A Big Data-Based Outbound Call Quality Monitoring and Optimization System

Abstract

This invention discloses a big data-based outbound call quality monitoring and optimization system, belonging to the field of outbound call quality optimization technology. It includes an outbound voice acquisition unit, a voice grouping unit, an outbound call quality assessment unit, an optimized voice library generation unit, and an outbound call optimization unit. The outbound voice acquisition unit acquires all outbound voices within a time period from the current point to a previous time node, obtaining a voice data array. The voice grouping unit connects to a voice analysis module to group each element in the voice data array. The outbound call quality assessment unit communicates with the voice grouping unit; obtains the assessment intersection, counts the number of evaluations, and then calculates the outbound call quality ratio. The optimized voice library generation unit performs intersection calculations on the evaluations and the grouping units to obtain the intersection. This big data-based outbound call quality monitoring and optimization system can monitor outbound call quality in real time and optimize outbound call behavior based on the outbound call quality, thereby improving outbound call quality.

Description

Technical Field

[0001] This invention specifically relates to an outbound call quality monitoring and optimization system based on big data, belonging to the field of outbound call quality optimization technology. Background Technology

[0002] In current intelligent dialogue systems based on Large Language Models (LLM), one approach involves retrieving a vector library and providing the LLM with candidate dialogues. The LLM then selects the optimal dialogue based on context and context. Currently, the evaluation standard relies on expert experience to influence the LLM's dialogue selection. However, with increasing business complexity, evaluating each dialogue based on expert experience and manually intervening in the LLM's dialogue selection becomes extremely inefficient and may even fail to objectively reflect the customer's communication intentions. Therefore, optimizing the dialogue selection process and improving the efficiency of intelligent dialogue systems are urgent problems that need to be addressed. Chinese Patent Publication No. CN119088918A discloses a method, apparatus, electronic device, and storage medium for optimizing dialogue. This method optimizes corpus data using a preset sorting method, that is, adding new corpus data or deleting corpus data with small weight values ​​according to the weight, thereby continuously optimizing the corpus data according to different scenarios, and generating a dialogue selection model that is more suitable for different scenarios. After receiving a dialogue request from a client terminal, it can more accurately respond with the target dialogue that matches the scenario, improving the accuracy of dialogue selection. However, this optimization method cannot monitor and provide feedback on outbound call quality, nor can it optimize outbound call quality in real time. Summary of the Invention

[0003] To address the aforementioned issues, this invention proposes a big data-based outbound call quality monitoring and optimization system, which can monitor outbound call quality in real time and optimize outbound call behavior based on the quality, thereby improving outbound call quality.

[0004] The outbound call quality monitoring and optimization system based on big data of the present invention includes:

[0005] The outbound call voice acquisition unit acquires all outbound call voices within a time period from the current point to a previous time node, thus obtaining a voice data array. , A previous time point was a specific time point or the time point when the optimization evaluation was last completed.

[0006] A voice grouping unit, which is connected to a voice analysis module, acquires a voice data array. Next, the voice data array Each element within the group is grouped as follows:

[0007] Voice data arrays are divided according to call duration. ,when The call duration of a certain element within is less than When this happens, the element is included in the invalid communication data array. ;when The duration of a call within a certain element is greater than ,and At that time, this element will be included in the effective communication data array. ;when The duration of a call within a certain element is greater than At that time, this element will be included in the inefficient communication data array. ;

[0008] The voice data array is divided according to the number of statistical interactions. ,when The number of times a certain element within the call is less than When this happens, the element is included in the invalid communication data array. ;when The number of times a certain element within the call is greater than ,and At that time, this element will be included in the effective communication data array. ;when The number of times a certain element within the call is greater than At that time, this element will be included in the inefficient communication data array. ;

[0009] The speech data array is divided according to whether the conversion was successful. ,when Once an element within the array is successfully converted, that element is added to the successfully converted data array. ;when If a transformation of an element fails, that element is added to the transformation failure data array. ;

[0010] Outbound call quality assessment unit, which communicates with voice packet unit; obtains assessment intersection. , = And statistics quantity Next, calculate the outbound call quality ratio. ,when When the value exceeds the set threshold and the set optimization period duration is exceeded, an optimization command is output.

[0011] (1)

[0012] in, For voice data array Total amount;

[0013] The optimized speech database generation unit communicates with the speech grouping unit; the optimized speech database generation unit acquires... , and ; and on , and Perform intersection calculation to obtain the intersection. , ;

[0014] Outbound call optimization unit, which is communicatively connected to outbound call quality assessment unit and optimized voice database generation unit; when outbound call optimization unit receives optimization instructions, it will... The data is sent to the outbound call optimization unit to obtain outbound call optimization data.

[0015] During operation, the outbound voice acquisition unit acquires all outbound voice calls within a set time period and sends them to the voice grouping unit. The voice grouping unit then groups the outbound voice calls, completing the grouping process. Finally, the outbound voice groups are sent to the outbound call quality evaluation unit, which evaluates the grouped outbound voice calls to obtain the outbound call quality ratio. And based on the outbound call quality ratio The system determines whether to output an optimization command based on the set optimization cycle duration. After the outbound call optimization unit receives the optimization command, it obtains the database generated by the optimized voice library generation unit and optimizes it to obtain outbound call optimization data.

[0016] Furthermore, the aforementioned The duration is 5-10 seconds. The duration is 3-5 minutes. The number of times is 2-3. The duration is 8-12 times. A complete call interaction includes an outbound call output and a voice segment receiving a response; the success is converted into a successful transaction of the outbound call, being transferred to a dedicated business person, or receiving a positive response from the called party.

[0017] Furthermore, the outbound call voice acquisition unit obtains the intersection. Next, the called party's voice segment, the calling party's voice segment, and the responding calling party's voice segment are extracted. The calling party's voice segment is the outbound voice segment that occurs before and is closest to the called party's voice segment. The responding calling party's voice segment is the outbound voice segment that occurs after and is closest to the called party's voice segment. Then, related words are extracted from the called party's voice segments. These related words are generated from an outbound call response dictionary. For each called party's voice segment containing related words, one or more related words are obtained as an index item. Each index item is connected to a responding calling party's voice segment. The intersection is then processed. Grouping, resulting in groups ,in ∈ ; For voice data array The Each element Each index entry corresponds to a voice segment from the caller.

[0018] Each index entry is linked to a segment of the caller's voice; this completes the intersection. Grouping, resulting in groups ,in ∈ ; For voice data array The Each element Each index entry corresponds to a voice segment from the calling party.

[0019] Furthermore, the outbound call optimization unit includes a high-frequency word extraction unit, a test set generation module, and a calculation module; the high-frequency word extraction unit works as follows:

[0020] The high-frequency word extraction unit obtains groups. Then, the voice segments of the responding caller corresponding to the same index item are grouped together to obtain N groups, each group being... Then on High-frequency word extraction is performed. For a certain set The first in The group responds to the caller's voice segment. , For collection Total number of data sets; obtain Afterwards, Perform single high-frequency word extraction, whereby a single high-frequency word is defined as a phrase appearing in the text. The number of times it appears in the text reaches a set threshold. The calculation is as follows:

[0021] (2)

[0022] in, for The number of times a certain phrase appears in the text, and the phrase appears in the context of the text. Number of times it appears in ,and When the number of occurrences of a phrase is greater than 4, the phrase is considered a single high-frequency word by default. for Total number of phrases;

[0023] Based on the above calculations, the sequence of high-frequency words for a single entry is obtained. , Next, for the sequence Perform secondary high-frequency word calculation to obtain the secondary high-frequency word sequence. ; This is the last secondary high-frequency word in the secondary high-frequency word sequence; the calculation process for the secondary high-frequency word is as follows:

[0024] Get the sequence Find any single high-frequency word in the text and count its frequency. In the sequence Number of times it appears ,when Will Send in ;

[0025] (3)

[0026] in, For sequence The total number of elements in it; For sequence Any element in;

[0027] The high-frequency word extraction unit obtains groups. and to High-frequency word extraction, extraction process and grouping The extraction process is consistent, and the output is a secondary high-frequency word sequence. After completing the calculation, the high-frequency word extraction unit outputs... and ;

[0028] The test set generation module obtains and And calculate the intersection. Union A= ; = ; obtained the first-level importance phrase and second-level importance phrases The first-level importance phrases The phrase corresponding to intersection A, the phrase with second-order importance. The phrase obtained by the union of set B and the intersection of set A;

[0029] The calculation module obtains , And calculus of sequences ;

[0030] The calculated sequence =( )- ;

[0031] Next, iterate through and calculate the sequence. And from the calculation sequence Get the contents sequence ;statistics Each set of data contains The elements in the set are used to obtain the corresponding data. Combine elements to complete. All Element combination operation, and combine identical elements. Element combination and merging, finally based on the same The optimized word groups are obtained by arranging the number of element combinations from high to low. The optimized word groups include... The corresponding phrases, and the first group in the order. Element combinations or multiple groups The elements are combined to form the corresponding word groups; finally, outbound call data containing optimized word groups is generated either manually or automatically by AI.

[0032] Furthermore, when the AI ​​automatically generates outbound call data, the number of characters in the outbound call data needs to be set. The number of characters in the outbound call data is the average value of the voice segment of the responding caller corresponding to each index item.

[0033] Furthermore, the outbound call quality assessment unit is connected to an assessment output module, which generates daily, monthly, quarterly, and annual assessment reports; the outbound call quality assessment unit calculates the outbound call quality ratio for each stage. And by evaluating the outbound call quality ratio output by the output module Simultaneously, it outputs the composition of the outbound call quality ratio, i.e., the output... , , and The percentage.

[0034] Furthermore, the outbound call quality assessment unit also includes an optimal time period outbound call assessment module, which acquires outbound call quality assessment data for each time period. , and The number of elements in the intersection and the number of elements in the union are used to determine the outbound call efficiency. , ;

[0035] By statistically analyzing the effective outbound call volume over various time periods, which consists of successful conversions and effective communication, and assigning appropriate proportional factors to these volumes, the quality of outbound calls can be comprehensively evaluated.

[0036] Compared with existing technologies, the big data-based outbound call quality monitoring and optimization system of the present invention can be applied to a combination of robots and humans. After the robot makes an outbound call and connects with the called party, the robot transfers the call to a human. The human can then test the outbound call data, record the entire test outbound call, evaluate the outbound call quality, optimize the evaluation results, obtain optimized results, and implement the new optimized results in the next round of outbound call process to improve outbound call quality. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the overall structure of the outbound call quality monitoring and optimization system based on big data according to the present invention.

[0038] Figure 2 This is a schematic diagram of the workflow structure of the outbound call quality monitoring and optimization system of the present invention.

[0039] Figure 3 This is a schematic diagram of the process structure for refining and grouping all outbound voice calls according to the present invention. Detailed Implementation

[0040] like Figures 1 to 3 The big data-based outbound call quality monitoring and optimization system shown includes:

[0041] The outbound call voice acquisition unit acquires all outbound call voices within a time period from the current point to a previous time node, thus obtaining a voice data array. , A previous time point was a specific time point or the time point when the optimization evaluation was last completed.

[0042] A voice grouping unit, which is connected to a voice analysis module, acquires a voice data array. Next, the voice data array Each element within the group is grouped as follows:

[0043] Voice data arrays are divided according to call duration. ,when The call duration of a certain element within is less than When this happens, the element is included in the invalid communication data array. ;when The duration of a call within a certain element is greater than ,and At that time, this element will be included in the effective communication data array. ;when The duration of a call within a certain element is greater than At that time, this element will be included in the inefficient communication data array. ;

[0044] The voice data array is divided according to the number of statistical interactions. ,when The number of times a certain element within the call is less than When this happens, the element is included in the invalid communication data array. ;when The number of times a certain element within the call is greater than ,and At that time, this element will be included in the effective communication data array. ;when The number of times a certain element within the call is greater than At that time, this element will be included in the inefficient communication data array. ;

[0045] The speech data array is divided according to whether the conversion was successful. ,when Once an element within the array is successfully converted, that element is added to the successfully converted data array. ;when If a transformation of an element fails, that element is added to the transformation failure data array. ;

[0046] Outbound call quality assessment unit, which communicates with voice packet unit; obtains assessment intersection. , = And statistics quantity Next, calculate the outbound call quality ratio. ,when When the value exceeds the set threshold and the set optimization period duration is exceeded, an optimization command is output.

[0047] (1)

[0048] in, For voice data array Total amount;

[0049] The optimized speech database generation unit communicates with the speech grouping unit; the optimized speech database generation unit acquires... , and ; and on , and Perform intersection calculation to obtain the intersection. , ;

[0050] Outbound call optimization unit, which is communicatively connected to outbound call quality assessment unit and optimized voice database generation unit; when outbound call optimization unit receives optimization instructions, it will... The data is sent to the outbound call optimization unit to obtain outbound call optimization data.

[0051] During operation, the outbound voice acquisition unit acquires all outbound voice calls within a set time period and sends them to the voice grouping unit. The voice grouping unit then groups the outbound voice calls, completing the grouping process. Finally, the outbound voice groups are sent to the outbound call quality evaluation unit, which evaluates the grouped outbound voice calls to obtain the outbound call quality ratio. And based on the outbound call quality ratio The system determines whether to output an optimization command based on the set optimization cycle duration. After the outbound call optimization unit receives the optimization command, it obtains the database generated by the optimized voice library generation unit and optimizes it to obtain outbound call optimization data.

[0052] The The duration is 5-10 seconds. The duration is 3-5 minutes. The number of times is 2-3. The duration is 8-12 times. A complete call interaction includes an outbound call output and a voice segment receiving a response; the success is converted into a successful transaction of the outbound call, being transferred to a dedicated business person, or receiving a positive response from the called party.

[0053] The outbound call voice acquisition unit obtains the intersection. Next, the called party's voice segment, the calling party's voice segment, and the responding calling party's voice segment are extracted. The calling party's voice segment is the outbound voice segment that occurs before and is closest to the called party's voice segment. The responding calling party's voice segment is the outbound voice segment that occurs after and is closest to the called party's voice segment. Then, related words are extracted from the called party's voice segments. These related words are generated from an outbound call response dictionary. For each called party's voice segment containing related words, one or more related words are obtained as an index item. Each index item is connected to a responding calling party's voice segment. The intersection is then processed. Grouping, resulting in groups ,in ∈ ; For voice data array The Each element Each index entry corresponds to a voice segment from the caller.

[0054] Each index entry is linked to a segment of the caller's voice; this completes the intersection. Grouping, resulting in groups ,in ∈ ; For voice data array The Each element Each index entry corresponds to a voice segment from the calling party.

[0055] The outbound call optimization unit includes a high-frequency word extraction unit, a test set generation module, and a calculation module; the high-frequency word extraction unit works as follows:

[0056] The high-frequency word extraction unit obtains groups. Then, the voice segments of the responding caller corresponding to the same index item are grouped together to obtain N groups, each group being... Then on High-frequency word extraction is performed. For a certain set The first in The group responds to the caller's voice segment. , For collection Total number of data sets; obtain Afterwards, Perform single high-frequency word extraction, whereby a single high-frequency word is defined as a phrase appearing in the text. The number of times it appears in the text reaches a set threshold. The calculation is as follows:

[0057] (2)

[0058] in, for The number of times a certain phrase appears in the text, and the phrase appears in the context of the text. Number of times it appears in ,and When the number of occurrences of a phrase is greater than 4, the phrase is considered a single high-frequency word by default. for Total number of phrases;

[0059] Based on the above calculations, the sequence of high-frequency words for a single entry is obtained. , Next, for the sequence Perform secondary high-frequency word calculation to obtain the secondary high-frequency word sequence. ; This is the last secondary high-frequency word in the secondary high-frequency word sequence; the calculation process for the secondary high-frequency word is as follows:

[0060] Get the sequence Find any single high-frequency word in the text and count its frequency. In the sequence Number of times it appears ,when Will Send in ;

[0061] (3)

[0062] in, For sequence The total number of elements in it; For sequence Any element in;

[0063] The high-frequency word extraction unit obtains groups. and to High-frequency word extraction, extraction process and grouping The extraction process is consistent, and the output is a secondary high-frequency word sequence. After completing the calculation, the high-frequency word extraction unit outputs... and ;

[0064] The test set generation module obtains and And calculate the intersection. Union A= ; = ; obtained the first-level importance phrase and second-level importance phrases The first-level importance phrases The phrase corresponding to intersection A, the phrase with second-order importance. The phrase obtained by the union of set B and the intersection of set A;

[0065] The calculation module obtains , And calculus of sequences ;

[0066] The calculated sequence =( )- ;

[0067] Next, iterate through and calculate the sequence. And from the calculation sequence Get the contents sequence ;statistics Each set of data contains The elements in the set are used to obtain the corresponding data. Combine elements to complete. All Element combination operation, and combine identical elements. Element combination and merging, finally based on the same The optimized word groups are obtained by arranging the number of element combinations from high to low. The optimized word groups include... The corresponding phrases, and the first group in the order. Element combinations or multiple groups The elements are combined to form the corresponding word groups; finally, outbound call data containing optimized word groups is generated either manually or automatically by AI.

[0068] When the AI ​​automatically generates outbound call data, the number of characters in the outbound call data needs to be set. The number of characters in the outbound call data is the average value of the voice segment of the responding caller corresponding to each index item.

[0069] The outbound call quality assessment unit is connected to an assessment output module, which generates daily, monthly, quarterly, and annual assessment reports; the outbound call quality assessment unit calculates the outbound call quality ratio for each stage. And by evaluating the outbound call quality ratio output by the output module Simultaneously, it outputs the composition of the outbound call quality ratio, i.e., the output... , , and The percentage.

[0070] The outbound call quality assessment unit also includes an optimal time period outbound call assessment module, which obtains outbound call quality assessment data for each time period. , and The number of elements in the intersection and the number of elements in the union are used to determine the outbound call efficiency. , ;

[0071] By statistically analyzing the effective outbound call volume over various time periods, which consists of successful conversions and effective communication, and assigning appropriate proportional factors to these volumes, the quality of outbound calls can be comprehensively evaluated.

[0072] The above embodiments are merely preferred embodiments of the present invention. Therefore, all equivalent changes or modifications made to the structure, features and principles described in the claims of the present invention are included within the scope of the present invention.

Claims

1. A big data-based outbound call quality monitoring and optimization system, characterized in that: include: The outbound call voice acquisition unit acquires all outbound call voices within a time period from the current point to a previous time node, thus obtaining a voice data array. , ; A voice grouping unit, which is connected to a voice analysis module, acquires a voice data array. Next, the voice data array Each element within the group is grouped as follows: Voice data arrays are divided according to call duration. ,when The call duration of a certain element within is less than When this happens, the element is included in the invalid communication data array. ;when The duration of a call within a certain element is greater than ,and At that time, this element will be included in the effective communication data array. ;when The duration of a call within a certain element is greater than At that time, this element will be included in the inefficient communication data array. ; The voice data array is divided according to the number of statistical interactions. ,when The number of times a certain element within the call is less than When this happens, the element is included in the invalid communication data array. ;when The number of times a certain element within the call is greater than ,and At that time, this element will be included in the effective communication data array. ;when The number of times a certain element within the call is greater than At that time, this element will be included in the inefficient communication data array. ; The speech data array is divided according to whether the conversion was successful. ,when Once an element within the array is successfully converted, that element is added to the successfully converted data array. ;when If a transformation of an element fails, that element is added to the transformation failure data array. ; Outbound call quality assessment unit, which communicates with voice packet unit; obtains assessment intersection. , And statistics quantity Next, calculate the outbound call quality ratio. ,when When the value exceeds the set threshold and the set optimization period duration is exceeded, an optimization command is output. （1）； in, For voice data array Total amount; The optimized speech library generation unit communicates with the speech grouping unit; the optimized speech library generation unit acquires... , and ; and on , and Perform intersection calculation to obtain the intersection. , ; Outbound call optimization unit, which is communicatively connected to outbound call quality assessment unit and optimized voice database generation unit; when outbound call optimization unit receives optimization instructions, it will... The data is sent to the outbound call optimization unit to obtain outbound call optimization data.

2. The outbound call quality monitoring and optimization system based on big data according to claim 1, characterized in that: The Duration 5-10 minutes , The duration is 3-5 minutes. The number of times is 2-3. The duration is 8-12 times. A complete call interaction includes an outbound call output and a voice segment receiving a response; the success is converted into a successful transaction of the outbound call, being transferred to a dedicated business person, or receiving a positive response from the called party.

3. The outbound call quality monitoring and optimization system based on big data according to claim 1, characterized in that: The outbound call voice acquisition unit obtains the intersection. Next, the called party's voice segment, the calling party's voice segment, and the responding calling party's voice segment are extracted. The calling party's voice segment is the outbound voice segment that occurs before and is closest to the called party's voice segment. The responding calling party's voice segment is the outbound voice segment that occurs after and is closest to the called party's voice segment. Then, related words are extracted from the called party's voice segments. These related words are generated from an outbound call response dictionary. For each called party's voice segment containing related words, one or more related words are obtained as an index item. Each index item is connected to a responding calling party's voice segment. The intersection is then processed. Grouping, resulting in groups ,in ; For voice data array The Each element Each index entry corresponds to a voice segment from the caller. Each index entry is linked to a segment of the caller's voice; this completes the intersection. Grouping, resulting in groups ,in ; For voice data array The Each element Each index entry corresponds to a voice segment from the calling party.

4. The outbound call quality monitoring and optimization system based on big data according to claim 3, characterized in that: The outbound call optimization unit includes a high-frequency word extraction unit, a test set generation module, and a calculation module; the high-frequency word extraction unit works as follows: The high-frequency word extraction unit obtains groups. Then, the voice segments of the responding caller corresponding to the same index item are grouped together to obtain N groups, each group being... Then on High-frequency word extraction is performed. For a certain set The first in The group responds to the caller's voice segment. , For collection Total number of data sets; obtain Afterwards, Perform single high-frequency word extraction, whereby a single high-frequency word is defined as a phrase appearing in the text. The number of times it appears in the text reaches a set threshold. The calculation is as follows: （2）； in, for Total number of phrases for The number of times a certain phrase appears in the text, and the phrase appears in the context of the text. Number of times it appears in ,and When the phrase is in When a phrase appears more than 4 times, it is considered a single high-frequency word by default. Based on the above calculations, the sequence of high-frequency words for a single entry is obtained. , Next, for the sequence Perform secondary high-frequency word calculation to obtain the secondary high-frequency word sequence. ; This is the last secondary high-frequency word in the secondary high-frequency word sequence; the calculation process for the secondary high-frequency word is as follows: Get the sequence Find any single high-frequency word and count its frequency. In the sequence Number of times it appears ,when At that time, Send in ; （3）； in, For sequence The total number of elements in it; For sequence Any element in; The high-frequency word extraction unit obtains groups. and to High-frequency word extraction, extraction process and grouping The extraction process is consistent, and the output is a secondary high-frequency word sequence. After completing the calculation, the high-frequency word extraction unit outputs... ; The test set generation module obtains and And calculate the intersection. Union , ; ; obtained the first-level importance phrase and second-level importance phrases The first-level importance phrases For intersection The corresponding phrases, second-level importance phrases By union -Intersection The resulting phrase; The calculation module obtains , And calculus of sequences ; The calculated sequence ; Next, iterate through and calculate the sequence. And from the calculation sequence Get the contents sequence ;statistics Each set of data contains The elements in the set are used to obtain the corresponding data. Combine elements to complete. All Element combination operation, and combine identical elements. Element combination and merging, finally based on the same The optimized word groups are obtained by arranging the number of element combinations from high to low. The optimized word groups include... The corresponding phrases, and the first group in the order. Element combinations or multiple groups The elements are combined to form the corresponding word groups; finally, outbound call data containing optimized word groups is generated either manually or automatically by AI.

5. The outbound call quality monitoring and optimization system based on big data according to claim 4, characterized in that: When the AI ​​automatically generates outbound call data, the number of characters in the outbound call data needs to be set. The number of characters in the outbound call data is the average value of the response caller's voice segment corresponding to each index item.

6. The outbound call quality monitoring and optimization system based on big data according to claim 1, characterized in that: The outbound call quality assessment unit is connected to an assessment output module, which generates daily, monthly, quarterly, and annual assessment reports.

7. The outbound call quality monitoring and optimization system based on big data according to claim 1, characterized in that: The outbound call quality assessment unit also includes an optimal time period outbound call assessment module, which obtains outbound call quality assessment data for each time period. , and The number of elements in the intersection and the number of elements in the union are calculated; then, the outbound call efficiency is obtained. .

Images