Scheduling method for intelligent cloud call center to call multi-manufacturer AI (advanced technology attachment) capabilityTechnical Field
The invention relates to the technical field of intelligent cloud call centers, in particular to a scheduling method for scheduling multi-manufacturer AI (advanced technology interface) capability of an intelligent cloud call center.
Background
With the development and maturity of artificial intelligence and voice interaction technology, AI intelligent voice gradually goes into consumer life. Call centers, also known as customer service centers, originated in the 30 s of the 20 th century, are service institutions consisting of a collection of service personnel in a relatively centralized location. The call center contains two main types of services, outgoing and incoming. The calling type service becomes more and more the mainstream service of the market, actively dials the telephone for the customers to carry out the telemarketing and market investigation, and becomes one of the more popular competitive means of the market. The AI robot outbound improves the popularization efficiency and saves the outbound popularization cost, and enterprises are more and more inclined and use the robot outbound for businesses with weak mobility.
In 2020, the total number of mobile phone users in China reaches 15.9 hundred million users and keeps rising trend, so that more and more enterprises and merchants begin to pay attention to telemarketing and begin to pay attention to business opportunities and necessity in outbound marketing. The call center outbound marketing has numerous advantages, and can accurately analyze customer demands, improve the success rate, improve the brand image and the like.
At present, aiming at a large number of projects of AI robot outbound business, a relatively common method is to specify a robot model outbound: a specific robot outbound model is established and used for matching service default manual outbound to users, and the model adopts a question-answering mode, so that the model is single, and the user experience is relatively hard. In the case of multiple service scene models, the requirements on the supporting concurrency of an AI service engine (ASR/TTS/NLP) are high, and the single-point failure rate is also high.
In addition, aiming at the prior art, a multi-AI manufacturer integrated call center system scheme is provided, and the single-point failure rate is reduced while the AI service concurrency is improved. The concurrency of services of each AI manufacturer is different from the maturity of the industry field which is good for each AI manufacturer, so that the traffic marketing conversion rate is uneven. Therefore, when the optimal outbound resource is obtained by actually adopting the resource scheduling algorithm, how to ensure the reasonable utilization of the resources of multiple AI service manufacturers (the benefit of the AI manufacturers is maximized) and the intelligent outbound quality and efficiency at the same time is the problem to be solved by the technical scheme.
For the AI intelligent outbound system, if the dynamic acquisition robot route can be calculated rapidly and accurately, the utilization rate of the concurrency of each AI service manufacturer is improved, the obtained benefits of each AI manufacturer can be ensured, and the global benefits in the telecommunication industry can not be influenced by one AI manufacturer.
Disclosure of Invention
The invention aims to solve the problem that the service quality cannot be improved due to uneven resource scheduling among intelligent outbound multi-service manufacturers of a call center in the prior art, and provides a scheduling method for calling multi-manufacturer AI (automatic identification) capability based on a dynamic statistical process.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: a scheduling method for an intelligent cloud call center to invoke multi-manufacturer AI capability comprises the following steps:
s1: extracting manufacturer data under certain conditions, and labeling the manufacturer flow;
S2: initiating an intelligent outbound call, acquiring the maximum total score of an intelligent call flow according to the object data mark and the algorithm model, and initiating a call according to an intelligent call service flow;
s3: for historical service data, reversely calculating the flow service quality and marking the data;
s4: and according to the step S2 and the step S3, realizing multi-resource scheduling among multiple manufacturers of intelligent calling, and selecting an optimal manufacturer route.
Further, in the step S1, the extracting the data attribute of the data includes: intelligent call flow ID, belonging to the field, vendor name, vendor flow identification, preset concurrency, remaining concurrency, call times, vendor yield, vendor field score, quality of service score.
Further, in the step S2, a total score is calculated according to the following formula, and the total score is used to select the priority:
Wherein Q [ t ] represents the intelligent flow priority at the time t, the larger the value is, the more the call selection is, Pt represents the unit time benefit of manufacturers at the time t, Ut represents the intelligent call concurrency utilization rate at the time t, Mt represents the service quality at the time t, St represents the field score at the time t,
Wherein:
In the formula P, N represents the number of unit data sets, and mu represents the manufacturer yield; in the formula U, Et represents the concurrency quantity of unit time, and E represents the unified total concurrency of the system.
Further, in the step S2, the call is ended, the service quality of the manufacturer is marked, and the call record attribute includes a call ID, an AI flow ID, a service achievement status, and a call time.
Further, in the step S3, the flow service quality is calculated according to the following formula:
M represents the quality of service.
Further, in the step S3, the data tag of the outbound algorithm of the resource scheduling is verified through the following formula, and the tag characteristic value is adjusted;
wherein, P [ i, t ] represents the income amount of the manufacturer in unit time, and U [ t ] represents the concurrent utilization rate at the moment t.
The beneficial effects of the invention are as follows: the invention has the advantages of high flexibility, strong processing capacity, high cost performance, high stability and the like; the invention improves the flexibility of intelligent outbound multi-manufacturer resource scheduling through an iterative algorithm, avoids the bottleneck problem of resource waste caused by a fixed parameter algorithm, improves the efficiency of intelligent call multi-manufacturer resource scheduling, realizes reasonable distribution of AI multi-manufacturer benefits, improves the service quality, reduces the intelligent call deployment risk, and ensures the global benefit of intelligent voice in the outbound field.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
Fig. 1 is a flow algorithm device diagram of the present invention.
Detailed Description
The invention will now be described in detail with reference to the drawings and specific examples.
The invention provides a scheduling method for calling multi-manufacturer AI capability by an intelligent cloud call center, which comprises the following steps:
S1: and extracting manufacturer data under certain conditions, and labeling the manufacturer flow.
S1.1: the data attributes include: intelligent call flow ID, belonging to the field, vendor name, vendor flow identification, preset concurrency, remaining concurrency, number of calls, vendor yield, vendor field score (5-10), quality of service score (0-100).
The data of this example are for example the following: y1{1, scene 1, domain 1, vendor A, A001, 200, 50, 500,0.01,9,2} Y2{2, scene 1, domain 1, vendor B, B001, 200, 30, 600,0.02,9,0.83}
S2: initiating an intelligent outbound call, acquiring the maximum total score of an intelligent call flow according to the object data mark and the algorithm model, and initiating a call according to an intelligent call service flow;
q represents the intelligent flow priority, with the larger the value the more forward the call selection is; the calculation results respectively include:
Q1＝0.01*500/60+150/(200+200)+10/500*100+9＝11.121；
Q2＝0.02*600/60+180/(200+200)+5/600*100+9＝10.483；
S2.1: ending the call, marking the service quality of a manufacturer, wherein the call record attribute comprises a call ID, an AI flow ID, a service achievement state and call time; x (i) = (X1,x2,...xi,...,xN).
S3: and (5) reversely calculating the flow service quality aiming at the historical service data and marking the data.
M1＝10/500*100＝2,M2＝5/600*100＝0.833
S3.1: and verifying the data mark of the resource scheduling outbound algorithm, and adjusting the characteristic value of the mark.
N-bit data sets are provided with mu manufacturer yield, and P represents manufacturer unit time yield;
P1＝0.01*500/60＝0.083,P2＝0.02*600/60＝0.2。
e concurrency per unit timeE System sense total concurrency, U represents the concurrency utilization rate at a certain moment;
U1＝150/(200+200)＝0.38,U2＝180/(200+200)＝0.45。
S4: and according to S2-3, the multi-resource scheduling among multiple manufacturers of intelligent calls is realized, and the optimal manufacturer route is selected, so that the maximization of concurrent use of manufacturers and the maximization of outbound business marketing benefit are achieved.
In conclusion, by adopting the algorithm to perform test operation on the intelligent outbound history records of the history items, the average occupation of the resource scheduling of each AI manufacturer is greatly improved compared with the prior art, the degree of relevance of the flow calling ratio and the market score of each scene among multiple operators is required to reach 95%, and the service quality is improved by 1.5 percent. Through practice, the algorithm quantifies indexes of the scheduling data of the historical manufacturers, balances the service yield of each AI manufacturer, and improves the total concurrency efficiency of the call center.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It should be understood by those skilled in the art that the above embodiments do not limit the scope of the present invention in any way, and all technical solutions obtained by equivalent substitution and the like fall within the scope of the present invention.
The invention is not related in part to the same as or can be practiced with the prior art.