An answerer matching method and apparatus, an electronic device, and a storage medium

Abstract

The present disclosure provides a method and device for matching an answerer, electronic equipment, readable storage medium and computer program product, and relates to the field of artificial intelligence and intelligent search. The specific implementation scheme is: determining an answerer vector corresponding to a candidate answerer based on a first keyword vector corresponding to a first keyword, wherein the first keyword is determined according to a question answered by the candidate answerer, and the answerer vector is a vector used to represent a question that can be answered by the candidate answerer; determining a target answerer matched with a question to be answered from the candidate answerer based on a second keyword vector corresponding to a second keyword and the answerer vector, wherein the second keyword is determined according to the question to be answered. The answerer vector can more effectively represent the question that can be answered by the answerer, and can express various questions to be answered, including new questions, in the form of a vector, so that the target answerer matched with the question to be answered can be more accurate.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence, specifically to intelligent search and big data technologies, which can be applied to scenarios such as intelligent search. Background Technology

[0002] For interactive knowledge-sharing platforms, the ability to accurately match users with answerers for their questions is a crucial factor in ensuring a good user experience. Summary of the Invention

[0003] This disclosure provides a method, apparatus, electronic device, readable storage medium, and computer program product for accurately matching respondents to questions to be answered.

[0004] According to one aspect of this disclosure, a method for matching respondents is provided, which may include the following steps:

[0005] Based on the first word vector corresponding to the first keyword, the answerer vector corresponding to the candidate answerer is determined. The first keyword is the keyword determined based on the candidate answerer's answered questions, and the answerer vector is the vector used to represent the questions that the candidate answerer can answer.

[0006] Based on the second word vector corresponding to the second keyword and the respondent vector, the target respondent that matches the question to be answered is determined from the candidate respondents. The second keyword is the keyword determined according to the question to be answered.

[0007] According to a second aspect of this disclosure, a matching device for respondents is provided, the device comprising:

[0008] The answerer vector determination unit is used to determine the answerer vector corresponding to the candidate answerer based on the first word vector corresponding to the first keyword. The first keyword is a keyword determined based on the candidate answerer's answered questions, and the answerer vector is a vector used to represent the questions that the candidate answerer can answer.

[0009] The target answerer matching unit is used to determine the target answerer that matches the question to be answered from the candidate answerers based on the second word vector corresponding to the second keyword and the answerer vector. The second keyword is the keyword determined according to the question to be answered.

[0010] According to another aspect of this disclosure, an electronic device is provided, comprising:

[0011] At least one processor; and

[0012] The memory is communicatively connected to the at least one processor; wherein,

[0013] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the methods in any embodiment of this disclosure.

[0014] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing a computer to perform the methods of any embodiment of this disclosure.

[0015] According to another aspect of this disclosure, a computer program product is provided, including a computer program / instructions, characterized in that the computer program / instructions, when executed by a processor, implement the methods in any embodiment of this disclosure.

[0016] The disclosed technique determines the answer vector representing the answerer's ability to answer questions by using the word vectors corresponding to the keywords of already answered questions. Therefore, it significantly reduces the problem of answer vectors from all answerers tending to be averaged, thus preventing the answer vectors from providing a personalized representation of the answerer. In this way, the answer vector can more effectively represent the answerer's ability to answer questions.

[0017] Representing the question to be answered by using word vectors corresponding to the keywords of the question allows for vector representation of various types of questions, including new questions. Therefore, by using the second word vector and the respondent vector, the target respondent that matches the question can be more accurately identified from the candidate respondents.

[0018] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0019] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0020] Figure 1 A flowchart illustrating a matching method for respondents provided in an embodiment of this disclosure;

[0021] Figure 2 This is a flowchart of a first method for determining the answer principal vector provided in the embodiments of this disclosure;

[0022] Figure 3 This is a flowchart of a second method for determining the answer principal vector provided in the embodiments of this disclosure;

[0023] Figure 4 This is a flowchart of a third method for determining the answer principal vector provided in the embodiments of this disclosure;

[0024] Figure 5 A schematic diagram of a matching device for respondents is provided for embodiments of this disclosure;

[0025] Figure 6 This is a schematic diagram of an electronic device provided for an embodiment of the present disclosure. Detailed Implementation

[0026] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0027] In related technologies, interactive knowledge Q&A sharing platforms typically use the following methods to match answerers to questions during the question-and-answer matching process:

[0028] First, the answered questions of candidate respondents are grouped into a question sequence, and then the Word2vec model is used to learn the word vector corresponding to each answered question. Second, the word vectors corresponding to multiple answered questions are summed and averaged to obtain the respondent vector used to represent the respondent. Third, when determining the question to be answered, the Word2vec model is used again to learn the word vector corresponding to the question to be answered. Finally, based on the respondent vector and the word vector corresponding to the question to be answered, the target respondent vector is matched.

[0029] However, in technologies related to answerer matching for questions awaiting answers, when an answerer has answered many questions, averaging the word vectors corresponding to multiple answered questions to obtain the answerer vector often leads to all answerer vectors becoming very similar. This results in the answerer vector failing to provide a personalized representation of the answerer, and consequently, failing to accurately match answerers for questions awaiting answers.

[0030] If the Item2ve model is used instead of the Word2vec model when learning word vectors for already answered questions, it can still learn word vectors corresponding to already answered questions and, to some extent, solve the problem of answerer vectors tending to be averaged and unable to effectively represent the answerer. However, the Item2ve model cannot learn word vectors for newly emerging unanswered questions. That is, it cannot obtain word vectors corresponding to unanswered questions, and therefore cannot use word vectors to represent unanswered questions, ultimately leading to the inability to match answerers for unanswered questions.

[0031] To address the aforementioned issues, this disclosure provides a method for matching respondents, as detailed in the embodiments below. Figure 1This is a flowchart illustrating a matching method for respondents provided in an embodiment of this disclosure. The method may include the following steps:

[0032] Step S101: Based on the first word vector corresponding to the first keyword, determine the answerer vector corresponding to the candidate answerer. The first keyword is the keyword determined based on the candidate answerer's answered questions, and the answerer vector is the vector used to represent the questions that the candidate answerer can answer.

[0033] Step S102: Based on the second word vector corresponding to the second keyword and the answerer vector, determine the target answerer that matches the question to be answered from the candidate answerers. The second keyword is the keyword determined according to the question to be answered.

[0034] In the embodiments of this disclosure, the answer vector representing the candidate answerer's ability to answer questions is determined by using word vectors corresponding to keywords in already answered questions. This significantly reduces the problem of answer vectors failing to provide personalized representations of answerers due to the tendency of all answerers' answer vectors to converge to an average. Thus, the answer vector can more effectively represent the questions that an answerer can answer.

[0035] Furthermore, by representing the question to be answered using word vectors corresponding to the keywords of the question, it is possible to express various questions to be answered, including new questions, using vectors.

[0036] Therefore, by using the second word vector and the respondent vector, the target respondent that matches the question to be answered can be determined from the candidate respondents, which can more accurately match the target respondent for the question to be answered.

[0037] In the embodiments of this disclosure, the number of candidate respondents is at least one. In practical applications, the number of candidate respondents is often multiple. For each candidate respondent, the number of questions answered is also at least one. In practical applications, the number of questions answered is often multiple. In the embodiments of this disclosure, there is no specific limitation on the number of candidate respondents or the number of questions answered by each candidate respondent.

[0038] For different numbers of candidate answerers, regardless of the number of questions each candidate answerer has answered, the specific implementation of determining the answerer vector corresponding to a candidate answerer based on the first word vector corresponding to the first keyword in the embodiments of this disclosure can be as follows: First, for each candidate answerer, determine the first keyword corresponding to all the questions answered by that answerer; then, for each candidate answerer, determine the corresponding first word vector for each first keyword corresponding to all the questions answered; finally, for each candidate answerer, determine the answerer vector based on all the first word vectors. That is, for multiple candidate answerers, the answerer vector corresponding to each candidate answerer is determined one by one.

[0039] Answered questions refer to questions that the candidate answerer has previously answered. This can be obtained by extracting the candidate answerer's historical answer data from their historical behavior data.

[0040] Keywords are words in the description of answered or unanswered questions that reach a preset importance threshold. A description may contain only one keyword or more. Keyword extraction can be performed using either supervised or unsupervised methods.

[0041] Supervised keyword extraction methods can utilize a pre-trained keyword extraction model to extract keywords from answered or unanswered questions. This pre-trained model can be a model trained based on sample statements and corresponding labeled keywords. During keyword extraction, the keywords to be processed (keywords from answered or unanswered questions) can be input into the pre-trained keyword extraction model to obtain the corresponding keywords.

[0042] For example, by extracting keywords from four answered questions from a candidate respondent: "Beijing's second-home policy", "What kind of pension insurance is better", "Is a certain mobile phone worth buying", and "The advantages and disadvantages of mobile phone system A and mobile phone system B", the keywords for each of the four answered questions can be obtained.

[0043] Specifically, the keywords for "Beijing's second home policy" are: "Beijing" and "second home"; the keywords for "which pension insurance is better" are: "pension insurance"; the keywords for "is a certain mobile phone worth buying" are: "certain"; and the keywords for "the advantages and disadvantages of mobile phone system A and mobile phone system B" are: "A" and "B".

[0044] While supervised keyword extraction models can extract keywords accurately and quickly, they require extensive manual data labeling, leading to high costs. Unsupervised keyword extraction, on the other hand, is less costly, but often produces less accurate results. In practice, a customized method can be selected based on predefined requirements to extract keywords.

[0045] For each candidate answerer, the specific method of "determining the answerer vector corresponding to the candidate answerer based on the first word vector corresponding to the first keyword" is the same. Therefore, in the embodiments of this disclosure, when describing in detail the specific implementation of determining the answerer vector corresponding to the candidate answerer based on the first word vector corresponding to the first keyword, only one candidate answerer will be used as an example for illustration.

[0046] To achieve more accurate answerer matching for questions to be answered, embodiments of this disclosure may first classify the questions to be answered and the already answered questions into question types. When determining the answerer vector corresponding to a candidate answerer, answerer vectors for different question types are determined based on different question types.

[0047] Please refer to the detailed steps. Figure 2 , Figure 2 This is a flowchart of a first method for determining the answer vector provided in an embodiment of this disclosure.

[0048] Step S201: Using a pre-set question type classification strategy, determine the first question type, which is at least one question type determined based on the questions that have been answered.

[0049] Step S202: Using the first word vector, determine the answer vector corresponding to the first question type.

[0050] From a word vector perspective, keywords in questions of the same category tend to have higher semantic similarity, and the word vectors corresponding to these keywords will also have higher similarity. Determining the answerer vector for different question types based on the first word vector corresponding to each question type allows for a more refined and accurate representation of the answerer.

[0051] A pre-defined question type classification strategy can refer to classifying questions based on a pre-trained question type classification model. This model is typically trained based on question samples and corresponding labeled question types. In this case, the specific implementation of determining the first question type using the pre-defined question type classification strategy is as follows: input the answered questions into the pre-trained question type classification model to obtain the first question type corresponding to the answered questions.

[0052] In addition, the pre-set problem type classification strategy can be a strategy for classifying problem types based on prior values ​​or pre-defined problem types.

[0053] In the embodiments of this disclosure, common problem types include, but are not limited to, lifestyle, electronics, technology, food, history, and military.

[0054] For example, the answered questions, such as "Beijing's second home policy", "What kind of pension insurance is better", "Is a certain mobile phone worth buying", and "The advantages and disadvantages of mobile phone system A and mobile phone system B", can be categorized as follows: 1. Lifestyle: "Beijing's second home policy" and "What kind of pension insurance is better"; 2. Electronics: "Is a certain mobile phone worth buying" and "The advantages and disadvantages of mobile phone system A and mobile phone system B".

[0055] Determining the respondent vector corresponding to the first question type means determining the respondent vector of the candidate respondent under different question types. For example, the respondent vector of the candidate respondent under life-related questions, and the respondent vector of the candidate respondent under electronics-related questions.

[0056] To more conveniently and clearly determine the answerer vectors for different question types, in the embodiments of this disclosure, the steps for determining the answerer vector corresponding to the first question type are as follows: Figure 3 As shown, Figure 3 This is a flowchart of a second method for determining the answer master vector provided in the embodiments of this disclosure.

[0057] Step S301: Extract keywords from the answered questions according to the first question type to obtain the first keyword corresponding to the first question type.

[0058] Step S302: Using the first keyword corresponding to the first question type, determine the first word vector corresponding to the first question type.

[0059] Step S303: Based on the first word vector corresponding to the first question type, determine the answer vector corresponding to the first question type.

[0060] Because keywords in questions of the same category tend to have higher semantic similarity, the word vectors corresponding to those keywords will also have higher similarity. Therefore, learning word vectors according to question type can reveal the implicit hypotheses of word vectors.

[0061] The specific implementation method of extracting keywords from answered questions according to the first question type is as follows: First, according to the question type, answered questions belonging to the same question type are treated as the same question sequence; then, keywords are extracted from the answered questions for different question sequences; finally, the first keyword corresponding to the first question type is obtained, and the first keyword corresponding to the first question type is treated as a keyword sequence.

[0062] Specifically, for the already answered questions: "Beijing's second home policy", "What kind of pension insurance is better", "Is a certain mobile phone worth buying", and "The advantages and disadvantages of mobile phone system A and mobile phone system B", keywords can be extracted according to the first question type to obtain the first keywords corresponding to the first type: 1. Life category: "Beijing", "second home" and "pension insurance"; 2. Electronic category: "certain", "A" and "B".

[0063] Determining the answerer vector by using the word vectors corresponding to keywords, compared to determining the answerer vector based on the word vectors corresponding to already answered questions, will significantly improve the problem of answerer vectors from different answerers tending to be averaged.

[0064] Furthermore, when matching target answerers, the word vectors corresponding to the keywords of the question to be answered are matched with the answerer's vector. That is, representing the question to be answered based on its keywords can avoid the situation where word vector representation cannot be performed for newly emerging questions.

[0065] Therefore, in the embodiments of this disclosure, when learning the word vectors corresponding to the answered questions and the questions to be answered, either the Word2vec model or the Item2ve model can be used.

[0066] Since the answerer vector can be more accurately representative, in the embodiments of this disclosure, the specific implementation of determining the answerer vector based on the first word vector is often as follows: First, calculate the mean of the vectors corresponding to the first word vector; then, use the mean of the vectors as the answerer vector.

[0067] To improve the matching efficiency and accuracy of target answerers, in the embodiments of this disclosure, the question type corresponding to the question to be answered can be determined first, and answerer vectors with the same question type can be filtered out. Then, based on the second word vector and the answerer vector corresponding to the second question type, the target answerer can be determined. For the specific implementation process, please refer to... Figure 4 , Figure 4 This is a flowchart of a third method for determining the answer master vector provided in the embodiments of this disclosure.

[0068] Step S401: In the respondent vector corresponding to the first question type, determine the respondent vector corresponding to the second question type, where the second question type is the question type corresponding to the question to be answered.

[0069] Step S402: Determine the target answerer based on the second word vector and the answerer vector corresponding to the second question type.

[0070] In the embodiments of this disclosure, in order to intuitively and conveniently determine the target answerer, regardless of whether the answerer vector is first filtered based on the question type, the method for determining the target answerer is as follows: First, calculate the vector similarity between the answerer vector and the second word vector; then, based on the vector similarity, determine the target answerer vector from the answerer vector; finally, take the candidate answerer corresponding to the target answerer vector as the target answerer.

[0071] When determining the target answerer vector, the answerer vector with the highest feature similarity to the second word vector can be used as the target answerer vector.

[0072] Furthermore, to more simply and accurately determine the target answerer, embodiments of this disclosure can also employ a pre-trained vector matching model to determine the target answerer vector based on vector similarity, thereby identifying the target answerer. The specific process is as follows: the answerer vector and the second word vector are input into the pre-trained vector matching model to obtain vector similarity. Based on the vector similarity, it is determined whether the candidate answerer matches the question to be answered. The candidate answerer that matches the question to be answered is selected as the target answerer.

[0073] In the embodiments of this disclosure, the pre-trained vector matching model includes: a model trained based on the matching results of sample word vectors and corresponding annotations.

[0074] Specifically, it can be a deep neural network (DNN) model consisting of only an input layer, hidden layers, and an output layer. In this case, the input layer can be a concatenated vector of the respondent and a second word vector, and the output layer outputs whether the candidate respondent matches the question to be answered.

[0075] like Figure 5 As shown, embodiments of this disclosure provide a matching device for respondents, the device comprising:

[0076] Answerer vector determination unit 501 is used to determine the answerer vector corresponding to the candidate answerer based on the first word vector corresponding to the first keyword. The first keyword is a keyword determined based on the candidate answerer's answered questions, and the answerer vector is a vector used to represent the questions that the candidate answerer can answer.

[0077] The target answerer matching unit 502 is used to determine the target answerer that matches the question to be answered from the candidate answerers based on the second word vector corresponding to the second keyword and the answerer vector. The second keyword is the keyword determined according to the question to be answered.

[0078] In another embodiment of this disclosure, the answer vector determination unit 501 may further include:

[0079] The first question type determination subunit is used to determine the first question type using a pre-set question type classification strategy. The first question type is at least one question type determined based on the questions that have been answered.

[0080] The first answer vector determination subunit is used to determine the answer vector corresponding to the first question type using the first word vector.

[0081] In another embodiment of this disclosure, the first answer principal vector determining subunit may further include:

[0082] The keyword acquisition sub-unit is used to extract keywords from answered questions according to the first question type, and obtain the first keyword corresponding to the first question type;

[0083] The word vector determination subunit is used to determine the first word vector corresponding to the first question type using the first keyword corresponding to the first question type;

[0084] The second answer master vector determination subunit is used to determine the answer master vector corresponding to the first question type based on the first word vector corresponding to the first question type.

[0085] In another embodiment of this disclosure, the answer vector determination unit 501 may further include:

[0086] The question type determination subunit is used to determine the answerer vector corresponding to the second question type from the answerer vector corresponding to the first question type. The second question type is the question type corresponding to the question to be answered.

[0087] The first target answerer determination subunit is used to determine the target answerer based on the second word vector and the answerer vector corresponding to the second question type.

[0088] In another embodiment of this disclosure, the target answerer matching unit 502 may further include:

[0089] The similarity calculation subunit is used to calculate the vector similarity between the answer vector and the second word vector.

[0090] The target answerer vector calculation subunit is used to determine the target answerer vector from the answerer vectors based on vector similarity.

[0091] The third target answerer determination sub-unit is used to identify the candidate answerers corresponding to the target answerer vector as the target answerers.

[0092] In another embodiment of this disclosure, the answer vector determination unit 501 may further include:

[0093] The vector mean calculation subunit is used to calculate the vector mean corresponding to the first word vector;

[0094] The third answer master vector subunit is used to take the vector mean as the answer master vector.

[0095] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0096] According to embodiments of this disclosure, this disclosure also provides an electronic device and a readable storage medium.

[0097] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0098] like Figure 6 As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 602 or a computer program loaded from storage unit 608 into random access memory (RAM) 603. RAM 603 may also store various programs and data required for the operation of device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0099] Multiple components in device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0100] The computing unit 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the answer matching method. For example, in some embodiments, the answer matching method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the answer matching method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the answer matching method by any other suitable means (e.g., by means of firmware).

[0101] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0102] Program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0103] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0104] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0105] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0106] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0107] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0108] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for matching answerers, comprising: Based on the first word vector corresponding to the first keyword, the answerer vector corresponding to the candidate answerer is determined. The first keyword is determined according to the keyword corresponding to each of the answered questions for each candidate answerer. The method for determining the first word vector includes: for each candidate answerer's each answered question, determining the first word vector corresponding to the first keyword respectively; the answerer vector is a vector used to represent the questions that the candidate answerer can answer, and the answerer vector is determined based on all the first word vectors corresponding to each candidate answerer; Based on the second word vector corresponding to the second keyword and the answerer vector, a target answerer matching the question to be answered is determined from the candidate answerers, wherein the second keyword is a keyword determined according to the question to be answered; The step of determining the answerer vector corresponding to the candidate answerer includes: Using a pre-set question type segmentation strategy, a first question type is determined, which is at least one question type determined based on the answered questions; using the first word vector, the answerer vector corresponding to the first question type is determined; the pre-set question type segmentation strategy is: performing question type segmentation based on a pre-trained question type segmentation model, wherein the question type segmentation model is a model trained based on question samples and corresponding labeled question types; or, the pre-set question type segmentation strategy is a strategy for performing question type segmentation based on prior values ​​or pre-defined question types; Wherein, determining the respondent vector corresponding to the first question type includes: Based on the first question type, keywords are extracted from the answered questions to obtain the first keywords corresponding to the first question type; Using the first keyword corresponding to the first question type, determine the first word vector corresponding to the first question type; Based on the first word vector corresponding to the first question type, determine the answerer vector corresponding to the first question type; The step of determining the target answerer matching the question to be answered from the candidate answerers includes: using a pre-trained vector matching model to determine the target answerer based on vector similarity; wherein the pre-trained vector matching model includes: a model trained based on the matching results of sample word vectors and corresponding annotations, including: a deep neural network model with an input layer, a hidden layer and an output layer, wherein the input of the input layer is: a concatenated answerer vector and a second word vector, and the output of the output layer is whether the candidate answerer matches the question to be answered.

2. The method according to claim 1, wherein the target respondent is determined based on vector similarity using a pre-trained vector matching model: In the respondent vector corresponding to the first question type, determine the respondent vector corresponding to the second question type, where the second question type is the question type corresponding to the question to be answered; The answerer vector and the second word vector are input into the pre-trained vector matching model to obtain vector similarity; Based on the vector similarity, it is determined whether the candidate respondent matches the question to be answered, and the candidate respondent that matches the question to be answered is taken as the target respondent.

3. The method according to claim 1 or 2, wherein, The steps for determining the answer principal vector include: Calculate the mean of the vectors corresponding to the first word vector; The mean of the vectors is used as the main vector of the answer.

4. A matching device for respondents, comprising: The answerer vector determination unit is used to determine the answerer vector corresponding to the candidate answerer based on the first word vector corresponding to the first keyword. The first keyword is determined according to the keywords corresponding to each answered question for each candidate answerer. The method for determining the first word vector includes: for each candidate answerer's each answered question, determining the first word vector corresponding to the first keyword respectively; the answerer vector is a vector used to represent the questions that the candidate answerer can answer, and the answerer vector is determined based on all the first word vectors corresponding to each candidate answerer; The target answerer matching unit is used to determine the target answerer that matches the question to be answered from the candidate answerers based on the second word vector corresponding to the second keyword and the answerer vector, wherein the second keyword is a keyword determined according to the question to be answered; The answer main vector determination unit includes: The first question type determination subunit is used to determine a first question type using a pre-set question type classification strategy. The first question type is at least one question type determined based on the answered questions. The pre-set question type classification strategy is: classifying question types based on a pre-trained question type classification model, wherein the question type classification model is a model trained based on question samples and corresponding labeled question types; or, the pre-set question type classification strategy is a strategy for classifying question types based on prior values ​​or pre-defined question types. The first answer vector determination subunit is used to determine the answer vector corresponding to the first question type using the first word vector; The first answer principal vector determines the sub-unit, including: The keyword acquisition subunit is used to extract keywords from the answered questions according to the first question type, and obtain the first keyword corresponding to the first question type; The word vector determination subunit is used to determine the first word vector corresponding to the first question type using the first keyword corresponding to the first question type; The second answer master vector determination subunit is used to determine the answer master vector corresponding to the first question type based on the first word vector corresponding to the first question type; The respondent vector determination unit is further configured to determine the target respondent based on vector similarity using a pre-trained vector matching model; wherein the pre-trained vector matching model includes a model trained based on the matching results of sample word vectors and corresponding annotations, comprising a deep neural network model with an input layer, a hidden layer, and an output layer, wherein the input layer is fed with a concatenated respondent vector and a second word vector, and the output layer is the result of whether the candidate respondent matches the question to be answered.

5. The apparatus according to claim 4, wherein, The answer vector determination unit includes: The question type determination subunit is used to determine the answerer vector corresponding to the second question type from the answerer vector corresponding to the first question type, wherein the second question type is the question type corresponding to the question to be answered; The first target answerer determination subunit is used to input the answerer vector and the second word vector into the pre-trained vector matching model to obtain vector similarity; based on the vector similarity, it is determined whether the candidate answerer matches the question to be answered, and the candidate answerer that matches the question to be answered is taken as the target answerer.

6. The apparatus according to claim 4 or 5, wherein, The answer vector determination unit includes: The vector mean calculation subunit is used to calculate the vector mean corresponding to the first word vector; The third answer master vector subunit is used to take the mean of the vectors as the answer master vector.

7. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 3.

8. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method described in any one of claims 1 to 3.

9. A computer program product comprising a computer program / instructions, wherein, When the computer program / instructions are executed by the processor, they implement the steps of the method described in claims 1 to 3.

Images