Search

CN-122001979-A - Telephone number selection outbound method, electronic device and storage medium

CN122001979ACN 122001979 ACN122001979 ACN 122001979ACN-122001979-A

Abstract

The application discloses a telephone number selection outbound method, electronic equipment and a storage medium, wherein the method comprises the steps of collecting historical outbound records of all outbound numbers in a target time range, and obtaining historical behavior characteristics of each outbound number; the method comprises the steps of obtaining compensation weights of outbound numbers of all lines according to historical behavior characteristics and lines where all outbound numbers are located, obtaining basic scores of all numbers according to the compensation weights and preset scores, obtaining behavior scores of all numbers according to the basic scores of all numbers and the last outbound result, sorting the behavior scores, preferentially selecting high-scoring outbound numbers for outbound, and updating the behavior scores in real time according to the outbound result after each outbound is finished. According to the technical scheme, the number use priority is optimized through the dynamic scoring mechanism, the accurate matching of the outbound number is realized by combining the historical behaviors of the user, the number survival period is prolonged, and the operation cost is reduced.

Inventors

  • ZHAO FENG
  • REN YANG
  • YUAN CHANGSHENG
  • XU XIAODONG
  • LI RUMAN
  • ZHAN ZHONGQIANG
  • DONG YUANYUAN

Assignees

  • 北京凌渡科技有限公司

Dates

Publication Date
20260508
Application Date
20260210

Claims (9)

  1. 1. A method for telephone number selection outbound, the method comprising: Collecting historical outbound records of all outbound numbers in a target time range, and obtaining historical behavior characteristics of each outbound number; acquiring compensation weights of the outbound numbers of all lines according to the historical behavior characteristics and the lines where the outbound numbers are located; according to the compensation weight and the preset score, obtaining the basic score of each number; Obtaining behavior scores of the numbers according to the basic scores of the numbers and the last outbound result; And after each outbound call is finished, updating the behavior scores in real time according to the outbound call result.
  2. 2. The method of claim 1, wherein the base score acquisition process includes taking a constant value as a preset score and taking a compensation weight x preset score result as a base score.
  3. 3. The method of claim 1, wherein the behavioral scoring results comprise a base score + an outbound score, the outbound score being obtained based on an outbound result classification.
  4. 4. The method according to claim 3, wherein the outbound score acquisition process comprises classifying outbound results into five categories according to outbound feedback, including A1 active on, A2 brief on, A3 user refusal, A4 unmanned answer, A5 user interception; Based on the outbound result classification, acquiring the outbound score of the corresponding type, and taking the outbound score as the outbound score.
  5. 5. The method of claim 4, wherein the A1 is effectively connected as an outbound call with a call duration greater than or equal to T, the A2 is briefly connected as an outbound call with a call duration greater than 0 but less than T, and the A4 is not connected as an outbound call with a ringing timeout, wherein T is set according to a called industry.
  6. 6. The method of claim 4, wherein the outbound call component acquisition process comprises, Acquiring a reference answering rate P_base according to the historical outbound records of the numbers, That is to say, p_base=n\u total_success N_total_calls; Calculating the condition answering rate P_i of the effective connection of the next outbound after each outbound classification in the historical outbound record occurs, and adopting Bayesian smoothing in the process of calculating the condition answering rate P_i to eliminate small probability errors, namely P_i= (N_success_after_ai+m) P_base)/(N_ai+m); From p_base and p_i, the probability difference Δp_i is calculated: ΔP_i = P_i - P_base; According to the Delta P_i and industry characteristics of the called party, setting a scaling factor to scale to obtain an outbound Score delta_score_i=round (Delta P_i×S); Wherein, the N_total_success represents the historical total number of listens; n_total_calls represents the historical total number of calls; n_ai represents the total number of occurrences of outbound classification A_i, A_i ε (A1-A5); N_success_after_ai is the number of times the next call is successful after action A_i occurs; m represents a Bayesian smoothing parameter; S represents a fractional scaling factor; round () represents a rounding calculation.
  7. 7. The method of claim 1, wherein the compensation weight acquisition process comprises: the original weighted call completing rate RawRate i of the outbound number i is obtained, ; Wherein Ni represents the historical total dialing times of the number i, isConnected { i, k } represents the result of the kth dialing with the number i, is turned on as 1, is not turned on as 0, decay (tk) represents a time Decay function, decay (tk) = e-lambda (T_ { current } -tk) }, wherein lambda is a Decay factor, T_ { current } -tk is the number of days from the current time; a global average on-coming rate GlobalRate is obtained, ; Wherein M represents the total number; the confidence weight ConfidenceWeight i for the number i is calculated, ; Wherein EffectiveCount i is the number i of effective dialing times, C is a preset constant, represents a confidence threshold, and C is set according to the number i of effective dialing times; Obtaining the external calling number BaseWeight of the line where the number i is located according to the original weighted call completing rate RawRate i , the global average call completing rate GlobalRate and the confidence weight ConfidenceWeight i of the number i i And calculating the average value of all numbers of the line according to BaseWeight i to be used as compensation weight.
  8. 8. An electronic device comprising one or more processors and a memory having stored therein computer readable instructions which, when executed by the one or more processors, implement the steps of the method of any of claims 1 to 7.
  9. 9. A readable storage medium having stored therein computer readable instructions which, when executed by one or more processors, cause the one or more processors to implement the steps of the method of any of claims 1 to 7.

Description

Telephone number selection outbound method, electronic device and storage medium Technical Field The invention belongs to the technical field of intelligent outbound call in the technical field of communication, and particularly relates to a telephone number selection outbound call method, electronic equipment and a storage medium. Background The existing outbound system generally adopts a polling number selection strategy, a random allocation strategy or a number selection strategy based on simple rules (such as generic matching), and has the problems that the resource allocation is stiff, the high-value number is excessively used due to fixed priority, and a blocking mechanism is easy to trigger. Lack of adjustment, the real-time optimization of the number selection strategy without combination of user behavior feedback. Disclosure of Invention In order to solve the technical problems, the application provides a telephone number selection outbound method, electronic equipment and a storage medium. The method and the technical scheme include that the method for calling out the telephone number selection comprises the following steps: Collecting historical outbound records of all outbound numbers in a target time range, and obtaining historical behavior characteristics of each outbound number; acquiring compensation weights of the outbound numbers of all lines according to the historical behavior characteristics and the lines where the outbound numbers are located; according to the compensation weight and the preset score, obtaining the basic score of each number; Obtaining behavior scores of the numbers according to the basic scores of the numbers and the last outbound result; And after each outbound call is finished, updating the behavior scores in real time according to the outbound call result. Further, the basic score obtaining process includes taking a constant value as a preset score, and taking the compensation weight multiplied by a preset score result as a basic score. Further, the behavior scoring results include a base score + an outbound score, the outbound score being obtained based on an outbound result classification. Further, the outbound score obtaining process includes classifying outbound results into five types according to outbound feedback, wherein the five types include effective connection of A1 (the call duration is more than or equal to T), short connection of A2 (the call duration is less than or equal to T0), refusal of A3 users, answering of A4 unmanned (ringing timeout) and interception of A5 users, wherein T is set according to the called industry; Based on the outbound result classification, acquiring the outbound score of the corresponding type, and taking the outbound score as the outbound score. Further, the procedure for acquiring the outbound call score comprises, Acquiring a reference answering rate P_base according to the historical outbound records of the numbers, That is to say, p_base=n\u total_success N_total_calls; Calculating the condition answering rate P_i of the effective connection of the next outbound after each outbound classification in the historical outbound record occurs, and adopting Bayesian smoothing in the process of calculating the condition answering rate P_i to eliminate small probability errors, namely P_i= (N_success_after_ai+m) P_base)/(N_ai+m); From p_base and p_i, the probability difference Δp_i is calculated: ΔP_i = P_i - P_base; According to the Delta P_i and industry characteristics of the called party, setting a scaling factor to scale to obtain an outbound Score delta_score_i=round (Delta P_i×S); Wherein, the N_total_success represents the historical total number of listens; n_total_calls represents the historical total number of calls; n_ai represents the total number of occurrences of outbound classification A_i, A_i ε (A1-A5); N_success_after_ai: the number of times the next call was successful after action A_i occurred. M represents a Bayesian smoothing parameter; S represents a fractional scaling factor; round () represents a rounding calculation. Further, the compensation weight obtaining process includes: the original weighted call completing rate RawRate i of the outbound number i is obtained, Wherein Ni represents the historical total dialing times of the number i, isConnected { i, k } represents the result of the kth dialing with the number i, is turned on as 1, is not turned on as 0, decay (tk) represents a time Decay function, decay (tk) = e-lambda (T_ { current } -tk) }, wherein lambda is a Decay factor, T_ { current } -tk is the number of days from the current time; a global average on-coming rate GlobalRate is obtained, Wherein M represents the total number; the confidence weight ConfidenceWeight i for the number i is calculated, Wherein EffectiveCount i is the number i of effective dialing times, C is a preset constant, represents a confidence threshold, and C is set according to the number i of effective dialing times;