is taken as the customer comprehensive score KFP at the corresponding time node; Extract the four values ​​of the comprehensive customer service score KFP at each time node in the time interval, which are the value corresponding to the first time node, the value corresponding to the last time node, the maximum value and the minimum value; subtract the value corresponding to the first time node from the minimum value as the first analysis value W1; subtract the maximum value from the value corresponding to the last time node as the second analysis value W2; subtract the value corresponding to the first time node from the value corresponding to the last time node as the third analysis value W3, and substitute the first analysis value W1, the second analysis value W2 and the third analysis value W3 into the formula: Thus, the score correction value corresponding to each customer service is obtained, where s1, s2 and s3 are the weight influence factors corresponding to the first analysis value W1, the second analysis value W2 and the third analysis value W3 respectively; Take the average of the customer service comprehensive score KFP at each time node as the service score of the corresponding customer service, add the customer service score and the score correction value of the corresponding customer service, and thus obtain the corrected service score T2 of each customer service; When the user chooses to access manual customer service, the number of people to be processed T1, the corrected service score T2, and the cumulative online time T3 of each online customer service are obtained, using the formula: Get the customer service reception index KFJ of each current online customer service, extract the top three customer service staff with the customer service reception index KFJ among the current online customer service staff, obtain the age, years of experience, number of people to be received, avatar and praise rate of the corresponding customer service staff, package and send to the user's interactive interface, and the user can choose which customer service to connect to.

Images