CN113192619A

CN113192619A - Object matching method, device, equipment and storage medium

Info

Publication number: CN113192619A
Application number: CN202110580218.9A
Authority: CN
Inventors: 刘宗节
Original assignee: Beijing Jingdong Tuoxian Technology Co Ltd
Current assignee: Beijing Jingdong Tuoxian Technology Co Ltd
Priority date: 2021-05-26
Filing date: 2021-05-26
Publication date: 2021-07-30

Abstract

The embodiment of the invention discloses an object matching method, device, equipment and storage medium. The method comprises the following steps: when an object matching event is detected, acquiring information to be matched of an object to be matched corresponding to the object matching event and candidate matching information of each candidate matching object; aiming at the candidate matching information of each candidate matching object, determining the matching degree between the object to be matched and the candidate matching object according to the information to be matched and the candidate matching information; determining a target matching object matched with the object to be matched from the candidate matching objects according to the matching degree of the candidate matching objects; the information to be matched can comprise required skill information and/or service timeliness requirement information, and the candidate matching information can comprise available skill information and/or service timeliness providing information. According to the technical scheme of the embodiment of the invention, the target matching object with higher matching degree in the aspects of skill and/or service timeliness can be matched for the object to be matched.

Description

Object matching method, device, equipment and storage medium

Technical Field

The embodiment of the invention relates to the field of computer application, in particular to an object matching method, device, equipment and storage medium.

Background

With the rapid development of the internet, an online inquiry medical system based on the internet hospital is gradually emerging, and users can perform online inquiry through the internet hospital.

In practical applications, a user often performs online inquiry through an inquiry function in an online inquiry medical system, and when the inquiry function is triggered, a triage assignment module in the online inquiry medical system assigns (i.e., matches) doctors to the user for receiving a consultation according to the comprehensive scores of the doctors.

In the process of implementing the invention, the inventor finds that the following technical problems exist in the prior art: at present, the matching degree between the matched doctor (i.e. the target matching object) and the user (i.e. the object to be matched) is low, which may result in high referral rate, high return rate, and the like, and the user experience is poor.

Disclosure of Invention

The embodiment of the invention provides an object matching method, device, equipment and storage medium, which aim to realize the effect of matching a target matching object with higher matching degree for an object to be matched.

In a first aspect, an embodiment of the present invention provides an object matching method, which may include:

when an object matching event is detected, acquiring information to be matched of an object to be matched corresponding to the object matching event and candidate matching information of each candidate matching object;

aiming at the candidate matching information of each candidate matching object, determining the matching degree between the object to be matched and the candidate matching object according to the information to be matched and the candidate matching information;

determining a target matching object matched with the object to be matched from the candidate matching objects according to the matching degree of the candidate matching objects;

the information to be matched comprises skill information to be received and/or service timeliness requirement information, and the candidate matching information comprises available skill information and/or service timeliness providing information.

In a second aspect, an embodiment of the present invention further provides an object matching apparatus, which may include:

the matching information acquisition module is used for acquiring the information to be matched of the object to be matched corresponding to the object matching event and the candidate matching information of each candidate matching object when the object matching event is detected;

the matching degree determining module is used for determining the matching degree between the object to be matched and the candidate matching object according to the information to be matched and the candidate matching information aiming at the candidate matching information of each candidate matching object;

the object matching module is used for determining a target matching object matched with the object to be matched from all the candidate matching objects according to the matching degree of all the candidate matching objects;

In a third aspect, an embodiment of the present invention further provides an object matching apparatus, which may include:

one or more processors;

a memory for storing one or more programs;

when the one or more programs are executed by the one or more processors, the one or more processors implement the object matching method provided by any embodiment of the present invention.

In a fourth aspect, the embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the object matching method provided in any embodiment of the present invention.

According to the technical scheme of the embodiment of the invention, the information to be matched of the object to be matched and the candidate matching information of each candidate matching object corresponding to the detected object matching event are obtained, wherein the information to be matched can comprise the information required to receive skill and/or the information required by service timeliness, and the candidate matching information can comprise the information provided with skill and/or the information provided by service timeliness; furthermore, for the candidate matching information of each candidate matching object, the matching degree between the object to be matched and the candidate matching object can be determined according to the information to be matched and the candidate matching information, which can indicate the matching performance of the object to be matched and the candidate matching object in the aspects of skills and/or service timeliness; accordingly, according to the matching degree of each candidate matching object, a target matching object matched with the object to be matched can be determined from the candidate matching objects, and the target matching object can be a candidate matching object which is more matched with the object to be matched in the aspects of skill and/or service timeliness. According to the technical scheme, the matching performance of the object to be matched and the candidate matching object in the aspects of skills and/or service timeliness is fully considered, so that the target matching object matched for the object to be matched and the object to be matched can have higher matching degree in the aspects of skills and/or service timeliness, when the target matching object and the object to be matched are applied to an online inquiry scene, the referral rate, the return rate and the like can be reduced to a greater extent, and the user experience is improved.

Drawings

FIG. 1 is a flowchart of an object matching method according to a first embodiment of the present invention;

FIG. 2 is a flowchart of an object matching method according to a second embodiment of the present invention;

FIG. 3 is a flowchart of an object matching method according to a third embodiment of the present invention;

fig. 4 is a schematic diagram of a first optional example in an object matching method in a third embodiment of the present invention;

fig. 5 is a schematic diagram of a second alternative example in an object matching method in the third embodiment of the present invention;

fig. 6 is a block diagram of an object matching apparatus according to a fourth embodiment of the present invention;

fig. 7 is a schematic structural diagram of an object matching apparatus in the fifth embodiment of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.

Before the embodiment of the present invention is described, an application scenario of the embodiment of the present invention is exemplarily described: in the online inquiry scene, since the triage assignment module only considers the comprehensive scores of doctors when matching the doctors, and ignores the matching degree between the user and the doctors, the doctors matching a certain user may be doctors who cannot treat the disease of the user, for example, the disease of a certain user belongs to the internal medicine, and the matched doctors also belong to the internal medicine, but there are many sub-department rooms below the internal medicine, such as the stomach department, the hepatobiliary department, and the like, and the diseases that can be treated by the doctors in each sub-department room are different, which easily results in a high referral rate (referral by the doctors); for another example, when a user needs a quick doctor visit, the matched doctor for the user may not be able to provide a quick doctor visit service immediately, which easily results in a high rate of return (from the user).

On the basis, in order to solve the technical problem, the object matching method set forth in the following embodiments of the present invention fully considers the matching degree between the user and the doctor, especially the matching degree in terms of skills and service timeliness, so that the user can be matched to meet the actual needs of the user. The specific implementation process of the object matching method is as follows.

Example one

Fig. 1 is a flowchart of an object matching method according to an embodiment of the present invention. The embodiment can be applied to the condition that the target object with higher matching degree in the aspects of skills and service timeliness is matched with the object to be matched. The method can be executed by the object matching device provided by the embodiment of the invention, the device can be realized by software and/or hardware, the device can be integrated on the object matching equipment, and the equipment can be various user terminals or servers.

Referring to fig. 1, the method of the embodiment of the present invention specifically includes the following steps:

s110, when an object matching event is detected, acquiring information to be matched of an object to be matched corresponding to the object matching event and candidate matching information of each candidate matching object, wherein the information to be matched comprises skill information to be received and/or service timeliness requirement information, and the candidate matching information comprises available skill information and/or service timeliness providing information.

The object matching event may be an event for matching a corresponding object to be matched with a target object with a higher matching degree in terms of skills and service timeliness, in different application scenarios, a trigger mechanism of the object matching event may have differences, for example, in an online inquiry scenario, when disease description information from a certain user is received, the object matching event may be triggered, and the object to be matched at this time may be the user; in a legal consultation scene, when legal question description information from a certain user is received, the object matching event can be triggered, and the object to be matched at the moment can also be the user; etc., and are not specifically limited herein.

Therefore, when the object matching event is detected, the information to be matched of the object to be matched corresponding to the object matching event can be acquired, and the information to be matched can show the skill information to be received and/or the service timeliness requirement information of the object to be matched. Specifically, the skill information to be accepted may be information related to the skill that the subject to be matched needs to accept, and the skill may have different meanings in different application scenarios, for example, in an online inquiry scenario, the skill may be a technical ability capable of diagnosing the condition of the subject to be matched that needs to be inquired; in a legal consultation scene, the skill can be the technical ability to answer the legal questions that the object to be matched needs to consult; and so on. The service aging requirement information may be information related to a requirement of the object to be matched on service aging, such as immediately completing a certain service, completing a certain service within a preset time period, and the like, and the service may have different meanings in different application scenarios, for example, a service for receiving a consultation in an online consultation scenario, a service for answering a consultation in a legal consultation scenario, and the like.

Meanwhile, candidate matching information of each candidate matching object corresponding to the object matching event can be obtained, wherein the candidate matching object can be an object which is determined from each matchable object and has a certain matching degree with the corresponding object to be matched, exemplarily, in an online inquiry scene, the matchable objects can be all doctors on an online inquiry medical system, and the candidate matching objects can be doctors belonging to departments related to the illness state of the object to be matched in each doctor; in the legal consultation scene, the matchable objects can be all lawyers in a certain law, and the candidate matching objects can be lawyers under legal branches related to legal problems consulted by the object to be matched in each lawyer; and so on. On the basis, for each candidate matching object, the candidate matching information of the candidate matching object can show the available skill information and/or the service aging providing information of the candidate matching object. Specifically, the providable skill information may be information related to the skills that can be provided by the candidate matching object, and the skills may have different meanings in different application scenarios, for example, in an online inquiry scenario, the skills may be technical abilities capable of diagnosing the disease condition that needs to be inquired by the corresponding object to be matched; in a legal consultation scene, the skill can be the technical ability to answer the legal questions to be consulted by the corresponding object to be matched; and so on. The service timeliness providing information may be information related to what service timeliness the candidate matching object can provide, such as being capable of completing a certain service immediately, being capable of completing a certain service within a preset time period, and the like, and the service may have different meanings in different application scenarios, such as being a service for receiving a consultation in an online consultation scenario, being a service for answering a consultation in a legal consultation scenario, and the like.

In addition, in practical applications, optionally, the information to be matched may further include historical interaction information between the object to be matched and the candidate matching object, object preference information of the object to be matched in the current time period, historical information (such as historical behavior information and historical attribute information) of the first object in the historical time period, and the like; the candidate match information may also include service completion information for the candidate matching object over the current time period, second object historical information over a historical time period (e.g., historical behavior information, historical preference information), and so on. The information to be matched and the candidate matching information may be understood as information capable of indicating the degree of matching between the object to be matched and the candidate matching object.

And S120, aiming at the candidate matching information of each candidate matching object, determining the matching degree between the object to be matched and the candidate matching object according to the information to be matched and the candidate matching information.

Since a target matching object with a high matching degree with the object to be matched needs to be determined from the candidate matching objects, the matching degree between the object to be matched and each candidate matching object can be determined respectively. Specifically, for each candidate matching object, the matching degree between the object to be matched and the candidate matching object may be determined according to the information to be matched of the object to be matched and the candidate matching information of the candidate matching object. In practical application, optionally, the matching degree may be determined in multiple ways, for example, feature representation is performed on the information to be matched and the candidate matching information respectively to obtain the feature to be matched and the candidate matching feature, and then the matching degree is determined according to the similarity between the feature to be matched and the candidate matching feature; inputting the information to be matched and the candidate matching information into the trained object matching model, and obtaining the matching degree according to the output result of the object matching model; etc., and are not specifically limited herein. It should be noted that, since the information to be matched may include information of skill to be accepted and/or information of service timeliness requirement, and the candidate matching information may include information of available skill and/or information of service timeliness provision, the matching degree obtained according to the two may be understood as the matching of the object to be matched and the candidate matching object in terms of skill and/or timeliness service.

And S130, determining a target matching object matched with the object to be matched from the candidate matching objects according to the matching degree of the candidate matching objects.

After the matching degrees of the object to be matched and each candidate matching object are obtained, a target matching object more matched with the object to be matched may be determined from the candidate matching objects according to the matching degrees, for example, the candidate matching object with the highest matching degree is used as the target matching object, then, for example, the matching degrees are sorted in a sequence from high to low, the candidate matching objects corresponding to the preset number of matching degrees sorted in the front are used as the target matching objects, and the like, which is not specifically limited herein. It should be noted that, since the matching degree may indicate the matching between the object to be matched and the corresponding candidate matching object in terms of skill and/or time-based service, the target matching object obtained thereby may be a candidate matching object that is more matched with the object to be matched in terms of skill and/or time-based service.

Example two

Fig. 2 is a flowchart of an object matching method provided in the second embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, the object to be matched includes an object to be interrogated, and the candidate matching object includes a candidate subject to be examined. The same or corresponding terms as those in the above embodiments are not explained in detail herein.

Referring to fig. 2, the method of the present embodiment may specifically include the following steps:

s210, when an object matching event is detected, acquiring information to be matched of an object to be inquired corresponding to the object matching event and candidate matching information of each candidate object to be examined, wherein the information to be matched comprises skill information to be received and/or service timeliness requirement information, and the candidate matching information comprises available skill information and/or service timeliness providing information.

The object to be investigated can be an object to be subjected to investigation, and the candidate subject to be subjected to investigation can be a candidate subject capable of being subjected to investigation. On the basis, optionally, the skill information to be received can be information related to the medical skill which the subject to be examined needs to receive and can treat the disease of the subject to be examined, and the provided skill information can be information related to the medical skill which the candidate subject to be examined can provide and can treat which disease; the service timeliness requirement information can be information related to the requirement of the subject to be diagnosed in terms of treatment timeliness, and the service timeliness providing information can be information related to treatment timeliness that the candidate treatment subject can provide.

And S220, aiming at the candidate matching information of each candidate diagnosis receiving object, determining the matching degree between the object to be inquired and the candidate diagnosis receiving object according to the information to be matched and the candidate matching information.

And S230, determining a target matching object matched with the object to be inquired from the candidate examination objects according to the matching degree of the candidate examination objects.

On the basis, optionally, an inquiry list of the subject to be inquired can be assigned to the target subject, so that the target subject can perform the inquiry on the subject to be inquired after receiving the inquiry list.

According to the technical scheme of the embodiment of the invention, when the appropriate target treatment object is matched for the object to be treated, the matching degree of the object to be treated and each candidate treatment object in terms of medical skill and treatment time is fully considered, so that the referral rate, the return rate and the like are reduced, and the treatment experience of the object to be treated is improved.

On this basis, optionally, when the information to be matched includes skill information to be received and the candidate matching information includes available skill information, acquiring the information to be matched of the object to be matched corresponding to the object matching event and the candidate matching information of each candidate matching object may include: acquiring disease description information input by a subject to be queried corresponding to the subject matching event, and determining skill information required to be received of the subject to be queried according to the disease description information; and aiming at each candidate receiving object corresponding to the object matching event, acquiring medical skill information of the candidate receiving object, and determining the provided skill information of the candidate receiving object according to the medical skill information, wherein the medical skill information comprises at least one of information of a department to which the candidate receiving object belongs, information of illness excellence and information of authority of issuing. When the technical skill is considered, the technical information to be accepted can be determined according to the disease description information input by the subject to be asked, and the disease description information can be information which is directly or indirectly input by the subject to be asked and is used for describing the current disease state of the subject to be asked; the providable skill information may be determined according to medical skill information of the candidate subject, where the medical skill information may indicate which skills the candidate subject has in medical treatment, and may include at least one of information about a department to which the candidate subject belongs, information about disease excellence, and information about issuing authority, and may also include other information related to medical skills, which is not specifically limited herein. The department information can indicate which department the candidate subject belongs to, the disease adequacy information can indicate which disease the candidate subject is adequacy in diagnosing, and the opening authority information can indicate whether the candidate subject has opening authority, which prescription has opening authority, and the like. The technical scheme provides the basis for receiving the skill information and providing the determination basis of the skill information in the on-line inquiry scene, thereby improving the accuracy of the determination of the skill information and the skill information.

Optionally, when the information to be matched includes service timeliness requirement information and the candidate matching information includes service timeliness providing information, acquiring the information to be matched of the object to be matched corresponding to the object matching event and the candidate matching information of each candidate matching object, which may include: judging whether the object to be inquired corresponding to the object matching event triggers an extremely fast inquiry function or not, and determining service timeliness requirement information of the object to be inquired according to a judgment result; and acquiring the treatment timeliness information of the candidate treatment object aiming at each candidate treatment object corresponding to the object matching event, and determining the service timeliness providing information of the candidate treatment object according to the treatment timeliness information. When considering from the aspect of service timeliness, the service timeliness requirement information can be determined according to whether the object to be interrogated triggers the extremely-fast interrogation function, because the object to be interrogated which triggers the extremely-fast interrogation function indicates that the object to be interrogated hopes to obtain the service to be interrogated as soon as possible, that is, the object to be interrogated which has a higher service timeliness is preferred as a candidate for being interrogated, and the service timeliness requirement information at this time can be information related to obtaining the service to be interrogated as soon as possible; on this basis, the service aging providing information may be determined according to the treatment aging information of the candidate treatment subject, because the treatment aging information may be information about how the candidate treatment subject can complete the treatment service with the same aging, which may include, for example, current treatment aging information of the candidate treatment subject in a current treatment time period and/or historical treatment aging information in a historical treatment time period, where the current treatment aging information may include at least one of subject activity, treatment switch on identifier, current treatment singular number and subject fatigue, and the historical treatment aging information may include at least one of subject average treatment duration, subject average first-return duration and subject average communication turn. Wherein the current visit time period may be a time period when the subject matching event is detected, such as the week, day, hour, etc. when it is detected; the historical visit period may be a period of time before the subject matching event is detected, such as a month, a week, a day, etc. before it is detected. The current treatment time efficiency information may be treatment time efficiency information in the current time period, wherein the object activity may indicate whether the candidate treatment object is actively receiving treatment currently, and the candidate treatment object in the active treatment state may provide treatment service quickly; the receiving switch opening mark can represent whether the candidate receiving object opens the receiving switch or not, which can represent whether the candidate receiving object is willing to receive the treatment or not, and the probability that the original candidate receiving object which receives the treatment can rapidly provide the receiving service is higher; the current treatment singular number can represent the number of the candidate treatment subjects in the current treatment time period, and the candidate treatment subjects with less current treatment singular number can rapidly provide treatment service with higher possibility; the fatigue of the subject may represent the fatigue of the candidate subject in the current time period, which may be determined by the current number of visits, for example, taking a sigmod function as an example, the independent variable is the current number of visits, the dependent variable is the fatigue of the subject, and when the current number of visits exceeds a preset threshold, the fatigue of the subject increases extremely rapidly, which may result in that the candidate subject cannot provide the visit service quickly. The historical treatment time-efficiency information can be treatment time-efficiency information in a historical time period, wherein the average treatment time length of the object can represent the average value of each treatment time length of the candidate treatment object in the historical treatment time period, and the average value can represent the treatment service completion efficiency of the candidate treatment object; the average first-time duration of the subject may represent an average duration from the beginning of the examination to the first-time response of the candidate examination subject in each examination within the historical examination time period, which may represent the first-time response efficiency of the candidate examination subject; the average communication round of the subject may represent an average value of communication rounds that the candidate subject needs to perform to complete the service of the examination in each examination within the historical time period of the examination, which may present the efficiency of completing the service of the candidate subject. The historical diagnosis receiving time-effect information can show whether the candidate diagnosis receiving object can rapidly complete the diagnosis receiving service, namely, the technical scheme provides the basis for determining the service time-effect demand information and the service time-effect providing information in the on-line inquiry scene, so that the accuracy of determining the service time-effect demand information and the service time-effect providing information is improved.

Optionally, the obtaining of the information to be matched of the object to be matched corresponding to the object matching event may include: acquiring object preference information of an object to be queried corresponding to the object matching event in the current query time period and/or first object history information in the historical query time period, and taking the object preference information and/or the first object history information as information to be matched. The current inquiry time period may be a time period when the object matching event is detected, such as the week, the day, the hour, and the like when the object matching event is detected, which may be the same as or different from the current visit time period, and is not specifically limited herein; the historical inquiry time period may be a time period before the object matching event is detected, such as a month, a week, a day, etc., before the object matching event is detected, and may be the same as or different from the historical visit time period, which is not specifically limited herein. The object preference information may be information related to the preference of the object to be asked in the current asking time period, and may include at least one of the number of browsing times of each candidate department, the number of clicks of each candidate department, the last-time clicked department, a traditional chinese medical institution access identifier, and a search doctor function trigger identifier, where the number of browsing times is equivalent to the number of exposure times of a certain candidate department, or is the number of loading times of a link corresponding to the candidate department, and in practical applications, when there is a personalized push service or there is a difference when a home page is returned from each candidate department, the number of browsing times of each candidate department is no longer the same; the click times can represent the times of clicking a candidate department by the object to be inquired, and the higher the click times is, the higher the interest degree of the object to be inquired in the candidate department is; the department clicked last time can show the candidate department clicked last by the object to be asked, namely the candidate department interested in the object to be asked in real time; the traditional Chinese medicine hospital access identifier can indicate whether the subject to be inquired wishes to access the traditional Chinese medicine hospital, namely whether the subject to be inquired is interested in the traditional Chinese medicine hospital; the function trigger of the searching physician may indicate whether the object to be queried triggers the function of the searching physician, which may present whether the object to be queried wishes to receive a diagnosis from a candidate subject with a higher matching degree in terms of skills, and at this time, the object to be queried may be matched with a candidate subject more similar to the target searching physician searched by the object to be queried. The first object history information may be information related to behaviors and/or attributes of the object to be interrogated in a history interrogation period, and may include at least one of an interrogation frequency for each candidate department, a last interrogation department, a new user identifier at the last interrogation, a member user identifier, and a chronic disease identifier, where the interrogation frequency for each candidate department and the last interrogation department may be history behavior information of the object to be interrogated, the new user identifier at the last interrogation, the member user identifier, and the chronic disease identifier may be history attribute information of the object to be interrogated, where the new user identifier and the member user identifier at the last interrogation are mainly set from the perspective of retaining a new user and providing better consultation service for a good user, and the chronic disease identifier may indicate whether the disease of the object to be interrogated is a chronic disease or not, if so, it indicates that multiple interrogations are required, it may be possible to match candidate treatment subjects that were previously scored well by the subject to be interrogated. According to the technical scheme, the information to be matched is considered from different angles, and the accuracy of subsequent matching degree determination is improved due to the multi-angle consideration of the information to be matched.

Optionally, the obtaining of candidate matching information of each candidate matching object corresponding to the object matching event may include: and aiming at each candidate diagnosis receiving object corresponding to the object matching event, acquiring diagnosis receiving completion information of the candidate diagnosis receiving object in the current diagnosis receiving time period and/or second object historical information in the historical diagnosis receiving time period, and taking the diagnosis receiving completion information and/or the second object historical information as candidate matching information. Wherein the referral completion information may be information related to the completion of each referral service of the candidate referral subject in the current referral time period, and may include at least one of a subject referral completion rate, a subject refusal rate, a subject retirement rate, a subject referral rate and a subject goodness rate, wherein the subject referral completion rate may be a ratio of the number of referral services completed by the candidate referral subject to the number of total referral services, the subject refusal rate may be a ratio of the number of referral services rejected by the candidate referral subject to the number of total referral services, the subject retirement rate may be a ratio of the number of referral services retired by the candidate referral subject to the number of total referral services, the subject referral rate may be a ratio of the number of referral services transferred by the candidate referral subject to the number of total referral services, the subject goodness of care may be a ratio of the number of the candidate subjects who receive goodness of care of the subject to be asked to the number of the total number of the candidate subjects. The second object history information may be information related to behavior and/or attributes of the candidate receiving object in the history receiving time period, and may include at least one of object gender, object occupation level, object occupation type, object traditional Chinese and western medicine type, and history receiving singular number, wherein the object gender, the object occupation level, the object occupation type, and the object traditional Chinese and western medicine type may be history attribute information of the candidate receiving object, and the history receiving singular number may be history behavior information of the candidate receiving object. According to the technical scheme, the candidate matching information considered from different angles is provided, and the accuracy of subsequent matching degree determination is improved due to the multi-angle consideration of the candidate matching information.

EXAMPLE III

Fig. 3 is a flowchart of an object matching method provided in the third embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, the trained object matching model includes a first feature representation layer, a second feature representation layer, and a matching layer, and determining a matching degree between the object to be matched and the candidate matching object according to the information to be matched and the candidate matching information may include: inputting information to be matched into a first feature representation layer to obtain features to be matched, and inputting candidate matching information into a second feature representation layer to obtain candidate matching features; inputting the feature to be matched and the candidate matching feature into a matching layer to obtain the matching degree between the object to be matched and the candidate matching object, wherein the matching layer calculates the matching degree according to the splicing result of the feature to be matched and the candidate matching feature. The same or corresponding terms as those in the above embodiments are not explained in detail herein.

Referring to fig. 3, the method of this embodiment may specifically include the following steps:

s310, when an object matching event is detected, acquiring information to be matched of an object to be matched corresponding to the object matching event and candidate matching information of each candidate matching object, wherein the information to be matched comprises skill information to be received and/or service timeliness requirement information, and the candidate matching information comprises available skill information and/or service timeliness providing information.

S320, aiming at the candidate matching information of each candidate matching object, inputting the information to be matched into a first feature representation layer in the trained object matching model to obtain the feature to be matched, and inputting the candidate matching information into a second feature representation layer in the object matching model to obtain the candidate matching feature.

The first feature representation layer may be a network layer for representing the information to be matched as the feature to be matched, so that the feature to be matched may be obtained after the information to be matched is input to the first feature representation layer; the second feature representation layer may be a network layer for representing the candidate matching information as a candidate matching feature, and thus the candidate matching feature may be obtained by inputting the candidate matching information to the second feature representation layer. In practical applications, the first feature representation layer and the second feature representation layer may be understood as network layers capable of feature representation, and the first feature representation layer and the second feature representation layer may differ in network structure due to different input information, and are not specifically limited herein.

S330, inputting the feature to be matched and the candidate matching feature into a matching layer in the object matching model to obtain the matching degree between the object to be matched and the candidate matching object, wherein the matching layer calculates the matching degree according to the splicing result of the feature to be matched and the candidate matching feature.

The matching layer may be a network layer for determining the matching degree according to the splicing result of the feature to be matched and the candidate matching feature, that is, the matching layer may splice the feature to be matched and the candidate matching feature input to the matching layer, and then determine the matching degree based on the splicing result. It should be noted that, compared with the technical scheme that dimension conversion is performed on the feature to be matched and/or the candidate matching feature to make the feature dimensions of the feature to be matched and/or the candidate matching feature consistent, then the spatial value is calculated based on the feature to be matched and the candidate matching feature with consistent feature dimensions, and the matching degree is determined according to the spatial value, the original feature to be matched and the candidate matching feature are directly input into a network layer (i.e., a matching layer) without performing any dimension conversion, and all the features are directly processed in the matching layer, so that the accuracy of determining the matching degree is improved.

And S340, determining a target matching object matched with the object to be matched from the candidate matching objects according to the matching degree of the candidate matching objects.

According to the technical scheme of the embodiment of the invention, the matching information is subjected to characteristic representation through the characteristic representation layer, and the obtained matching characteristics are input into the matching layer for determining the matching degree according to the splicing result among the matching characteristics, so that the matching degree with high determination precision is obtained.

On this basis, optionally, when the information to be matched is skill information to be received, inputting the information to be matched into the first feature representation layer to obtain the feature to be matched, which may include: and performing feature representation on the skill information to be received based on a first multi-layer bidirectional conversion encoder in a first feature representation layer to obtain the feature to be matched of the skill information to be received. Because the skill information to be received belongs to the text information, in order to realize the feature representation of the text information, the feature representation can be performed on the text information based on the first multilayer bidirectional conversion encoder to obtain the corresponding feature to be matched. In practical applications, optionally, the first multi-layer Bidirectional conversion Encoder may be a Bidirectional Encoder Representation (BERT) model from converters (Transformers), and the application of the first multi-layer Bidirectional conversion Encoder improves the representation efficiency and representation accuracy when the skill information is required to be received for feature representation.

Optionally, when the candidate matching information is available to provide skill information, inputting the candidate matching information into the second feature representation layer to obtain a candidate matching feature, which may include: and performing feature representation on the available skill information based on a second multi-layer bidirectional conversion encoder in a second feature representation layer to obtain candidate matching features of the available skill information. The second multi-layer bi-directional transcoder is similar to the first multi-layer bi-directional transcoder described above, and is not described herein again. It should be noted that, because the input information of the two is different, the network structures of the two may be the same or different, and are not specifically limited herein.

Optionally, the information to be matched further includes historical interaction information between the object to be matched and the candidate matching object, and the information to be matched is input into the first feature representation layer to obtain the feature to be matched, which may include: and obtaining the to-be-matched features of the historical interactive information based on the key value relationship pre-stored in the first feature representation layer. The key-value relationship comprises key information (key) and value information (value), the key information comprises an object identifier to be matched of an object to be matched and a candidate matching object identifier of a candidate matching object, the object identifier to be matched can be a unique identifier of the object to be matched, the candidate matching object identifier can be a unique identifier of the candidate matching object, and the correspondence between the two identifiers can be stored in the key information, so that the key-value relationship corresponding to the candidate matching object and the object to be matched can be found from the multiple key-value relationship after the candidate matching object and the object to be matched are obtained. The value information may include each occurred interaction information between the object to be matched and the candidate matching object, and the occurred interaction information may be interaction information that has occurred between the object to be matched and the candidate matching object, and taking an online inquiry scene as an example, it may be that the object to be inquired has performed an inquiry to the candidate inquiry object, the object to be inquired has given a good comment to the candidate inquiry object that has been inquired, the object to be inquired has performed a back-diagnosis to the candidate inquiry object, the candidate inquiry object has performed a transfer and/or back-diagnosis to the object to be inquired, and so on. The information to be matched may further include historical interaction information between the object to be matched and the candidate matching object, where the historical interaction information may be interaction information that may or may not occur between the object to be matched and the candidate matching object within a historical matching time period, and the online inquiry scene is taken as an example continuously, whether the object to be inquired has performed an inquiry on the candidate inquiry object, whether the object to be inquired has already given a good comment on the inquired candidate inquiry object, whether the object to be inquired has performed a back-call on the candidate inquiry object, whether the candidate inquiry object has performed a transfer and/or back-call on the object to be inquired, and so on. Therefore, after the historical interaction information is input into the first feature representation layer, the first feature representation layer can find the key value relationship corresponding to the interaction parties (namely, the object to be matched and the candidate matching object) of the historical interaction information based on the key information in the pre-stored key value relationships, determine whether the historical interaction information occurs based on the value information in the key value relationship, and obtain the feature to be matched of the historical interaction information according to the determination result, for example, the feature to be matched of the sent historical interaction information is taken as 1, and otherwise, the feature to be matched of the historical interaction information is 0. According to the technical scheme, the possibility that the object to be matched and the history matching object are interacted in the history matching time period is fully considered, so that the characteristic to be matched of the history interaction information is determined through the prestored key value relation by taking the generated interaction information as a reference, and the effect of determining the characteristic to be matched in multiple angles is achieved.

Optionally, the object matching model may be obtained by pre-training through the following steps: acquiring to-be-historical information of an object to be historical, candidate historical information of a candidate historical object, and matching labels set for service completion information of the object to be historical according to the candidate historical object, and taking the to-be-historical information, the candidate historical information and the matching labels as a group of training samples; and training the original neural network model based on a plurality of groups of training samples to obtain an object matching model, wherein the original neural network model comprises a first feature representation layer, a second feature representation layer and a matching layer which are not trained, and a learning target layer for learning according to the output result and the matching label of the matching layer which are not trained. The matching label may be a label set according to the service completion information of the candidate history object for the object to be subjected to history, for example, if the service completion information indicates that the candidate history object completes the service of the object to be subjected to history and obtains a favorable comment of the object to be subjected to history on the service, the matching label may be 1; for example, if the service completion information indicates that the candidate history object does not serve the to-be-processed history object, the candidate history object does not serve the to-be-processed history object but does not complete the service, and the candidate history object completes the to-be-processed history object but does not receive a good comment of the to-be-processed history object on the service, the matching label may be 0; etc., and are not specifically limited herein. That is, the matching label is set based on whether the candidate history object has completed the service of this time well, and can reflect the matching degree between the candidate history object and the object to be historical. On the basis, the learning target layer can be a network layer for learning according to the output result of the untrained matching layer and the matching label, so that the trained object matching model can output a matching degree which is more similar to the matching label.

To better understand the training process of the object matching model as a whole, it is exemplarily described below with reference to specific examples. Illustratively, taking a Deep Structured Semantic Model (DSSM), i.e., a two-tower model, which is improved in conjunction with an online inquiry scenario as an example, see fig. 4, which includes a network structure at the patient end and a network structure at the doctor end, input information of the network structure at the patient end may include disease description information, real-time information of the patient, such as browsing information (i.e., the number of times the patient browses each candidate department), click information (i.e., the number of times the patient clicks each candidate department), doctor-patient matching information (i.e., historical interaction information between the patient and the doctor), etc., and offline information of the patient, such as the number of inquiries to each candidate department, last inquiry department, new user identification at last inquiry, member user identification, chronic disease identification, etc., it should be noted that the real-time information (i.e., online information) may be considered as information obtained in the current inquiry period, offline information can be considered as information obtained during a historical interrogation period, i.e., various information of a patient can be divided from an online and offline perspective; the input information of the network structure at the doctor end may include medical skill information, real-time information such as liveness (i.e., activity of the subject), fatigue (i.e., fatigue of the subject), and the like, and offline information such as job level (i.e., occupation level of the subject), goodness of rating (i.e., goodness of rating of the subject), average length of treatment (i.e., average length of treatment of the subject), and the like, wherein the medical skill information may also be attributed to the offline information. The network structure of the patient side and the network structure of the doctor side can be used as input layers, in addition, the double-tower model can further comprise a matching layer and a learning target layer, and the detailed description is as follows:

a first layer: input layer

The model training is divided into two different "towers" (i.e., neural networks) that can be used to generate patient-end features (embedding) and physician-end features (embedding). In particular, the method comprises the following steps of,

input information for a network architecture on the patient side

1, the patient end performs characteristic representation (32-dimensional) on the disease description information by using BRRT;

2, real-time information (44 dimension) including information of real-time browsing and clicking of the patient and doctor-patient matching information, wherein the real-time information can be acquired from online acquisition in advance and can be obtained by reversely calculating the operation time of the patient according to the time of the assignment order, and the specific information is as follows:

information such as real-time browsing and clicking (40-dimensional):

the browsing times of the patient to each candidate department (such as each primary main department) (16-dimensional in total, each department corresponds to one dimension), the click times of the patient to each primary main department (16-dimensional in total, each department corresponds to one dimension), the last click department of the patient (4-dimensional), whether the patient triggers the rapid inquiry function, the function trigger mark of the searching doctor and the visit mark of the traditional Chinese medical hospital.

Doctor-patient matching information (4-dimensional):

whether the patient's disease matches the physician's expertise, whether the patient has performed an inquiry to the physician, whether the patient has given a good comment on the physician who has been inquired, whether the patient has performed a back-visit to the physician, whether the physician has rejected the patient.

3 offline information (23 dimensions):

the number of times of the patient's inquiry to each candidate department within the last three months (16-dimensional, each department corresponds to one dimension), the last inquiry department (i.e. the last candidate department) (4-dimensional), the new user identifier at the last inquiry (i.e. whether the last inquiry is a new user), the member user identifier (i.e. whether the last inquiry is a PLUS member) and the chronic disease identifier (i.e. whether the last inquiry is a chronic disease patient).

The patient side can respectively perform feature representation on the three information parts, and the feature representation results are spliced to generate 99-dimensional patient side features.

Input information of network structure of doctor side

1 real-time information (7 dimensions):

the doctor activity, the examination receiving switch opening identification (namely whether the examination receiving switch is opened or not), the current examination receiving odd number, the doctor fatigue, the doctor daily refusal rate, the doctor daily goodness rate and the doctor daily referral rate.

2 offline information (32 dimensions):

department information (4-dimension), disease adequacy information (namely, doctor adequacy diseases), doctor gender (2-dimension), doctor occupation level (1-dimension), doctor occupation type (2-dimension), authority information for starting (namely, whether the doctor has authority for starting), Chinese and western medicine type (2-dimension) of the doctor, historical examination receiving odd number, doctor's good appraisal rate, doctor's average examination receiving duration, doctor's average first-return duration, doctor's average communication turn, doctor's average communication time, doctor's preference disease severity (1-dimension), doctor's rejection rate, doctor's referral rate and doctor's examination receiving completion rate.

The patient side can respectively perform feature representation on the two pieces of information, and the feature representation results are spliced to generate 39-dimensional patient side features.

A second layer: matching layer

The 99-dimensional features of the patient side and the 39-dimensional features of the doctor side are spliced (concat) and then go through a full connect (full connect) layer and 2 hidden (hidden) layers, wherein the first hidden layer comprises 64 neurons, the second hidden layer comprises 32 neurons, and each hidden layer uses a relu activation function.

And a third layer: learning target layer

The final learning target of the double-tower model is the matching degree between doctors and patients (namely matching label, 0 or 1), the learning target layer is connected with the output of the matching layer in 32 dimensions, the output of the learning target layer is in 2 dimensions, on the basis, the optional loss function is the cross entropy, the optimizer is adam, and the learning rate is 0.001.

In order to verify the effectiveness of the training step, the Area Under the ROC Curve (AUC) defined by the coordinate axes is used as a test index for testing the double-tower model trained by the training step, wherein the ROC Curve may be called a receiver operating characteristic Curve (receiver operating characteristic Curve). Specifically, the test index may include two parts: firstly, accuracy and recall rate; second, good rate and long time for receiving a diagnosis. The total number of data sets of the test is 20000, wherein the number of training sets is 15000, the number of test sets is 5000, the accuracy of the double-tower model after training on the test sets is 90.2%, and the recall rate is 83.6%; meanwhile, A/B Test is carried out after the model is on line (namely applied) to obtain the time length of the treatment and the favorable assessment rate, wherein the favorable assessment rate is increased from 94% to 96%, the referral rate is reduced from 8% to 2%, the rejection rate is reduced from 5% to 1.5%, and the return rate is reduced from 10% to 4%. Therefore, the obtained double-tower model can effectively solve the problems of high referral rate, matrix rate and return rate in the triage and order dispatching process, and improves the user experience.

It should be noted that, besides the above-mentioned two-tower model, other models capable of matching features of two ends may be used, such as wrde-and-deep model capable of matching shallow features and deep features of the doctor end and the patient end, which is not specifically limited herein.

In addition, in order to better understand the working process of the online inquiry medical system as a whole, the following can be taken as an illustrative description with reference to specific examples. Illustratively, as shown in fig. 5, it is a workflow diagram of an online inquiry medical system, and the object matching method corresponds to calculating the matching degree between doctors and patients based on a DSSM model, and then the important links in the workflow are briefly described. The patient describes a session (i.e., disease description information) from which the intelligent disease analysis module determines what disease the patient needs to consult and matches based on the disease and the physician's domain of expertise and department, which is prepared online. The predicted saturation degree is that whether the doctors can process and complete the orders or not is predicted according to the possible orders in the future time period and the current doctor number, if yes, the saturation is not achieved, and otherwise, the saturation is achieved. The comprehensive score is calculated according to information such as the favorable rating of doctors, the average doctor receiving time and the like, and the reason that a plurality of doctors exist in a department, and from the aspect of realizability, all the doctors are difficult to recall and input into the DSSM model for matching degree calculation is considered, so that some doctors can be screened out according to the comprehensive score and then can be found. The isolation rule may be whether the doctor has an illegal action such as refusal of a doctor, long time non-treatment, etc., which may affect the penalty factor.

Example four

Fig. 6 is a block diagram of an object matching apparatus according to a fourth embodiment of the present invention, where the apparatus is configured to execute an object matching method according to any of the embodiments. The object matching method of the present invention is not limited to the above embodiments, and the embodiments of the object matching method may be referred to for details that are not described in detail in the embodiments of the object matching device. Referring to fig. 6, the apparatus may specifically include: a matching information obtaining module 410, a matching degree determining module 420 and an object matching module 430. Wherein the content of the first and second substances,

a matching information obtaining module 410, configured to, when an object matching event is detected, obtain to-be-matched information of an object to be matched corresponding to the object matching event and candidate matching information of each candidate matching object;

a matching degree determining module 420, configured to determine, according to the information to be matched and the candidate matching information, a matching degree between the object to be matched and the candidate matching object for the candidate matching information of each candidate matching object;

the object matching module 430 is configured to determine, according to the matching degree of each candidate matching object, a target matching object that matches the object to be matched from each candidate matching object;

Optionally, the trained object matching model includes a first feature representation layer, a second feature representation layer, and a matching layer, and the matching degree determining module 420 may include:

the matching feature obtaining unit is used for inputting the information to be matched into the first feature representation layer to obtain the feature to be matched and inputting the candidate matching information into the second feature representation layer to obtain the candidate matching feature;

and the matching degree determining unit is used for inputting the features to be matched and the candidate matching features into the matching layer to obtain the matching degree between the object to be matched and the candidate matching object, wherein the matching layer calculates the matching degree according to the splicing result of the features to be matched and the candidate matching features.

On this basis, optionally, when the information to be matched is skill information to be received, the matching feature obtaining unit may include: the first to-be-matched feature obtaining subunit is used for performing feature representation on skill information to be received on the basis of a first multilayer bidirectional conversion encoder in a first feature representation layer to obtain to-be-matched features of the skill information to be received; and/or the presence of a gas in the gas,

the information to be matched may further include historical interaction information between the object to be matched and the candidate matching object, and the matching feature obtaining unit may include: the second to-be-matched feature obtaining subunit is used for obtaining to-be-matched features of the historical interaction information based on a key value relationship pre-stored in the first feature representation layer, wherein key information in the key value relationship comprises to-be-matched object identifiers of the to-be-matched objects and candidate matching object identifiers of the candidate matching objects, and value information in the key value relationship comprises interaction information which has occurred between the to-be-matched objects and the candidate matching objects; and/or the presence of a gas in the gas,

when the candidate matching information is the available skill information, the matching feature obtaining unit may include: and the candidate matching feature obtaining subunit is used for performing feature representation on the providable skill information based on a second multi-layer bidirectional conversion encoder in a second feature representation layer to obtain candidate matching features which can provide the skill information.

Optionally, the object matching model may be obtained by pre-training through the following modules:

a training sample obtaining module, configured to obtain to-be-historical information of an object to be historical, candidate historical information of a candidate historical object, and matching labels set for service completion information of the object to be historical according to the candidate historical object, and use the to-be-historical information, the candidate historical information, and the matching labels as a set of training samples;

and the object matching model obtaining module is used for training the original neural network model based on a plurality of groups of training samples to obtain an object matching model, wherein the original neural network model comprises a first feature representation layer, a second feature representation layer and a matching layer which are not trained, and a learning target layer which is used for learning according to the output result and the matching label of the matching layer which are not trained.

Optionally, the object to be matched includes an object to be interrogated, and the candidate matching object includes a candidate subject to be examined.

On this basis, optionally, when the information to be matched includes skill information to be accepted and the candidate matching information includes available skill information, the matching information obtaining module 410 may include:

the skill information to be accepted determining unit is used for acquiring disease description information input by the object to be inquired corresponding to the object matching event and determining skill information to be accepted of the object to be inquired according to the disease description information;

the system comprises a providable skill information determining unit, a providing authority information determining unit and a judging unit, wherein the providable skill information determining unit is used for acquiring medical skill information of candidate treatment subjects for each candidate treatment subject corresponding to a subject matching event, and determining providable skill information of the candidate treatment subjects according to the medical skill information, and the medical skill information comprises at least one of information of a department to which the subject belongs, information of illness excellence and information of authority of an issuing party; and/or the presence of a gas in the gas,

when the information to be matched includes service timeliness requirement information and the candidate matching information includes service timeliness providing information, the matching information obtaining module 410 may include:

the service timeliness requirement information determining unit is used for judging whether the object to be inquired corresponding to the object matching event triggers the extremely fast inquiry function or not and determining the service timeliness requirement information of the object to be inquired according to the judgment result;

the service timeliness providing information determining unit is used for acquiring the service timeliness information of the candidate receiving objects aiming at each candidate receiving object corresponding to the object matching event, and determining the service timeliness providing information of the candidate receiving objects according to the receiving timeliness information, wherein the receiving timeliness information comprises the current receiving timeliness information of the candidate receiving objects in the current receiving time period and/or the historical receiving timeliness information in the historical receiving time period, the current receiving timeliness information comprises at least one of object activity, receiving switch opening identification, current receiving singular number and object fatigue, and the historical receiving timeliness information comprises at least one of object average receiving duration, object average first-return duration and object average communication rounds.

Still optionally, the matching information obtaining module 410 may include:

the system comprises a to-be-matched information acquisition unit, a to-be-matched information acquisition unit and a matching unit, wherein the to-be-matched information acquisition unit is used for acquiring object preference information of an object to be inquired corresponding to an object matching event in a current inquiry time period and/or first object history information in a historical inquiry time period, the object preference information and/or the first object history information are used as to-be-matched information, the object preference information comprises at least one of the browsing times of each candidate department, the clicking times of each candidate department, the last clicking department, a traditional Chinese medicine hospital access identifier and a search doctor function trigger identifier, and the first object history information comprises at least one of the inquiry times of each candidate department, the last inquiry department, a new user identifier at the last inquiry, a member user identifier and a chronic disease identifier; and/or the presence of a gas in the gas,

and the candidate matching information acquisition unit is used for acquiring the receiving completion information of the candidate receiving object in the current receiving time period and/or the second object history information in the historical receiving time period aiming at each candidate receiving object corresponding to the object matching event, and taking the receiving completion information and/or the second object history information as the candidate matching information, wherein the receiving completion information comprises at least one of object receiving completion rate, object rejection rate, object return rate, object referral rate and object qualification rate, and the second object history information comprises at least one of object gender, object occupation level, object occupation type, object traditional Chinese and western type and historical receiving singular number.

In the object matching device provided by the fourth embodiment of the present invention, the matching information obtaining module obtains the information to be matched of the object to be matched corresponding to the detected object matching event and the candidate matching information of each candidate matching object, where the information to be matched may include information required to receive skills and/or information required to provide service timeliness, and the candidate matching information may include information capable of providing skills and/or information provided to provide service timeliness; further, the matching degree determining module may determine, for the candidate matching information of each candidate matching object, the matching degree between the object to be matched and the candidate matching object according to the information to be matched and the candidate matching information, which may indicate the matching property of the object to be matched and the candidate matching object in the aspect of skill and/or service timeliness; accordingly, the object matching module may determine, according to the matching degree of each candidate matching object, a target matching object that matches the object to be matched from among the candidate matching objects, where the target matching object may be a candidate matching object that is more matched with the object to be matched in terms of skills and/or service timeliness. The device sufficiently considers the matching performance of the object to be matched and the candidate matching object in the aspects of skills and/or service timeliness, so that the target matching object matched for the object to be matched and the object to be matched can have higher matching degree in the aspects of skills and/or service timeliness, when the device is applied to an online inquiry scene, the referral rate, the return rate and the like can be reduced to a greater extent, and the user experience is improved.

The object matching device provided by the embodiment of the invention can execute the object matching method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method.

It should be noted that, in the embodiment of the object matching apparatus, each included unit and module are merely divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be implemented; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.

EXAMPLE five

Fig. 7 is a schematic structural diagram of an object matching apparatus according to a fifth embodiment of the present invention, and referring to fig. 7, the apparatus includes a memory 510, a processor 520, an input device 530, and an output device 540. The number of processors 520 in the device may be one or more, and one processor 520 is taken as an example in fig. 7; the memory 510, processor 520, input device 530, and output device 540 in the apparatus may be connected by a bus or other means, such as by bus 550 in fig. 7.

The memory 510 is used as a computer-readable storage medium for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the object matching method in the embodiment of the present invention (for example, the matching information acquisition module 410, the matching degree determination module 420, and the object matching module 430 in the object matching apparatus). The processor 520 executes various functional applications of the device and data processing by executing software programs, instructions, and modules stored in the memory 510, that is, implements the object matching method described above.

The memory 510 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of the device, and the like. Further, the memory 510 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, memory 510 may further include memory located remotely from processor 520, which may be connected to devices through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The input device 530 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the device. The output device 540 may include a display device such as a display screen.

EXAMPLE six

An embodiment of the present invention provides a storage medium containing computer-executable instructions, which when executed by a computer processor, are configured to perform a method for object matching, the method including:

Of course, the storage medium provided by the embodiment of the present invention contains computer-executable instructions, and the computer-executable instructions are not limited to the operations of the method described above, and may also perform related operations in the object matching method provided by any embodiment of the present invention.

From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly, can also be implemented by hardware, but the former is a better embodiment in many cases. With this understanding, the technical solutions of the present invention may be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.

It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. An object matching method, comprising:

determining a target matching object matched with the object to be matched from each candidate matching object according to the matching degree of each candidate matching object;

2. The method of claim 1, wherein the trained object matching model comprises a first feature representation layer, a second feature representation layer and a matching layer, and the determining the matching degree between the object to be matched and the candidate matching object according to the information to be matched and the candidate matching information comprises:

inputting the information to be matched into the first feature representation layer to obtain a feature to be matched, and inputting the candidate matching information into the second feature representation layer to obtain a candidate matching feature;

inputting the feature to be matched and the candidate matching feature into the matching layer to obtain the matching degree between the object to be matched and the candidate matching object, wherein the matching layer calculates the matching degree according to the splicing result of the feature to be matched and the candidate matching feature.

3. The method according to claim 2, wherein when the information to be matched is the skill information to be accepted, the inputting the information to be matched into the first feature representation layer to obtain a feature to be matched comprises: performing feature representation on the skill information to be received based on a first multi-layer bidirectional conversion encoder in the first feature representation layer to obtain the feature to be matched of the skill information to be received;

and/or the presence of a gas in the gas,

the information to be matched further includes historical interaction information between the object to be matched and the candidate matching object, and the step of inputting the information to be matched into the first feature representation layer to obtain the feature to be matched includes: obtaining the to-be-matched features of the historical interaction information based on a key value relationship pre-stored in the first feature representation layer, wherein the key information in the key value relationship comprises an object identifier to be matched of the to-be-matched object and a candidate matching object identifier of the candidate matching object, and the value information in the key value relationship comprises interaction information which has occurred between the to-be-matched object and the candidate matching object;

and/or the presence of a gas in the gas,

when the candidate matching information is the providable skill information, the inputting the candidate matching information into the second feature representation layer to obtain a candidate matching feature includes: and performing feature representation on the provided skill information based on a second multi-layer bidirectional conversion encoder in the second feature representation layer to obtain candidate matching features of the provided skill information.

4. The method of claim 2, wherein the object matching model is pre-trained by:

acquiring to-be-historical information of an object to be historical, candidate historical information of a candidate historical object and a matching label set for service completion information of the object to be historical according to the candidate historical object, and taking the to-be-historical information, the candidate historical information and the matching label as a group of training samples;

training an original neural network model based on a plurality of groups of training samples to obtain the object matching model, wherein the original neural network model comprises the first feature representation layer, the second feature representation layer and the matching layer which are not trained, and a learning target layer for learning according to the output result of the matching layer which is not trained and the matching labels.

5. The method of claim 1, wherein the object to be matched comprises an object to be interrogated and the candidate matched object comprises a candidate subject to be examined.

6. The method according to claim 5, wherein when the information to be matched includes the skill information to be accepted and the candidate matching information includes the providable skill information, the obtaining of the information to be matched of the object to be matched corresponding to the object matching event and the candidate matching information of each candidate matching object includes:

acquiring disease description information input by the object to be inquired corresponding to the object matching event, and determining the skill information to be received of the object to be inquired according to the disease description information;

acquiring medical skill information of the candidate diagnosis receiving object aiming at each candidate diagnosis receiving object corresponding to the object matching event, and determining the provided skill information of the candidate diagnosis receiving object according to the medical skill information, wherein the medical skill information comprises at least one of information of a department to which the object belongs, information of illness excellence and information of authority of an issuer;

and/or the presence of a gas in the gas,

when the information to be matched includes the service timeliness requirement information and the candidate matching information includes the service timeliness providing information, the obtaining of the information to be matched of the object to be matched corresponding to the object matching event and the candidate matching information of each candidate matching object includes:

judging whether the object to be interrogated corresponding to the object matching event triggers an extremely fast interrogation function or not, and determining the service timeliness requirement information of the object to be interrogated according to the judgment result;

and for each candidate receiving object corresponding to the object matching event, acquiring receiving aging information of the candidate receiving object, and determining the service aging providing information of the candidate receiving object according to the receiving aging information, wherein the receiving aging information comprises current receiving aging information of the candidate receiving object in a current receiving time period and/or historical receiving aging information in a historical receiving time period, the current receiving aging information comprises at least one of object activity, a receiving switch opening identifier, current receiving singular number and object fatigue, and the historical receiving aging information comprises at least one of object average receiving duration, object average first-return duration and object average communication rounds.

7. The method according to claim 5, wherein the obtaining of the information to be matched of the object to be matched corresponding to the object matching event comprises:

acquiring object preference information of the object to be inquired corresponding to the object matching event in the current inquiry time period and/or first object history information in the historical inquiry time period, and taking the object preference information and/or the first object history information as the information to be matched, wherein the object preference information comprises at least one of browsing times of each candidate department, clicking times of each candidate department, last clicking department, traditional Chinese medicine hospital access identification and search doctor function trigger identification, and the first object history information comprises at least one of inquiry times of each candidate department, last inquiry department, new user identification at last inquiry, member user identification and chronic disease identification;

and/or the presence of a gas in the gas,

the obtaining of candidate matching information of each candidate matching object corresponding to the object matching event includes:

and acquiring the receiving completion information of the candidate receiving objects in the current receiving time period and/or the second object historical information in the historical receiving time period aiming at each candidate receiving object corresponding to the object matching event, and taking the receiving completion information and/or the second object historical information as the candidate matching information, wherein the receiving completion information comprises at least one of object receiving completion rate, object rejection rate, object return rate, object referral rate and object evaluation rate, and the second object historical information comprises at least one of object gender, object occupation level, object occupation type, object traditional Chinese and western medicine type and historical receiving singular number.

8. An object matching apparatus, comprising:

a matching degree determining module, configured to determine, according to the information to be matched and the candidate matching information, a matching degree between the object to be matched and the candidate matching object for the candidate matching information of each candidate matching object;

the object matching module is used for determining a target matching object matched with the object to be matched from each candidate matching object according to the matching degree of each candidate matching object;

9. An object matching apparatus, characterized by comprising:

one or more processors;

a memory for storing one or more programs;

when executed by the one or more processors, cause the one or more processors to implement the object matching method of any of claims 1-7.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the object matching method according to any one of claims 1 to 7.