CN-119694347-B - Outbound quality monitoring and optimizing system based on big data

CN119694347BCN 119694347 BCN119694347 BCN 119694347BCN-119694347-B

Abstract

The invention discloses an outbound quality monitoring and optimizing system based on big data, and belongs to the technical field of outbound quality optimization. The voice data array comprises an outbound direct address voice acquisition unit, a voice grouping unit, an outbound quality evaluation unit, an optimized voice library generation unit and an outbound optimization unit, wherein the outbound direct address voice acquisition unit acquires all outbound voices in a time period from the current time to a previous time node to obtain a voice data array The voice grouping unit is connected to the voice analysis module and is used for voice data array Each element in the voice packet is grouped, and the outbound quality assessment unit is communicated with the voice grouping unit to obtain an assessment intersection And count Number of (3) Then calculates the outbound quality ratio The optimized voice library generating unit pair 、 And Performing intersection calculation to obtain intersection The outbound quality monitoring and optimizing system based on big data can monitor the outbound quality in real time, optimize the outbound behavior according to the outbound quality and improve the outbound quality.

Inventors

CHEN ZHIJIAN
Shi Bocang
Li xirui
MAO HONGYING

Assignees

泰安泰盈信息科技有限公司

Dates

Publication Date: 20260505
Application Date: 20241227

Claims (7)

1. The outbound quality monitoring and optimizing system based on big data is characterized by comprising: an outer direct address sound acquisition unit for acquiring all outbound voices in a time period from the current time node to a previous time node to obtain a voice data array , ; The voice grouping unit is connected with the voice analysis module and acquires a voice data array Then to the voice data array Each element in the group is grouped, and the grouping process is as follows: dividing voice data array according to call duration When (when) The conversation duration of a certain element in the network is lower than that of the network When the element is incorporated into the invalid communication data array When (1) The conversation time of a certain element in the system is longer than that of the system And (2) and When the element is incorporated into the active communication data array When (1) The conversation time of a certain element in the system is longer than that of the system When the element is incorporated into the inefficient communication data array ; Dividing a speech data array according to statistical interaction times When (when) Talking to a certain element in the network the interaction times are lower than When the element is incorporated into the invalid communication data array When (1) The number of conversation interactions of a certain element in the system is larger than And (2) and When the element is incorporated into the active communication data array When (1) The number of conversation interactions of a certain element in the system is larger than When the element is incorporated into the inefficient communication data array ; Dividing arrays of voice data based on whether or not conversion is successful When (when) After a certain element in the data array is successfully converted, the element is incorporated into the data array which is successfully converted When (1) After conversion failure of an element in the data array, the element is incorporated into the conversion failure data array ; An outbound quality assessment unit in communication with the voice packet unit for obtaining an assessment intersection , And then count Number of (3) Then calculates the outbound quality ratio When (when) Outputting an optimization instruction when the optimization time is greater than the set threshold value and is greater than the set optimization period duration; (1); Wherein, the For arrays of speech data A total amount; An optimized voice library generating unit which communicates with the voice grouping unit and acquires 、 And And is matched with 、 And Performing intersection calculation to obtain intersection , ; The outbound optimization unit is in communication connection with the outbound quality assessment unit and the optimized voice library generation unit, and when the outbound optimization unit obtains an optimization instruction, the outbound optimization unit outputs the optimized instruction to the voice library generation unit Sending the outbound optimization data to an outbound optimization unit to obtain outbound optimization data.
2. The outbound quality monitoring and optimization system based on big data according to claim 1, wherein the outbound quality monitoring and optimization system is characterized in that The duration is 5-10 , The duration is 3 to 5 minutes, The times are 2 to 3 times, The time length is 8-12 times, one complete call interaction comprises one outbound output and a voice section which is answered, and the success is converted into the successful success of the outbound, transferred to special service personnel or affirmed by the outbound party.
3. The outbound quality monitoring and optimizing system based on big data according to claim 1, wherein the outbound direct address tone acquisition unit acquires intersections Extracting the called part voice section, the leading-out calling part voice section and the responding calling part voice section, wherein the leading-out calling part voice section is the outer direct address voice section which occurs before the called part voice section and is closest to the called part voice section, the responding calling part voice section is the outer direct address voice section which occurs after the called part voice section and is closest to the called part voice section, the related words are extracted from the called part voice section, the related words are generated by an outbound reply dictionary library, one related word or a combination of a plurality of related words is obtained for each called part voice section with the related words as index items, each index item is connected with one responding calling part voice section, and intersection is completed Grouping to obtain groups Wherein ; For arrays of speech data Is the first of (2) Of individual elements Responding to the calling party voice segments corresponding to the index items; each index item is connected with a voice section of the calling party to finish intersection Grouping to obtain groups Wherein ; For arrays of speech data Is the first of (2) Of individual elements And leading out the voice segments of the calling party corresponding to the index items.
4. The outbound quality monitoring and optimizing system based on big data according to claim 3, wherein the outbound optimizing unit comprises a high-frequency word extracting unit, a test set generating module and an algorithm module, and the working process of the high-frequency word extracting unit is as follows: the high-frequency word extraction unit acquires a group And collecting the voice segments of the responding calling party corresponding to the same index item to obtain N groups of collection sets, wherein each group of collection sets Then to Extracting high-frequency words, wherein For a certain collection The first of (3) The group responds to the caller's speech segments, , To collect together The total data set quantity is obtained After that, to Extracting single high-frequency word defined as a phrase in a certain phrase The number of occurrences in the pattern reaches a set threshold value, and the threshold value is set The calculation is as follows: (2); Wherein, the Is that Is used for the number of the total word groups, Is that The number of times a phrase appears, and the phrase is Number of occurrences of (a) And (2) and When the phrase is in When the occurrence number of the phrase is more than 4, defaulting the phrase to be a single high-frequency word; Obtaining the sequence of single high-frequency words according to the calculation , Then, the array is Performing secondary high-frequency word calculation to obtain secondary high-frequency word sequence ; The method comprises the steps of calculating and calculating the last secondary high-frequency word in a secondary high-frequency word sequence, wherein the calculation process of the secondary high-frequency word is as follows: Acquiring a series of numbers Any one of the single high-frequency words and counting the single high-frequency words In a plurality of rows The number of occurrences of (a) When (when) When it will Feeding in ; (3); Wherein, the Is in a plurality of columns The total amount of elements in (a); Is in a plurality of columns Any element; the high-frequency word extraction unit acquires a group And is opposite to Extracting high-frequency words, and grouping The extraction process is consistent, and a secondary high-frequency word sequence is output After the operation is completed, the high-frequency word extraction unit outputs ; The test set generation module obtains And And calculate the intersection And union set , ; Obtaining the first-level importance phrase And second-level importance phrase The first-level importance phrase Is an intersection of Corresponding phrase, two-level importance phrase From union Intersection set The obtained phrase; the calculation module obtains 、 Sum operation array ; The algorithm array ; Then, the algorithm sequence is traversed And from the calculation sequence The acquisition of comprises Is a series of (a) of (b) Statistics of Each set of data in (1) contains Obtaining the corresponding elements of the group of data Element combination, finish All of (3) Element combination operation and will be the same Combining and merging elements, and finally according to the same The element combination quantity is sequentially arranged from high to low to obtain an optimized phrase, and the optimized phrase comprises Corresponding phrase, and 1 group ordered at the front Combinations or groups of elements Finally, generating outbound data containing optimized phrase through manual editing or AI automatically.
5. The system for monitoring and optimizing outbound quality based on big data as claimed in claim 4, wherein the AI is configured to set an outbound data word number when the AI automatically generates outbound data, the outbound data word number being an average value of voice segments of a responding caller corresponding to each index item.
6. The outbound quality monitoring and optimizing system based on big data of claim 1, wherein the outbound quality assessment unit is connected with an assessment output module, and the assessment output module generates day, month, quarter and year assessment reports.
7. The big data based outbound quality monitoring and optimization system of claim 1, wherein the outbound quality assessment unit further comprises an optimal time period outbound assessment module, the optimal time period outbound assessment module obtaining the outbound information in each time period 、 And The number of intersection elements and the number of union elements, and then, obtaining the outbound effective rate 。

Description

Outbound quality monitoring and optimizing system based on big data Technical Field The invention particularly relates to an outbound quality monitoring and optimizing system based on big data, and belongs to the technical field of outbound quality optimization. Background In the current intelligent dialogue system based on a Large Language Model (LLM), one implementation mode is to perform preferential selection on an alternative dialogue by combining up-down questions after the alternative dialogue is returned to the LLM by searching a vector library, the current preferential judgment standard is to interfere with the dialogue selection capability of the LLM by expert experience, but each dialogue is judged by expert experience along with the improvement of service complexity, the efficiency of manually interfering with the dialogue selection capability of the LLM is very low, even client communication will cannot be objectively reflected, how to optimize the dialogue selection of the intelligent dialogue system is performed, the intelligent dialogue efficiency is the problem to be solved at present, for this purpose, chinese patent publication No. CN119088918A discloses a dialogue optimization method, a device, electronic equipment and a storage medium. Disclosure of Invention In order to solve the problems, the invention provides an outbound quality monitoring and optimizing system based on big data, which can monitor outbound quality in real time, optimize outbound behaviors according to the outbound quality and improve the outbound quality. The invention relates to a big data-based outbound quality monitoring and optimizing system, which comprises: an outer direct address sound acquisition unit for acquiring all outbound voices in a time period from the current time node to a previous time node to obtain a voice data array ,The previous time node is a set time node or a time node for completing the optimization evaluation last time; the voice grouping unit is connected with the voice analysis module and acquires a voice data array Then to the voice data arrayEach element in the group is grouped, and the grouping process is as follows: dividing voice data array according to call duration When (when)The conversation duration of a certain element in the network is lower than that of the networkWhen the element is incorporated into the invalid communication data arrayWhen (1)The conversation time of a certain element in the system is longer than that of the systemAnd (2) andWhen the element is incorporated into the active communication data arrayWhen (1)The conversation time of a certain element in the system is longer than that of the systemWhen the element is incorporated into the inefficient communication data array; Dividing a speech data array according to statistical interaction timesWhen (when)Talking to a certain element in the network the interaction times are lower thanWhen the element is incorporated into the invalid communication data arrayWhen (1)The number of conversation interactions of a certain element in the system is larger thanAnd (2) andWhen the element is incorporated into the active communication data arrayWhen (1)The number of conversation interactions of a certain element in the system is larger thanWhen the element is incorporated into the inefficient communication data array; Dividing arrays of voice data based on whether or not conversion is successfulWhen (when)After a certain element in the data array is successfully converted, the element is incorporated into the data array which is successfully convertedWhen (1)After conversion failure of an element in the data array, the element is incorporated into the conversion failure data array; An outbound quality assessment unit in communication with the voice packet unit for obtaining an assessment intersection,=And then countNumber of (3)Then calculates the outbound quality ratioWhen (when)Outputting an optimization instruction when the optimization time is greater than the set threshold value and is greater than the set optimization period duration; (1) Wherein, the For arrays of speech dataA total amount; An optimized voice library generating unit which communicates with the voice grouping unit and acquires 、AndAnd is matched with、AndPerforming intersection calculation to obtain intersection,; The outbound optimization unit is in communication connection with the outbound quality assessment unit and the optimized voice library generation unit, and when the outbound optimization unit obtains an optimization instruction, the outbound optimization unit outputs the optimized instruction to the voice library generation unitSending the outbound optimization data to an outbound optimization unit to obtain outbound optimization data. In operation, the exo direct address voice acquisition unit acquires all exotic voices in a set time period, sends all exotic direct address voices to the voice grouping unit, groups all exotic direct addr