Label-based agent customer matching method, device and equipment and storage medium
By calculating the label similarity and cosine similarity between agents and customers for two-way matching, the problem of low adaptability between agents and customers is solved, service quality and customer stickiness are improved, and customer experience and agent efficiency are enhanced.
Patent Information
- Application Number
- CN202510879385.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-26
- Publication Date
- 2025-09-26
AI Technical Summary
In existing technologies, the low compatibility between agents and customers leads to poor service quality and insufficient customer stickiness, ignoring the two-way connection and understanding between agents and customers.
By obtaining the labels and marking times of agents and customers, calculating semantic similarity and cosine similarity, building label pairs, and performing two-way matching between agents and customers based on the matching value, the customer is assigned to the agent if the matching value is greater than the preset threshold.
It improves the compatibility between agents and customers, enhances service quality and customer stickiness, enhances customer experience and agent work efficiency, and reduces training costs.
Smart Images
Figure CN120705606A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the fields of natural language processing and artificial intelligence, and is applicable to Internet finance scenarios. In particular, it relates to a tag-based agent-customer matching method, device, equipment, and storage medium. Background Art
[0002] Crowd tagging, a key tool for business operations, personalizes products by tagging user behaviors and characteristics. This allows for personalized service based on tags, fostering a customer-centric approach. Crowd tagging is widely used in the internet finance industry. However, the industry currently focuses excessively on customer tagging, emphasizing a one-way cycle of "from customer to customer," neglecting the compatibility between agents and customers. This failure to effectively establish a two-way connection and understanding between agents and customers results in poor service quality and insufficient customer retention. Summary of the Invention
[0003] The present invention provides a tag-based agent-customer matching method, device, equipment and storage medium to solve the technical problems of poor service quality and insufficient customer stickiness caused by low compatibility between agents and customers in the existing technology when performing agent-customer matching.
[0004] In a first aspect, a tag-based agent-customer matching method is provided, the tag-based agent-customer matching method comprising:
[0005] Get all the agent tags and corresponding marking times of the agents to be matched, and get all the customer tags and corresponding marking times of the customers to be matched;
[0006] Calculating the semantic similarity between each of the agent labels and each of the customer labels, and selecting the agent labels and the customer labels whose semantic similarity is greater than a first preset threshold to form a label pair;
[0007] Performing cosine similarity calculation based on the number of times the agent label and the customer label in the label pair are respectively marked and the cosine similarity formula to obtain a matching value between the to-be-matched agent and the to-be-matched customer;
[0008] If the matching value is greater than a second preset threshold, the customer to be matched is assigned to the agent to be matched.
[0009] In a second aspect, a tag-based agent-customer matching device is provided, wherein the tag-based agent-customer matching device comprises:
[0010] The tag acquisition module is used to obtain all the agent tags and corresponding tag times of the agents to be matched, and to obtain all the customer tags and corresponding tag times of the customers to be matched;
[0011] A first matching module is configured to calculate the semantic similarity between each of the agent labels and each of the customer labels, and select the agent labels and the customer labels whose semantic similarity is greater than a first preset threshold to form a label pair;
[0012] A second matching module is configured to perform cosine similarity calculation based on the number of times the agent label and the customer label in the label pair are marked and a cosine similarity formula to obtain a matching value between the to-be-matched agent and the to-be-matched customer;
[0013] The determination allocation module is configured to allocate the customer to be matched to the agent to be matched if the matching value is greater than a second preset threshold.
[0014] In a third aspect, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned tag-based agent-customer matching method are implemented.
[0015] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned tag-based agent-customer matching method are implemented.
[0016] In the solution implemented by the above-mentioned label-based agent-customer matching method, device, equipment and storage medium, the agent label of the agent to be matched and the corresponding number of markings, and the customer label of the customer to be matched and the corresponding number of markings are first obtained. Then, the agent label and customer label whose semantic similarity is greater than the first preset threshold are selected as the agent similarity label and customer similarity label. Finally, based on the marking times of the agent similarity label and the marking times of the customer similarity label and the cosine similarity formula, the cosine similarity is calculated to obtain the matching value of the agent to be matched and the customer to be matched. If the matching value is greater than the second preset threshold, the customer to be matched is assigned to the agent to be matched. Bidirectional matching of agents and customers based on labels increases the adaptability between agents and customers, improves service quality and customer stickiness. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0018] Figure 1 The figure is a schematic diagram of an application environment of a tag-based agent-customer matching method provided by an embodiment of the present invention.
[0019] Figure 2 The figure is a flowchart of an implementation method of a tag-based agent-customer matching method provided by an embodiment of the present invention.
[0020] Figure 3 yes Figure 2 Schematic diagram of a flow chart of a specific implementation of step S30.
[0021] Figure 4 yes Figure 2 A schematic flow chart of a specific implementation method before step S20.
[0022] Figure 5 yes Figure 4 A flowchart of a specific implementation of step S10 is shown in FIG.
[0023] Figure 6 yes Figure 2 A flowchart of a specific implementation of step S30 is shown in FIG.
[0024] Figure 7 This is a flowchart of another implementation of the tag-based agent-customer matching method provided by one embodiment of the present invention.
[0025] Figure 8 This is a schematic diagram of a tag-based agent-customer matching device in one embodiment of the present invention.
[0026] Figure 9 It is a structural diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION
[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.
[0028] The tag-based agent-customer matching method provided by the embodiment of the present invention can be applied in the following situations: Figure 1 In the application environment, Figure 1 A schematic diagram of an application environment of a tag-based agent-customer matching method provided by an embodiment of the present invention is shown as follows: Figure 1As shown, this application environment includes a terminal 10 and a server 20. The terminal 10 and the server 20 jointly execute the tag-based agent-customer matching method. It should be noted that the terminal 10 and the server 20 can be, but are not limited to, smartphones, tablet computers, laptop computers, desktop computers, etc. The terminal 10 and the server 20 can be connected via Bluetooth, USB (Universal Serial Bus), or other communication connection methods, and the present invention is not limited thereto.
[0029] Figure 2 A flowchart of an implementation method of a tag-based agent-customer matching method provided by an embodiment of the present invention is shown as follows: Figure 2 As shown, the tag-based agent-customer matching method may specifically include the following steps:
[0030] S20: Obtain all agent tags and corresponding marking times of the agents to be matched, and obtain all customer tags and corresponding marking times of the customers to be matched.
[0031] In this step, both the agent to be matched and the customer to be matched have multiple tags, each of which has a corresponding tag count. When obtaining the agent tag of the agent to be matched, the tag count corresponding to each agent tag is also obtained. When obtaining the customer tag of the customer to be matched, the tag count for each customer tag is also obtained. The tag count of an agent tag can, to a certain extent, reflect the degree of compatibility between the agent tag and the agent to be matched. A higher tag count indicates a higher degree of compatibility between the agent tag and the agent to be matched. Similarly, the tag count of a customer tag can also reflect the degree of compatibility between the customer tag and the customer to be matched.
[0032] For the labels of prospective agents and customers to be matched obtained based on basic attribute information, such as gender, age, etc., the number of markings is usually set to 1. For the labels of prospective agents and customers to be matched obtained based on historical interaction information, the number of markings is related to the historical interaction records. For example, if the prospective agent has recently received more than three customer complaints, the number of markings for the label "poor customer feedback" is 3; if the prospective customer has recently purchased pet insurance 5 times, the number of markings for the label "pet insurance" is 5. For the labels of prospective agents and customers to be matched obtained based on recent status information, the number of markings is obtained according to their recent behavioral logic. For example, if the prospective agent has recently participated in new energy marketing courses 3 times, the number of markings for the label "potential to develop new energy vehicle insurance business" is 3; if the prospective customer has browsed new energy vehicle insurance 20 times on the platform, the number of markings for the label "tendency to purchase new energy vehicle insurance" is 20.
[0033] S30: Calculate the semantic similarity between each of the agent tags and each of the customer tags, and select the agent tags and the customer tags whose semantic similarity is greater than a first preset threshold to form a tag pair.
[0034] In this step, the semantic similarity between each agent tag and each customer tag is calculated one by one. Agent tags and customer tags with semantic similarity greater than a first preset threshold are selected to form tag pairs. The setting of this first preset threshold is generally related to the tag configuration rules, and the specific value can be set according to actual circumstances. If no tag pair exists, it indicates that the agent and customer to be matched have no similar tags, and the subsequent steps are not required to conclude that the agent and customer to be matched are not a match. If a tag pair exists, the subsequent steps are performed. The number of tag pairs may be one or more.
[0035] S40: Perform cosine similarity calculation based on the number of times the agent label and the customer label in the label pair are marked and a cosine similarity formula to obtain a matching value between the to-be-matched agent and the to-be-matched customer.
[0036] In this step, the number of times the agent tag in the tag pair has been marked is obtained, and the number of times the customer tag in the tag pair has been marked is also obtained. It is understood that the number of times the agent tag has been marked can reflect the degree of compatibility between the agent tag and the agent to be matched, and the number of times the customer tag has been marked can also reflect the degree of compatibility between the customer tag and the customer to be matched. The agent tag and the customer tag in the tag pair are similar tags with a semantic similarity greater than a first preset threshold, and can reflect the same attribute of the agent to be matched and the customer to be matched. The closer their number of marks, the higher the degree of compatibility between the agent to be matched and the customer to be matched. Therefore, based on the cosine similarity calculation based on the number of marks corresponding to the agent tag and the customer tag in the tag pair, the resulting result, i.e., the matching value, can intuitively reflect the degree of matching between the agent to be matched and the customer to be matched.
[0037] S50: If the matching value is greater than a second preset threshold, assign the customer to be matched to the agent to be matched.
[0038] In this step, the cosine similarity calculation results are calculated in the range [-1, 1]. The closer the result is to 1, the higher the degree of match between the agent and the customer. The closer the result is to -1, the lower the degree of match between the agent and the customer. A second preset threshold can be set as needed, typically between [0.5, 1]. When the matching value is greater than the second preset threshold, it indicates a high degree of match between the agent and the customer. In this case, the customer is assigned to the agent.
[0039] The tag-based agent-customer matching method of this embodiment first obtains the agent tag and corresponding tag count of the agent to be matched, and the customer tag and corresponding tag count of the customer to be matched. Then, agent tags and customer tags whose semantic similarity exceeds a first preset threshold are selected as agent similarity tags and customer similarity tags. Finally, cosine similarity is calculated based on the tag counts of the agent similarity tags and the customer similarity tags and the cosine similarity formula to obtain a matching value for the agent to be matched and the customer to be matched. If the matching value exceeds a second preset threshold, the customer to be matched is assigned to the agent to be matched. This tag-based two-way matching of agents and customers increases the compatibility between agents and customers, improving service quality and customer loyalty.
[0040] Furthermore, associating customers with agents not only strengthens customer loyalty but also enhances the customer experience when interacting with familiar or well-regarded agents, allowing them to lower their guard. This also allows for targeted agent training, tailored to specific agent types, saving employee time and energy and improving agent efficiency. Furthermore, making customer feedback a factor in matching customers with agents enhances their sense of engagement and control.
[0041] In a specific embodiment, in an Internet finance scenario, all agent labels and corresponding marking times of agent A to be matched are obtained, and all customer labels and corresponding marking times of customer B to be matched are obtained. Agent A's labels are 80s (1), best performance last month (1), good service attitude (10), car insurance (20), critical illness insurance (4), good reviews from female customers (6), timely feedback (15), and high affinity (9). Customer B's labels are 80s (1), married (1), female (1), purchased critical illness insurance (3), may purchase car insurance (6), and likes customer service with high affinity (3). The number in the brackets is the number of markings corresponding to the label. Calculate the semantic similarity of each agent label and each customer label, and select the agent label and customer label whose semantic similarity is greater than the first preset threshold to construct a label pair. A cosine similarity calculation is performed based on the number of markings corresponding to the seat label and the customer label in the label pair to obtain a matching value between the to-be-matched seat A and the to-be-matched customer B. If the matching value is greater than a second preset threshold, the to-be-matched customer A is assigned to the to-be-matched seat B.
[0042] Figure 3 yes Figure 2 A flow chart of a specific implementation method in step S30 is shown in FIG. Figure 3 As shown, in some embodiments, step S30 includes:
[0043] S31. Input all the agent labels and all the customer labels into a pre-trained label classification model to calculate the semantic similarity between each agent label and each customer label, and output the agent label and the customer label whose semantic similarity is greater than a first preset threshold to form a label pair.
[0044] In this step, the agent label and customer label are input into a pre-trained label classification model, and then the agent label and customer label are subjected to text preprocessing and vectorization to obtain the agent label vector and customer label vector. The cosine similarity between the agent label vector and the customer label vector is then calculated. By judging whether the similarity reaches a first preset threshold, it is determined whether the input agent label and customer label constitute a label pair.
[0045] In some implementations, the cosine similarity formula in step S40 includes:
[0046]
[0047] Among them, x i Indicates the number of times the seat label in the i-th label pair is marked, y i = ∑ i = 1, i.e., the number of times the customer label is marked in the i-th label pair, and n is the number of label pairs. cosθ is the cosine similarity, i.e., the matching value between the agent to be matched and the customer to be matched.
[0048] In this step, the number of times the agent label and the number of times the customer label in all label pairs are marked are input into the cosine similarity calculation formula to obtain the matching value of the agent to be matched and the customer to be matched.
[0049] In a specific embodiment, there are 7 tag pairs between the to-be-matched agent A and the to-be-matched customer B. The tags and tag times in the tag pairs are shown in Table 1. The numbers in brackets in Table 1 are the tag times.
[0050] Table 1
[0051] Label to serial number Seat A to be matched Customer B to be matched 1 Post-90s(1) Post-90s(1) 2 Shenzhen (1) Shenzhen (1) 3 Sweet voice (10) Like sweet voice (4) 4 Pet Insurance (20) Buy pet insurance (3) 5 Young Killer (20) Youth (1) 6 Many VIP customers (3) VIP Customer(1) 7 Potential to develop auto insurance business (3) Tendency to purchase auto insurance (5)
[0052] Substituting the data in Table 1 into the above cosine similarity calculation formula, we can get:
[0053]
[0054] That is, the matching degree between agent A and customer B is 0.63.
[0055] Figure 4 yes Figure 2 A specific implementation flow diagram before step S20 is shown in FIG. Figure 4As shown, in some embodiments, it also includes:
[0056] S10: Construct the agent tag based on the data information of the agent to be matched, and construct the customer tag based on the data information of the customer to be matched.
[0057] In this embodiment, an agent tag is constructed based on the agent's data. This data includes basic attribute information, historical interaction information, and recent status information. Basic attribute information includes information reflecting the agent's basic attributes, such as gender, age, and hours worked. Historical interaction information includes customer reviews, historical transaction volume, and other information about interactions with customers. Recent status information includes information about the agent's recent physical condition, work status, and other status. The agent tag constructed based on this data provides a relatively comprehensive overview of the agent's overall situation.
[0058] Similarly, the customer tag is constructed based on the data information of the customer to be matched, and the data information of the customer to be matched includes basic attribute information, historical interaction information and recent status information. Among them, the basic attribute information includes information reflecting the basic attributes of the customer, such as gender, age, and actively set personal tendencies. Historical interaction information includes historical consumption records, service evaluations, and other interaction information with agents and platforms. Recent status information includes recent status information such as the customer's browsing history on the platform and consumption behavior that is inconsistent with historical consumption behavior. The customer tag constructed based on the above-mentioned data information of the customer to be matched can more comprehensively reflect the overall situation of the customer to be matched.
[0059] Figure 5 yes Figure 4 A specific implementation flow diagram of step S10 is shown in FIG. Figure 5 As shown, step S10 includes:
[0060] S11. Construct a first agent tag based on basic attribute information of the agent to be matched, construct a second agent tag based on historical interaction information of the agent to be matched, and construct a third agent tag based on recent status information of the agent to be matched.
[0061] S12: construct a first customer tag based on basic attribute information of the customer to be matched, construct a second customer tag based on historical interaction information of the customer to be matched, and construct a third customer tag based on recent status information of the customer to be matched.
[0062] In this embodiment, the first, second, and third agent labels are constructed based on the basic attribute information, historical interaction information, and recent status information of the agent to be matched. The first, second, and third customer labels are constructed based on the basic attribute information, historical interaction information, and recent status information of the customer to be matched. Constructing corresponding agent and customer labels based on the same type of data improves the efficiency of subsequent similarity calculations between agent and customer labels.
[0063] Figure 6 yes Figure 2 A specific implementation flow diagram of step S30 is shown in FIG. Figure 6 As shown, in some embodiments, step S30 may include:
[0064] S31, calculating the semantic similarity between each first agent tag and each first customer tag, and
[0065] Calculate the semantic similarity between each second agent label and each second customer label, and
[0066] The semantic similarity between each of the third agent tags and each of the third customer tags is calculated.
[0067] In this embodiment, when calculating the semantic similarity between each agent tag and each customer tag, the similarity of the corresponding agent tag and customer tag is calculated separately, which can reduce the amount of calculation and thus improve the efficiency of calculating the similarity between the agent tag and the customer tag.
[0068] Figure 7 FIG. 1 is a flow chart of another implementation of the tag-based agent-customer matching method provided by an embodiment of the present invention. Figure 7 As shown, in some embodiments, after step S50, the method further includes:
[0069] S51: If the matching value is less than a second preset threshold, assign the customer to be matched to another agent to be matched for agent-customer matching.
[0070] In this step, if the matching value is less than the second preset threshold, it means that the compatibility between the customer to be matched and the agent to be matched is insufficient. At this time, the customer to be matched is assigned to another agent to be matched and the agent-customer matching method based on the label in any of the above embodiments is re-executed until the customer to be matched is matched with an agent that meets the compatibility requirements.
[0071] The tag-based agent-customer matching method provided in the embodiment of the present invention can be constructed based on artificial intelligence, and relevant data can be acquired and processed based on artificial intelligence technology to achieve unattended tag-based agent-customer matching. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. Basic artificial intelligence technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. Artificial intelligence software technology mainly includes several major directions such as computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.
[0072] In one embodiment, a tag-based agent-customer matching device is provided, which corresponds to the tag-based agent-customer matching method in the above embodiment. Figure 8 As shown, the tag-based agent-customer matching device includes:
[0073] The tag acquisition module 200 is used to obtain all agent tags and corresponding tag times of the agents to be matched, and obtain all customer tags and corresponding tag times of the customers to be matched;
[0074] A first matching module 300 is configured to calculate the semantic similarity between each of the agent tags and each of the customer tags, and select the agent tags and the customer tags whose semantic similarity is greater than a first preset threshold to form a tag pair;
[0075] The second matching module 400 is configured to perform cosine similarity calculation based on the number of times the agent label and the customer label in the label pair are marked and the cosine similarity formula to obtain a matching value between the to-be-matched agent and the to-be-matched customer;
[0076] The determination allocation module 500 is configured to allocate the customer to be matched to the agent to be matched if the matching value is greater than a second preset threshold.
[0077] Specifically, the first matching module 300 includes a model matching unit.
[0078] The model matching unit is used to input all the seat labels and all the customer labels into a pre-trained label classification model to calculate the semantic similarity of each seat label and each customer label, so as to output the seat label and the customer label whose semantic similarity is greater than a first preset threshold to form a label pair.
[0079] Specifically, the second matching module 400 is further configured to calculate the cosine similarity using the following formula:
[0080]
[0081] Among them, x i Indicates the number of times the seat label in the i-th label pair is marked, y i = ∑ i = 1, i.e., the number of times the customer label is marked in the i-th label pair, and n is the number of label pairs. cosθ is the cosine similarity, i.e., the matching value between the agent to be matched and the customer to be matched.
[0082] Specifically, the tag-based agent-customer matching device further includes a tag building module.
[0083] The label construction module is used to construct the agent label based on the data information of the agent to be matched, and to construct the customer label based on the data information of the customer to be matched.
[0084] Specifically, the label construction module includes an agent label construction unit and a customer label construction unit.
[0085] The agent label construction unit is configured to construct a first agent label based on the basic attribute information of the agent to be matched, a second agent label based on the historical interaction information of the agent to be matched, and a third agent label based on the recent status information of the agent to be matched.
[0086] The customer tag construction unit is configured to construct a first customer tag based on basic attribute information of the customer to be matched, a second customer tag based on historical interaction information of the customer to be matched, and a third customer tag based on recent status information of the customer to be matched.
[0087] Specifically, the first matching module 300 includes a classification matching unit.
[0088] A classification matching unit is configured to calculate the semantic similarity between each of the first agent labels and each of the first customer labels, calculate the semantic similarity between each of the second agent labels and each of the second customer labels, and calculate the semantic similarity between each of the third agent labels and each of the third customer labels.
[0089] Specifically, the determination allocation module 500 is further configured to allocate the customer to be matched to another agent to be matched for agent-customer matching if the matching value is less than a second preset threshold.
[0090] The specific limitations of the tag-based agent-customer matching device can be found in the limitations of the tag-based agent-customer matching method described above and will not be further elaborated here. Each module within the tag-based agent-customer matching device described above may be implemented in whole or in part via software, hardware, or a combination thereof. Each of these modules may be embedded in or independent of a processor within a computer device in hardware form, or may be stored in a computer device memory in software form, allowing the processor to invoke and execute the corresponding operations of each module.
[0091] Figure 9 FIG. 1 is a schematic structural diagram of an embodiment of a computer device according to the present application, as shown in FIG. Figure 9 As shown, the present application also provides a computer device, including:
[0092] The memory and the processor store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes any step of the above-mentioned tag-based agent-customer matching method.
[0093] The present application also provides a computer-readable storage medium. When the computer-readable instructions are executed by one or more processors, the one or more processors execute any step in the label-based agent-customer matching method. It can be understood that the readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.
[0094] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.
Claims
1. A tag-based agent-customer matching method, characterized in that: include: Get all the agent tags and corresponding marking times of the agents to be matched, and get all the customer tags and corresponding marking times of the customers to be matched; Calculating the semantic similarity between each of the agent labels and each of the customer labels, and selecting the agent labels and the customer labels whose semantic similarity is greater than a first preset threshold to form a label pair; Performing cosine similarity calculation based on the number of times the agent label and the customer label in the label pair are respectively marked and the cosine similarity formula to obtain a matching value between the to-be-matched agent and the to-be-matched customer; If the matching value is greater than a second preset threshold, the customer to be matched is assigned to the agent to be matched.
2. The tag-based agent-customer matching method according to claim 1, wherein: The calculating the semantic similarity between each of the seat labels and each of the customer labels, and selecting the seat labels and the customer labels whose semantic similarity is greater than a first preset threshold to form a label pair, includes: All the seat labels and all the customer labels are input into a pre-trained label classification model to calculate the semantic similarity of each seat label and each customer label, so as to output the seat label and the customer label whose semantic similarity is greater than a first preset threshold to construct a label pair.
3. The tag-based agent-customer matching method according to claim 1, wherein: The cosine similarity formula includes: Among them, x i Indicates the number of times the seat label in the i-th label pair is marked, y i It represents the number of times the customer label is marked in the i-th label pair, and n represents the number of label pairs.
4. The tag-based agent-customer matching method according to claim 1, wherein: Before obtaining all the agent tags and corresponding marking times of the agents to be matched, and obtaining all the customer tags and corresponding marking times of the customers to be matched, the method further includes: The agent tag is constructed based on the data information of the to-be-matched agent, and the customer tag is constructed based on the data information of the to-be-matched customer.
5. The tag-based agent-customer matching method according to claim 4, characterized in that: The step of constructing the agent tag based on the data information of the agent to be matched and constructing the customer tag based on the data information of the customer to be matched includes: Constructing a first agent tag based on the basic attribute information of the to-be-matched agent, constructing a second agent tag based on the historical interaction information of the to-be-matched agent, and constructing a third agent tag based on the recent status information of the to-be-matched agent; A first customer tag is constructed based on the basic attribute information of the customer to be matched, a second customer tag is constructed based on the historical interaction information of the customer to be matched, and a third customer tag is constructed based on the recent status information of the customer to be matched.
6. The tag-based agent-customer matching method according to claim 5, characterized in that: The calculating of the semantic similarity between each of the agent labels and each of the customer labels includes: Calculate the semantic similarity between each first agent label and each first customer label, and Calculate the semantic similarity between each second agent label and each second customer label, and The semantic similarity between each of the third agent tags and each of the third customer tags is calculated.
7. The tag-based agent-customer matching method according to claim 1, wherein: After obtaining the matching value of the to-be-matched seat and the to-be-matched customer, the method further includes: If the matching value is less than a second preset threshold, the customer to be matched is assigned to another agent to be matched for agent-customer matching.
8. A tag-based agent-customer matching device, characterized in that: include: The tag acquisition module is used to obtain all the agent tags and corresponding tag times of the agents to be matched, and to obtain all the customer tags and corresponding tag times of the customers to be matched; A first matching module is configured to calculate the semantic similarity between each of the agent labels and each of the customer labels, and select the agent labels and the customer labels whose semantic similarity is greater than a first preset threshold to form a label pair; A second matching module is configured to perform cosine similarity calculation based on the number of times the agent label and the customer label in the label pair are marked and a cosine similarity formula to obtain a matching value between the to-be-matched agent and the to-be-matched customer; The determination allocation module is configured to allocate the customer to be matched to the agent to be matched if the matching value is greater than a second preset threshold.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the tag-based agent-customer matching method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the tag-based agent-customer matching method according to any one of claims 1 to 7 are implemented.