WO2015192668A1 - 语音业务的评价处理方法及装置 - Google Patents

语音业务的评价处理方法及装置 Download PDF

Info

Publication number
WO2015192668A1
WO2015192668A1 PCT/CN2015/072609 CN2015072609W WO2015192668A1 WO 2015192668 A1 WO2015192668 A1 WO 2015192668A1 CN 2015072609 W CN2015072609 W CN 2015072609W WO 2015192668 A1 WO2015192668 A1 WO 2015192668A1
Authority
WO
WIPO (PCT)
Prior art keywords
voice service
data
attribute data
service related
long tail
Prior art date
Application number
PCT/CN2015/072609
Other languages
English (en)
French (fr)
Inventor
刘义俊
潘璐伽
田光见
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP15807778.4A priority Critical patent/EP3002692A4/en
Priority to US14/971,103 priority patent/US9813549B2/en
Publication of WO2015192668A1 publication Critical patent/WO2015192668A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M3/00—Automatic or semi-automatic exchanges
    • H04M3/22—Arrangements for supervision, monitoring or testing
    • H04M3/2227—Quality of service monitoring
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor

Definitions

  • the present invention relates to communication technologies, and in particular, to a method and an apparatus for evaluating and processing a voice service.
  • the basic voice services still occupy the majority of the proportion.
  • the method mainly used for the evaluation of the voice service includes: if the interval between two consecutive communication is extremely short, the first communication is considered to be a communication of poor quality. That is, the communication quality is judged only based on the time interval of two consecutive communications.
  • the embodiment of the invention provides a method and a device for evaluating and processing a voice service, which are used to solve the problem that the voice service evaluation has limitations in the prior art.
  • a first aspect of the embodiments of the present invention provides a method for evaluating and processing a voice service, including:
  • the voice service evaluation model obtained according to the voice service related record data is used to evaluate the voice service related record data to be evaluated before include:
  • the voice service related record data includes: voice service related attribute data, correspondingly, according to the The voice service related record data, and the obtaining the voice service evaluation model specifically includes:
  • the acquiring the voice service evaluation model according to the voice service related attribute data includes:
  • Clustering algorithm is used to cluster the attribute data of the voice service to obtain a preset number of attribute data categories
  • the machine learning method is used to train the superior and bad ranking results to obtain the voice business evaluation model.
  • the classification metric value corresponding to each category includes: an average value and a standard deviation corresponding to each category .
  • the foregoing according to the classification metric value corresponding to each category, sorting each category , get good and bad sort results, including:
  • the classifications with the same average value are sorted according to the corresponding standard deviation, and the second ranking result is obtained;
  • the acquiring the voice according to the voice service related attribute data Before the business evaluation model also includes:
  • the voice service related attribute data includes two voice attribute related attribute data whose attribute value distribution has relevance, any one of the two voice service related attribute data is deleted.
  • the acquiring the voice according to the voice service related attribute data Before the business evaluation model also includes:
  • the voice service related attribute data includes voice service related attribute data having a long tail effect
  • the long tail data in the voice service related attribute data having the long tail effect is eliminated.
  • the determining whether each of the voice service related attribute data has a long tail effect comprises:
  • the eliminating the long tail data in the voice service related attribute data with the long tail effect including:
  • the long tail data of the voice service related attribute data having the long tail effect is eliminated according to the long tail data threshold.
  • the voice service related attribute data includes any one of the following or any combination thereof: ringing time, answering time, allocating time, and authenticating Request time, encryption request time, service request reception time, hang up time, end communication time, release time, and call time.
  • a second aspect of the embodiments of the present invention provides an apparatus for evaluating and processing a voice service, including:
  • An obtaining module configured to acquire data related to the voice service to be evaluated
  • a first evaluation module configured to use a voice service evaluation model obtained according to the voice service related record data, to evaluate the voice service related record data to be evaluated, and obtain an evaluation value of the voice service related record data to be evaluated;
  • the second evaluation module is configured to perform an evaluation process on the voice service corresponding to the voice service related record data to be evaluated according to the evaluation value.
  • the device further includes:
  • a model establishing module configured to acquire voice service related record data, where the voice service related record data includes voice service related attribute data; and the voice service evaluation model is obtained according to the voice service related attribute data.
  • the voice service related record data includes: voice service related attribute data, correspondingly,
  • the model establishing module is specifically configured to acquire the voice service evaluation model according to the voice service related attribute data.
  • the model building module includes:
  • a clustering unit configured to cluster the voice service related attribute data by using a clustering algorithm to obtain a preset number of attribute data categories
  • a calculating unit configured to calculate, according to the attribute data in each category, a classification metric value corresponding to each category
  • a sorting unit configured to sort the categories according to the classification metric value corresponding to each category, and obtain a ranking result
  • the learning unit is configured to use the machine learning method to train the superior and bad ranking results to obtain the voice business evaluation model.
  • the classification metric value corresponding to each category includes: an average value and a standard deviation corresponding to each category .
  • the sorting unit is specifically configured to use an average value corresponding to each category and attribute data in each category Attribute characteristics, sorting the respective categories to obtain a first sorting result; if the first sorting result includes the same average value, sorting the same average of the averages according to the corresponding standard deviation Obtaining a second sorting result; obtaining the pros and cons sorting result according to the first sorting result and the second sorting result.
  • the device further includes:
  • a first optimization module if the voice service related attribute data includes voice service related attribute data with correlation of two attribute value distributions, and deleting any one of the two voice service related attribute data Attribute data.
  • the device further includes:
  • a second optimization module configured to determine whether each of the voice service related attribute data has a long tail effect; if the voice service related attribute data includes voice service related attribute data having a long tail effect, canceling the long tail The long tail data in the attribute data of the effect of the voice service.
  • the second optimization module is specifically configured to use an attribute value corresponding to each voice service related attribute data. Distribution, determining whether each of the voice service related attribute data has a long tail effect;
  • the voice service related attribute data includes the following One or any combination thereof: ringing time, answering time, allocation time, authentication request time, encryption request time, service request receiving time, hanging time, ending communication time, release time, and calling time.
  • the record data related to the voice service to be evaluated is obtained, and the voice service evaluation model obtained according to the voice related record data is used, and the record data related to the voice service is evaluated, and the evaluation value of the record data related to the voice service to be evaluated is obtained.
  • the voice service corresponding to the voice service related record data is further evaluated and processed, thereby providing a more reliable basis for improving and optimizing the voice service.
  • the true state of the network can be reflected, and the network indicators can be further adjusted according to the network conditions that are reacted.
  • Embodiment 1 is a schematic flowchart of Embodiment 1 of a method for evaluating and processing a voice service according to the present invention
  • Embodiment 2 is a schematic flowchart of Embodiment 2 of a method for evaluating and processing a voice service according to the present invention
  • Embodiment 3 is a schematic flowchart of Embodiment 3 of a data evaluation method for a voice service provided by the present invention
  • Embodiment 4 is a schematic structural diagram of Embodiment 1 of an apparatus for evaluating and processing a voice service according to the present invention
  • FIG. 5 is a schematic structural diagram of Embodiment 2 of an apparatus for evaluating and processing a voice service according to the present invention
  • Embodiment 3 is a schematic structural diagram of Embodiment 3 of an apparatus for evaluating and processing a voice service according to the present invention
  • FIG. 7 is a schematic structural diagram of Embodiment 4 of an apparatus for evaluating and processing a voice service according to the present invention.
  • FIG. 8 is a schematic structural diagram of Embodiment 5 of an apparatus for evaluating and processing a voice service according to the present invention.
  • Embodiment 1 is a schematic flowchart of Embodiment 1 of a method for evaluating and processing a voice service according to the present invention. As shown in FIG. 1, the method includes:
  • the voice service related record data refers to the data in the voice service process that is completely recorded in the background during the voice service.
  • S102 Using a voice service evaluation model obtained according to the voice related record data, evaluating the record data related to the voice service to be evaluated, and obtaining an evaluation value of the record data related to the voice service to be evaluated.
  • the voice service evaluation model is obtained according to a large amount of historical voice related record data, and the obtained voice service related record data to be evaluated may also be used to establish a voice service evaluation model in the future.
  • the voice service evaluation model is used to obtain the evaluation value of the voice service related record data to be evaluated, and then comprehensively analyze and evaluate the voice service according to the evaluation value, so that the voice service can be comprehensively and reliably evaluated.
  • the true state of the network can be reflected, and the network indicators can be further adjusted according to the network conditions that are reacted.
  • the method before the evaluating the voice service related record data to be evaluated according to the voice service evaluation model obtained by using the voice related record data, includes: acquiring voice service related record data; and acquiring the voice according to the voice service related record data.
  • Business evaluation model before the evaluating the voice service related record data to be evaluated according to the voice service evaluation model obtained by using the voice related record data, includes: acquiring voice service related record data; and acquiring the voice according to the voice service related record data.
  • the voice service related record data includes voice service related attribute data.
  • a voice service related record data may include multiple voice service related attribute data.
  • the foregoing voice service evaluation model is obtained according to the voice service related record data, and specifically, the voice service evaluation model is obtained according to the voice service related attribute data.
  • the voice service related attribute data may include: Mobile Originated Call (MOC) data, Mobile Teminated Call (MTC) data, and the like, but is not limited thereto.
  • MOC Mobile Originated Call
  • MTC Mobile Teminated Call
  • the record data includes a large amount of attribute data, and the attribute values corresponding to the attribute data and the distribution of the attribute values can be obtained while obtaining the attribute data.
  • the received record data of the voice service to be evaluated is evaluated as a whole, and the evaluation value is obtained, instead of being evaluated for one of the attribute data.
  • the foregoing voice service related attribute data may include any one of the following or any combination thereof: ringing time, receiving time, allocation time, authentication request time, encryption request time, service request receiving time, hanging time, End communication time, release time, and call time.
  • these attribute data may be the above MOC. Specific attribute fields in data or MTC data, but not limited to this.
  • FIG. 2 is a schematic flowchart of Embodiment 2 of a method for processing a voice service according to the present invention.
  • the foregoing voice service evaluation model is obtained according to the attribute data of the voice service, which may be specifically:
  • the clustering algorithm is not limited here, and an appropriate clustering algorithm can be selected according to a specific application scenario.
  • each category includes at least one attribute data
  • each attribute data may correspond to a plurality of different attribute values, and according to the attribute values, the classification metric value corresponding to each category may be calculated.
  • the machine learning method can use Support Vector Machine (SVM), but it is not limited to this.
  • SVM Support Vector Machine
  • the model can be continuously updated based on the new data.
  • classification metric corresponding to each category may include: an average value and a standard deviation corresponding to each category. But it is not limited to this.
  • Table 1 records the average values corresponding to the attribute data in the five categories
  • Table 2 records the standard deviation corresponding to the attribute data in the five categories.
  • FIG. 3 is a schematic flowchart of a third embodiment of a data evaluation method for a voice service according to the present invention.
  • an average value and a standard deviation corresponding to each category are used as an example, and the foregoing classification metrics corresponding to each category are used.
  • Value, sort each category, get the result of good or bad, can be:
  • the attribute feature of the attribute data in each category is used to indicate that the attribute attribute is that the larger the attribute value is, the better, or the smaller the attribute value is. Therefore, in the specific sorting, the average values corresponding to each attribute data are first sorted for each category, and then the first sorting result of each category is comprehensively determined.
  • the same average classification is sorted according to the corresponding standard deviation, and the second sorting result is obtained.
  • the standard deviation is used to indicate the degree of aggregation of the classification.
  • the result of the ranking can be used as the evaluation value of each category.
  • the evaluation value is marked on the corresponding data for learning the acquisition model.
  • the sorting process is arranged by taking the data in Table 1 and Table 2 as an example.
  • Table 3 shows the sorting results.
  • Each column in Table 3 represents the sorting of each category corresponding to one attribute data, wherein the first column indicates the comprehensive evaluation. The value is 5 points for the first place, 4 points for the second place, and so on.
  • “ ⁇ ” in Table 3 indicates that the attribute value corresponding to the attribute data is as small as possible, and the attribute value corresponding to the attribute data of “ ⁇ ” is better.
  • c2 contains the most number of attribute data in the first sort, so c2 is optimal, 5 points, then there are 2 c4 in the first line, so c4 is second, 4 points From the first line, it is impossible to determine the order of c1, c3, and c5. Further, there are four in the second line c3, two in c1, and one in c5. Therefore, the order is c3, the third, the third, the third, the c1 2 points, 2 points, c5, the fifth point, 1 point.
  • the method further includes: if the voice service related attribute data includes two voice attribute related attribute data having relevance And deleting any voice service related attribute data in the above two voice service related attribute data.
  • the attribute values of the voice service related attribute data may be compared in turn, and if the two attribute values are found to have relevance, one of them may be deleted, and the remaining one may continue.
  • the method may further include: determining whether each voice service related attribute data has a long tail effect, if the voice service related attribute data includes The attribute data of the long tail effect eliminates the long tail data in the voice business related attribute data with the long tail effect.
  • the foregoing determines whether the attribute data of each voice service has a long tail effect. Specifically, it is determined whether the attribute data of each voice service has a long tail effect according to the attribute value distribution corresponding to the attribute data of each voice service.
  • the foregoing long tail data in the voice service related attribute data with the long tail effect is eliminated, specifically: determining a long tail data threshold according to the attribute value distribution corresponding to each voice service related attribute data; and according to the long tail data threshold , eliminating long tail data in the voice service related attribute data with long tail effect. More specifically, according to the attribute value distribution corresponding to each attribute data, the “the proportion of the long tail data in all attribute values” and the “long tail data value domain interval ratio in the attribute data” may be determined in each attribute data. Then determining long tail data according to the data that “the proportion of the long tail data in all the attribute values” is smaller than the first preset threshold and the “the ratio of the long tail data range of the attribute data is larger than the second preset threshold” Threshold.
  • the specific attribute data "authentication request time” is used as an example. It is assumed that there are 10000 attribute values corresponding to the attribute data, and the data of the range of values in the range of "10 to 1000" is 300, and then the long tail data is in all attributes. The value of the value is 3%. Assume that the value span of the attribute data is "1 to 1000". According to the distribution of the corresponding attribute values, most of the attribute values are found in "1 to 10", and a few values are distributed in "10. ⁇ 1000”, at this time, the long-tail data range of the attribute data is “10 to 1000”, that is, the long-tail data range is 99%.
  • the data of "the proportion of long tail data in all attribute values" is greater than or equal to 5%, and the long tail data range of the attribute data is retained.
  • Data that is greater than or equal to 80%, data that is less than 5% based on "the proportion of long tail data in all attribute values” and "the ratio of the long tail data range of the attribute data" are less than 80%.
  • the long tail data threshold is described.
  • the value span of the attribute data can be divided into multiple ranges to determine, for example, “1 to 1000” is divided into: “1 to 50”, “51 to 100”, and “101”. ⁇ 150”...“951 ⁇ 1000”, and then calculate the ratio of the number of attribute values in each range of values to the number of all attribute values to determine the range of the long tail data. Then determine "the proportion of long tail data in all attribute values” and "the proportion of long tail data in all attribute values”.
  • FIG. 4 is a schematic structural diagram of Embodiment 1 of an evaluation processing apparatus for a voice service according to the present invention. Intended, as shown in FIG. 1, the apparatus includes an acquisition module 401, a first evaluation module 402, and a second evaluation module 403. among them:
  • the obtaining module 401 is configured to obtain the voice service related record data to be evaluated.
  • the first evaluation module 402 is configured to perform the evaluation of the voice service related record data to be evaluated according to the voice service evaluation model obtained by the voice service related record data, and obtain the evaluation value of the record data related to the voice service to be evaluated.
  • the second evaluation module 403 is configured to perform an evaluation process on the voice service corresponding to the voice service related record data to be evaluated according to the evaluation value.
  • FIG. 5 is a schematic structural diagram of Embodiment 2 of a voice processing evaluation processing apparatus provided by the present invention.
  • the apparatus may further include: a model establishing module 404, configured to acquire a voice service related Recording data; acquiring the voice service evaluation model according to the voice service related record data.
  • the voice service related record data includes: voice service related attribute data.
  • the model establishing module 404 is specifically configured to acquire the voice service evaluation model according to the voice service related attribute data.
  • the model building module 404 can further include: a clustering unit 501, a computing unit 502, a sorting unit 503, and a learning unit 504. among them:
  • the clustering unit 501 is configured to cluster the voice service related attribute data by using a clustering algorithm to obtain a preset number of attribute data categories.
  • the calculating unit 502 is configured to calculate, according to the attribute data in each category, a classification metric value corresponding to each category.
  • the sorting unit 503 is configured to sort the classified categories according to the classification metric value corresponding to each category, and obtain a ranking result.
  • the learning unit 504 is configured to use the machine learning method to train the pros and cons ranking results to obtain the voice service evaluation model.
  • model establishing module 404 may be integrated into the foregoing device, or may be an independent device, and the established model may be transmitted to the evaluation processing device of the voice service.
  • classification metric value corresponding to each category includes: an average value and a standard deviation corresponding to the respective categories.
  • the sorting unit 503 is specifically configured to be flat according to each category Mean and attribute characteristics of the attribute data in each category, sorting the categories to obtain a first sorting result; if the first sorting result includes the same average value, the same average of the average values And sorting according to the corresponding standard deviation, obtaining a second sorting result; and obtaining the ranking result according to the first sorting result and the second sorting result.
  • FIG. 6 is a schematic structural diagram of Embodiment 3 of an apparatus for evaluating and processing a voice service according to the present invention.
  • the apparatus may further include: a first optimization module 601, if The voice service related attribute data includes two voice attribute related attribute data whose attribute value distribution has relevance, and then any one of the two voice service related attribute data is deleted.
  • FIG. 7 is a schematic structural diagram of Embodiment 4 of an apparatus for evaluating and processing a voice service according to the present invention.
  • the apparatus may further include: a second optimization module 701, configured to determine each location. Whether the voice service related attribute data has a long tail effect; if the voice service related attribute data includes voice service related attribute data having a long tail effect, eliminating the length of the voice service related attribute data having the long tail effect Tail data.
  • the second optimization module 701 determines whether each of the voice service related attribute data has a long tail effect according to the attribute value distribution corresponding to each voice service related attribute data. Correspondingly, determining a long tail data threshold according to the attribute value distribution corresponding to each voice service related attribute data; and eliminating long tail data of the voice service related attribute data having the long tail effect according to the long tail data threshold .
  • the first optimization module 601 and the second optimization module 701 may exist at the same time.
  • the voice service related attribute data includes any one of the following or any combination thereof: ringing time, receiving time, allocation time, authentication request time, encryption request time, service request receiving time, hanging time, ending communication Time, release time, call time.
  • FIG. 8 is a schematic structural diagram of Embodiment 5 of an apparatus for evaluating and processing a voice service according to the present invention.
  • the apparatus includes: a memory 801 and a processor 802. among them:
  • the memory 801 is used to store a set of instructions.
  • the processor 802 is configured to call a memory
  • the instruction set in the 801 is configured to: obtain the record data related to the voice service to be evaluated; and use the voice service evaluation model obtained according to the record data of the voice service to evaluate the record data related to the voice service to be evaluated, and obtain the And an evaluation value of the voice service related record data to be evaluated; and the voice service corresponding to the voice service related record data to be evaluated is evaluated according to the evaluation value.
  • the processor 802 is further configured to: before the evaluation of the voice service related record data to be evaluated according to the voice service evaluation model acquired according to the voice service related record data, obtain the voice service related record data; The voice service related record data is obtained, and the voice service evaluation model is obtained.
  • the voice service related record data includes: voice service related attribute data.
  • the processor 802 is specifically configured to acquire the voice service evaluation model according to the voice service related attribute data.
  • the processor 802 is specifically configured to perform clustering on the voice service related attribute data to obtain a preset number of attribute data classifications, and calculate classification metric values corresponding to each category according to the attribute data in each category. And sorting the categories according to the classification metric values corresponding to the respective categories, and obtaining the ranking results; and using the machine learning method to train the superior and bad ranking results to obtain the voice service evaluation model.
  • the classification metric value corresponding to each category includes: an average value and a standard deviation corresponding to the respective categories.
  • the processor 802 according to the average value corresponding to each category and the attribute feature of the attribute data in each category, sort the categories to obtain a first sorting result; if the first sorting result includes If the average value is the same, the classifications in which the average values are the same are sorted according to the corresponding standard deviation, and the second sort result is obtained; and the pros and cons are obtained according to the first sort result and the second sort result. Sort results.
  • the processor 802 is further configured to: before the acquiring the voice service evaluation model according to the voice service related attribute data, if the voice service related attribute data includes two voice attribute related attribute data with relevance of attribute value distribution And deleting any voice service related attribute data in the two voice service related attribute data.
  • the processor 802 is further configured to: before acquiring the voice service evaluation model according to the voice service related attribute data, determining whether each of the voice service related attribute data has a long tail effect; if the voice service related attribute data is Including voice service related attribute data having a long tail effect, the long tail data in the voice service related attribute data having the long tail effect is eliminated.
  • the processor 802 determines, according to the attribute value distribution corresponding to the voice service related attribute data, whether each of the voice service related attribute data has a long tail effect. And correspondingly, determining a long tail data threshold according to the attribute value distribution corresponding to each voice service related attribute data; and eliminating long tail data of the voice service related attribute data having the long tail effect according to the long tail data threshold .
  • the voice service related attribute data includes any one of the following or any combination thereof: ringing time, receiving time, allocation time, authentication request time, encryption request time, service request receiving time, hanging time, ending communication Time, release time, call time.
  • the disclosed apparatus and method may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the unit is only a logical function division.
  • there may be another division manner for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be in the form of hardware Implementation can also be implemented in the form of hardware plus software functional units.
  • the above-described integrated unit implemented in the form of a software functional unit can be stored in a computer readable storage medium.
  • the above software functional unit is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to perform the methods of the various embodiments of the present invention. Part of the steps.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program codes. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Quality & Reliability (AREA)
  • General Business, Economics & Management (AREA)
  • Operations Research (AREA)
  • Tourism & Hospitality (AREA)
  • Marketing (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Development Economics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Telephonic Communication Services (AREA)

Abstract

本发明实施例提供一种语音业务的评价处理方法及装置，该方法包括：获取待评价语音业务相关记录数据；采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价，获取所述待评价语音业务相关记录数据的评价值；根据所述评价值对所述待评价语音业务相关记录数据对应的语音业务进行评价处理。本发明实施例中，为改进和优化语音业务提供了更为可靠的依据。同时，根据对语音业务相关记录数据的评价还可以反应出网络的真实状况，可以根据所反应的网络状况进一步对网络指标进行调整。

Description

语音业务的评价处理方法及装置 技术领域
本发明涉及通信技术，尤其涉及一种语音业务的评价处理方法及装置。
背景技术
当前运营商提供的语音业务中，基础语音服务依旧占据着绝大部分的比例。为了能够有针对性的改进和优化语音业务，提高业务水平，需要对语音业务数据进行分析评价。
现有技术中，对于语音业务的评价主要采用的方法包括：如果相同的通信双方，连续发生两次通信的间隔时间特别短，则认为第一次通信为质量较差的通信。即仅根据连续两次通信的时间间隔来判断通信质量。
采用现有技术，使得对于语音业务的评价存在局限性，且误判和漏判的概率较大。
发明内容
本发明实施例提供一种语音业务的评价处理方法及装置，用于解决现有技术中语音业务评价存在局限性的问题。
本发明实施例第一方面提供一种语音业务的评价处理方法，包括：
获取待评价语音业务相关记录数据；
采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价，获取所述待评价语音业务相关记录数据的评价值；
根据所述评价值对所述待评价语音业务相关记录数据对应的语音业务进行评价处理。
结合第一方面，在第一方面的第一种可能的实施方式中，所述采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价之前，还包括：
获取语音业务相关记录数据；
根据所述语音业务相关记录数据，获取所述语音业务评价模型。
结合第一方面的第一种可能的实施方式，在第一方面的第二种可能的实施方式中，所述语音业务相关记录数据包括：语音业务相关属性数据，对应地，所述根据所述语音业务相关记录数据，获取所述语音业务评价模型具体包括：
根据所述语音业务相关属性数据，获取所述语音业务评价模型。
结合第一方面的第二种可能的实施方式，在第一方面的第三种可能的实施方式中，所述根据所述语音业务相关属性数据，获取所述语音业务评价模型，包括：
采用聚类算法，对所述语音业务相关属性数据进行聚类，获取预设数目个属性数据分类；
根据各分类中的属性数据，计算各分类对应的分类度量指标值；
根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果；
采用机器学习法，对所述优劣排序结果进行训练，获取所述语音业务评价模型。
结合第一方面的第三种可能的实施方式，在第一方面的第四种可能的实施方式中，所述各分类对应的分类度量指标值包括：所述各分类对应的平均值和标准差。
结合第一方面的第四种可能的实施方式，在第一方面的第五种可能的实施方式中，所述根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果，包括：
根据所述各分类对应的平均值和各分类中属性数据的属性特 征，对所述各分类进行排序，获取第一排序结果；
若所述第一排序结果中包括平均值相同的分类，则将各所述平均值相同的分类按照对应的标准差进行排序，获取第二排序结果；
根据所述第一排序结果和所述第二排序结果，获取所述优劣排序结果。
结合第一方面的第二种至第五种可能的实施方式中任一项，在第一方面的第六种可能的实施方式中，所述根据所述语音业务相关属性数据，获取所述语音业务评价模型之前，还包括：
若所述语音业务相关属性数据中包括两个属性值分布具有相关性的语音业务相关属性数据，则在所述两个语音业务相关属性数据中删除任一个语音业务相关属性数据。
结合第一方面的第二种至第五种可能的实施方式中任一项，在第一方面的第七种可能的实施方式中，所述根据所述语音业务相关属性数据，获取所述语音业务评价模型之前，还包括：
判断各所述语音业务相关属性数据是否具有长尾效应；
若所述语音业务相关属性数据中包括具有长尾效应的语音业务相关属性数据，则消除所述具有长尾效应的语音业务相关属性数据中的长尾数据。
结合第一方面的第七种可能的实施方式，在第一方面的第八种可能的实施方式中，所述判断各所述语音业务相关属性数据是否具有长尾效应，包括：
根据各所述语音业务相关属性数据对应的属性值分布，判断各所述语音业务相关属性数据是否具有长尾效应；
所述消除所述具有长尾效应的语音业务相关属性数据中的长尾数据，包括：
根据各所述语音业务相关属性数据对应的属性值分布，确定长尾数据阈值；
根据所述长尾数据阈值，消除所述具有长尾效应的语音业务相关属性数据的长尾数据。
结合第一方面的第二中可能的实施方式至第八种可能的实施方 式中任一项，在第一方面的第九种可能的实施方式中，所述语音业务相关属性数据包括下述任一项或其任意组合：响铃时间、接听时间、分配时间、鉴权请求时间、加密请求时间、业务请求接收时间、挂断时间、结束通信时间、释放时间、呼叫时间。
本发明实施例第二方面提供一种语音业务的评价处理装置，包括：
获取模块，用于获取待评价语音业务相关记录数据；
第一评价模块，用于采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价，获取所述待评价语音业务相关记录数据的评价值；
第二评价模块，用于根据所述评价值对所述待评价语音业务相关记录数据对应的语音业务进行评价处理。
结合第二方面，在第二方面的第一种可能的实施方式中，所述装置还包括：
模型建立模块，用于获取语音业务相关记录数据，其中，所述语音业务相关记录数据包括语音业务相关属性数据；根据所述语音业务相关属性数据，获取所述语音业务评价模型。
结合第二方面的第一种可能的实施方式，在第二方面的第二种可能的实施方式中，所述语音业务相关记录数据包括：语音业务相关属性数据，对应地，
所述模型建立模块，具体用于根据所述语音业务相关属性数据，获取所述语音业务评价模型。
结合第二方面的第二种可能的实施方式，在第二方面的第三种可能的实施方式中，所述模型建立模块包括：
聚类单元，用于采用聚类算法，对所述语音业务相关属性数据进行聚类，获取预设数目个属性数据分类；
计算单元，用于根据各分类中的属性数据，计算各分类对应的分类度量指标值；
排序单元，用于根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果；
学习单元，用于采用机器学习法，对所述优劣排序结果进行训练，获取所述语音业务评价模型。
结合第二方面的第三种可能的实施方式，在第二方面的第四种可能的实施方式中，所述各分类对应的分类度量指标值包括：所述各分类对应的平均值和标准差。
结合第二方面的第四种可能的实施方式，在第二方面的第五种可能的实施方式中，所述排序单元，具体用于根据所述各分类对应的平均值和各分类中属性数据的属性特征，对所述各分类进行排序，获取第一排序结果；若所述第一排序结果中包括平均值相同的分类，则将各所述平均值相同的分类按照对应的标准差进行排序，获取第二排序结果；根据所述第一排序结果和所述第二排序结果，获取所述优劣排序结果。
结合第二方面的第二种至第五种可能的实施方式中任一项，在第二方面的第五种可能的实施方式中，所述装置还包括：
第一优化模块，用于若所述语音业务相关属性数据中包括两个属性值分布具有相关性的语音业务相关属性数据，则在所述两个语音业务相关属性数据中删除任一个语音业务相关属性数据。
结合第二方面的第二种至第五种可能的实施方式中任一项，在第二方面的第六种可能的实施方式中，所述装置还包括：
第二优化模块，用于判断各所述语音业务相关属性数据是否具有长尾效应；若所述语音业务相关属性数据中包括具有长尾效应的语音业务相关属性数据，则消除所述具有长尾效应的语音业务相关属性数据中的长尾数据。
结合第二方面的第六种可能的实施方式，在第二方面的第七种可能的实施方式中，所述第二优化模块，具体用于根据各所述语音业务相关属性数据对应的属性值分布，判断各所述语音业务相关属性数据是否具有长尾效应；
根据各所述语音业务相关属性数据对应的属性值分布，确定长尾数据阈值；根据所述长尾数据阈值，消除所述具有长尾效应的语音业务相关属性数据的长尾数据。
结合第二方面的第二中可能的实施方式至第八种可能的实施方式中任一项，在第二方面的第九种可能的实施方式中，所述语音业务相关属性数据包括下述任一项或其任意组合：响铃时间、接听时间、分配时间、鉴权请求时间、加密请求时间、业务请求接收时间、挂断时间、结束通信时间、释放时间、呼叫时间。
本发明实施例中，获取待评价语音业务相关记录数据，采用根据语音相关记录数据获取的语音业务评价模型，对待评价语音业务相关记录数据进行评价，获取待评价语音业务相关记录数据的评价值，并根据这些语音业务相关记录数据的评价值进一步对待评价语音业务相关记录数据对应的语音业务进行综合评价处理，从而为改进和优化语音业务提供了更为可靠的依据。同时，根据对语音业务相关记录数据的评价还可以反应出网络的真实状况，可以根据所反应的网络状况进一步对网络指标进行调整。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1为本发明提供的语音业务的评价处理方法实施例一的流程示意图；
图2为本发明提供的语音业务的评价处理方法实施例二的流程示意图；
图3为本发明提供的语音业务的数据评价方法实施例三的流程示意图；
图4为本发明提供的语音业务的评价处理装置实施例一的结构示意图；
图5为本发明提供的语音业务的评价处理装置实施例二的结构示意图；
图6为本发明提供的语音业务的评价处理装置实施例三的结构示意图；
图7为本发明提供的语音业务的评价处理装置实施例四的结构示意图；
图8为本发明提供的语音业务的评价处理装置实施例五的结构示意图。
具体实施方式
为使本发明实施例的目的、技术方案和优点更加清楚，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
图1为本发明提供的语音业务的评价处理方法实施例一的流程示意图，如图1所示，该方法包括：
S101、获取待评价语音业务相关记录数据。
语音业务相关记录数据是指语音业务进行过程中，后台完整记录的语音业务过程中的数据。
S102、采用根据语音相关记录数据获取的语音业务评价模型，对上述待评价语音业务相关记录数据进行评价，获取该待评价语音业务相关记录数据的评价值。
具体地，这里语音业务评价模型是根据大量的历史语音相关记录数据获取的，上述获取的待评价语音业务相关记录数据未来也可以用于建立语音业务评价模型。
S103、根据上述评价值对上述待评价语音业务相关记录数据对应的语音业务进行评价处理。
通过该语音业务评价模型来获取待评价语音业务相关记录数据的评价值，然后进一步根据该评价值来全面的综合地分析评价语音业务，可以对语音业务进行全面可靠的评价。
本实施例中，获取待评价语音业务相关记录数据，采用根据语音相关记录数据获取的语音业务评价模型，对待评价语音业务相关记录数据进行评价，获取待评价语音业务相关记录数据的评价值，并根据这些语音业务相关记录数据的评价值进一步对待评价语音业务相关记录数据对应的语音业务进行综合评价处理，从而为改进和优化语音业务提供了更为可靠的依据。同时，根据对语音业务相关记录数据的评价还可以反应出网络的真实状况，可以根据所反应的网络状况进一步对网络指标进行调整。
具体地，上述采用根据语音相关记录数据获取的语音业务评价模型，对上述待评价语音业务相关记录数据进行评价之前，包括：获取语音业务相关记录数据；根据这些语音业务相关记录数据，获取上述语音业务评价模型。
其中，语音业务相关记录数据包括语音业务相关属性数据。一般情况下，一条语音业务相关记录数据可以包括多个语音业务相关属性数据。对应地，上述根据这些语音业务相关记录数据，获取上述语音业务评价模型，具体为根据上述语音业务相关属性数据，获取语音业务评价模型。
语音业务相关属性数据可以包括：移动主叫(Mobile Originated Call，简称MOC)数据、移动被叫(Mobile Teminated Call，简称MTC)数据等，但并不以此为限。需要说明的是，在进行语音通信的过程中，后台会完整的记录整个通信过程中的所有数据，在建立模型之前，在数据库中获取大量与语音业务相关记录数据。这些记录数据中包括大量属性数据，获取属性数据的同时可以获取这些属性数据对应的属性值以及属性值的分布。需要说明的是，具体评价时，是对接收到的待评价语音业务相关记录数据整体进行评价，获取评价值，而不是针对其中的某个属性数据进行评价。
更具体地，上述语音业务相关属性数据可以包括下述任一项或其任意组合：响铃时间、接听时间、分配时间、鉴权请求时间、加密请求时间、业务请求接收时间、挂断时间、结束通信时间、释放时间、呼叫时间。具体实现过程中，这些属性数据可能是上述MOC 数据或MTC数据中的具体属性字段，但并不以此为限。
图2为本发明提供的语音业务的评价处理方法实施例二的流程示意图，如图2所示，上述根据语音业务相关属性数据，获取上述语音业务评价模型，具体可以为：
S201、采用聚类算法，对上述语音业务相关属性数据进行聚类，获取预设数目个属性数据分类。即可以根据具体需要，预设将属性数据聚类为几类。
此处对聚类算法不作限制，可以根据具体的应用场景选择合适的聚类算法。
S202、根据各分类中的属性数据，计算各分类对应的分类度量指标值。
具体地，每个分类下包括至少一种属性数据，每个属性数据可能会对应多个不同的属性值，根据这些属性值就可以计算出每个分类对应的分类度量指标值。
前述获取语音业务相关属性数据时，可以同时获取到每个属性数据对应的大量属性值，以及这些属性值的分布情况。
S203、根据各分类对应的分类度量指标值，对各分类进行排序，获取优劣排序结果。
S204、采用机器学习法，对上述优劣排序结果进行训练，获取上述语音业务评价模型。
这里机器学习法可以选用支持向量机(Support Vector Machine，简称SVM)，但并不以此为限。当然，随着数据的不断更新，可以不断的根据新数据来更新模型。
更进一步地，上述各分类对应的分类度量指标可以包括：各分类对应的平均值和标准差。但并不以此为限。
假设聚类后有5种分类，举例说明，表1记载这5类中属性数据分别对应的平均值，表2记载这5类中属性数据对应的标准差。
表一
Figure PCTCN2015072609-appb-000001
Figure PCTCN2015072609-appb-000002
表2
Figure PCTCN2015072609-appb-000003
图3为本发明提供的语音业务的数据评价方法实施例三的流程示意图，如图3所示，以各分类对应的平均值和标准差为例进行说明，上述根据各分类对应的分类度量指标值，对各分类进行排序，获取优劣排序结果，可以为：
S301、根据上述分类对应的平均值和各分类中属性数据的属性特征，对上述各分类进行排序，获取第一排序结果。
具体地，各分类中属性数据的属性特征用于表明该属性特性是属性值越大越优，还是属性值越小越优。因此，具体排序时，先按每种属性数据对应的平均值分别为各分类排序，然后再综合确定各分类的第一排序结果。
S302、若该第一排序结果中包括平均值相同的分类，则将各平均值相同的分类按照对应的标准差进行排序，获取第二排序结果。
假设聚类后有5种分类(cluster)，记为：c1、c2、c3、c4、c5，根据平均值排序后，在第一排序结果中c3和c4并列，即它们平均值相同，那么再按照它们的标准差对c3和c4进行排序，获取第二排序结果。
与平均值不同的是，标准差用于表示分类的聚集度，标准差越小表示对应分类中的属性数据越集中，因此，按标准差排列时都是标准差越小则对应的分类越优。
S303、根据第一排序结果和第二排序结果，获取上述优劣排序结果。即综合第一排序结果和第二排序结果，就可以整理获得最终的优劣排序结果。
具体地，该优劣排序结果就可以作为各分类的评价值。将该评价值标记在对应的数据上，用于学习获取模型。
举例说明排序过程，以表1和表2中的数据为例进行排列，表3显示排序结果，表3中每一列表示一种属性数据对应的各分类的排序，其中，第1列表示综合评价值，即排第1位的记5分，排第2位的记4分，依次类推。需要说明的是，表3中“↓”表示对应属性数据的属性值越小越优，“↑”对应属性数据的属性值越大越优。
表3
Figure PCTCN2015072609-appb-000004
从第1行可以看出，c2在排序第1的属性数据中包含的个数最多，因此c2最优，记5分，然后第1行中有2个c4，因此c4第二，记4分，从第1行中无法确定c1、c3、c5的排序，进而参照第2行c3有4个，c1有2个，c5有1个，因此排序依次为c3排第3记3分，c1排第4记2分，c5排第5记1分。
进一步地，为了优化样本数据，上述根据语音业务相关属性数据，获取上述语音业务评价模型之前，还包括：若上述语音业务相关属性数据中包括两个属性值分布具有相关性的语音业务相关属性数据，则在上述两个语音业务相关属性数据中删除任一个语音业务相关属性数据。具体实现过程中，可以将语音业务相关属性数据的属性值依次两两进行比较，两两比较时，如果发现这两个属性值分布具有相关性，则删除其中之一，剩余的一个可以继续进行比较，以此类推，就可以在属性值分布特性相同或相似的多个属性数据中只保留一个属性数据，以避免不必要的加权累计效应。
另外，为了优化样本数据，上述根据语音业务相关属性数据，获取上述语音业务评价模型之前，还可以包括：判断各语音业务相关属性数据是否具有长尾效应，若这些语音业务相关属性数据中包括具有长尾效应的属性数据，则消除具有长尾效应的语音业务相关属性数据中的长尾数据。
上述判断各语音业务相关属性数据是否具有长尾效应，具体为：根据各语音业务相关属性数据对应的属性值分布，判断各语音业务相关属性数据是否具有长尾效应。
进一步地，上述消除上述具有长尾效应的语音业务相关属性数据中的长尾数据，具体为：根据各语音业务相关属性数据对应的属性值分布，确定长尾数据阈值；根据该长尾数据阈值，消除具有长尾效应的语音业务相关属性数据中的长尾数据。更具体地，可以根据各属性数据对应的属性值分布，确定各属性数据中“长尾数据在所有属性值中的占比”和“该属性数据中的长尾数据值域区间占比”，然后根据“长尾数据在所有属性值中的占比”小于第一预设阈值的数据和“该属性数据的长尾数据值域区间占比”大于第二预设阈值的数据确定长尾数据阈值。
以具体一个属性数据“鉴权请求时间”举例说明，假设该属性数据对应的属性值有10000个，值域分布在“10～1000”区间的数据为300个，那么此时长尾数据在所有属性值中的占比为3％；假设该属性数据的值域跨度为“1～1000”，根据对应的属性值分布，发现大部分属性值集中在“1～10”，少数值分布在“10～1000”，此时该属性数据的长尾数据值域区间为“10～1000”，即长尾数据值域区间占比为99％。假设第一阈值为5％，第二阈值为80％，那么保留“长尾数据在所有属性值中的占比”大于等于5％的数据，保留“该属性数据的长尾数据值域区间占比”大于等于80％的数据，根据“长尾数据在所有属性值中的占比”小于5％的数据和“该属性数据的长尾数据值域区间占比”小于80％的数据确定所述长尾数据阈值。
具体实现过程中，可以将属性数据的值域跨度切分为多个值域区间来进行判断，例如将“1～1000”切分为：“1～50”、“51～100”、“101～150”……“951～1000”，进而计算确定每个值域区间中的属性值数量占所有属性值数量的比例，来确定长尾数据的值域区间。然后再确定“长尾数据在所有属性值中的占比”和“长尾数据在所有属性值中的占比”。
图4为本发明提供的语音业务的评价处理装置实施例一的结构示 意图，如图1所示，该装置包括：获取模块401、第一评价模块402和第二评价模块403。其中：
获取模块401，用于获取待评价语音业务相关记录数据。
第一评价模块402，用于采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价，获取所述待评价语音业务相关记录数据的评价值。
第二评价模块403，用于根据所述评价值对所述待评价语音业务相关记录数据对应的语音业务进行评价处理。
图5为本发明提供的语音业务的评价处理装置实施例二的结构示意图，如图5所示，在图4的基础上，该装置还可以包括：模型建立模块404，用于获取语音业务相关记录数据；根据所述语音业务相关记录数据，获取所述语音业务评价模型。
具体地，所述语音业务相关记录数据包括：语音业务相关属性数据。相应地，模型建立模块404，具体用于根据所述语音业务相关属性数据，获取所述语音业务评价模型。
继续参照图5，模型建立模块404还可以包括：聚类单元501、计算单元502、排序单元503以及学习单元504。其中：
聚类单元501，用于采用聚类算法，对所述语音业务相关属性数据进行聚类，获取预设数目个属性数据分类。计算单元502，用于根据各分类中的属性数据，计算各分类对应的分类度量指标值。排序单元503，用于根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果。学习单元504，用于采用机器学习法，对所述优劣排序结果进行训练，获取所述语音业务评价模型。
需要说明的是，具体实现过程中，模型建立模块404可以集成于上述装置，也可以是一个独立的装置，将建立完成的模型传送给上述语音业务的评价处理装置即可。
进一步地，所述各分类对应的分类度量指标值包括：所述各分类对应的平均值和标准差。
更进一步地，排序单元503，具体用于根据所述各分类对应的平 均值和各分类中属性数据的属性特征，对所述各分类进行排序，获取第一排序结果；若所述第一排序结果中包括平均值相同的分类，则将各所述平均值相同的分类按照对应的标准差进行排序，获取第二排序结果；根据所述第一排序结果和所述第二排序结果，获取所述优劣排序结果。
图6为本发明提供的语音业务的评价处理装置实施例三的结构示意图，如图6所示，在图5的基础上，该装置还可以包括：第一优化模块601，用于若所述语音业务相关属性数据中包括两个属性值分布具有相关性的语音业务相关属性数据，则在所述两个语音业务相关属性数据中删除任一个语音业务相关属性数据。
图7为本发明提供的语音业务的评价处理装置实施例四的结构示意图，如图7所示，在图5的基础上，该装置还可以包括：第二优化模块701，用于判断各所述语音业务相关属性数据是否具有长尾效应；若所述语音业务相关属性数据中包括具有长尾效应的语音业务相关属性数据，则消除所述具有长尾效应的语音业务相关属性数据中的长尾数据。
更具体地，第二优化模块701，根据各所述语音业务相关属性数据对应的属性值分布，判断各所述语音业务相关属性数据是否具有长尾效应。并相应地，根据各所述语音业务相关属性数据对应的属性值分布，确定长尾数据阈值；根据所述长尾数据阈值，消除所述具有长尾效应的语音业务相关属性数据的长尾数据。
需要说明的是，在某些应用场景下，上述第一优化模块601和第二优化模块701可以同时存在。
另外，所述语音业务相关属性数据包括下述任一项或其任意组合：响铃时间、接听时间、分配时间、鉴权请求时间、加密请求时间、业务请求接收时间、挂断时间、结束通信时间、释放时间、呼叫时间。
图8为本发明提供的语音业务的评价处理装置实施例五的结构示意图，该装置包括：存储器801和处理器802。其中：
存储器801用于存储指令集。该处理器802被配置为调用存储器 801中的指令集，以执行如下流程：获取待评价语音业务相关记录数据；采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价，获取所述待评价语音业务相关记录数据的评价值；根据所述评价值对所述待评价语音业务相关记录数据对应的语音业务进行评价处理。
进一步地，处理器802，还用于在所述采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价之前，获取语音业务相关记录数据；根据所述语音业务相关记录数据，获取所述语音业务评价模型。
具体地，所述语音业务相关记录数据包括：语音业务相关属性数据。相应地，处理器802，具体用于根据所述语音业务相关属性数据，获取所述语音业务评价模型。
处理器802，具体用于采用聚类算法，对所述语音业务相关属性数据进行聚类，获取预设数目个属性数据分类；根据各分类中的属性数据，计算各分类对应的分类度量指标值；根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果；采用机器学习法，对所述优劣排序结果进行训练，获取所述语音业务评价模型。
其中，所述各分类对应的分类度量指标值包括：所述各分类对应的平均值和标准差。
更具体地，处理器802，根据所述各分类对应的平均值和各分类中属性数据的属性特征，对所述各分类进行排序，获取第一排序结果；若所述第一排序结果中包括平均值相同的分类，则将各所述平均值相同的分类按照对应的标准差进行排序，获取第二排序结果；根据所述第一排序结果和所述第二排序结果，获取所述优劣排序结果。
处理器802，还用于在根据所述语音业务相关属性数据，获取所述语音业务评价模型之前，若所述语音业务相关属性数据中包括两个属性值分布具有相关性的语音业务相关属性数据，则在所述两个语音业务相关属性数据中删除任一个语音业务相关属性数据。
处理器802，还用于在根据所述语音业务相关属性数据，获取所述语音业务评价模型之前，判断各所述语音业务相关属性数据是否具有长尾效应；若所述语音业务相关属性数据中包括具有长尾效应的语音业务相关属性数据，则消除所述具有长尾效应的语音业务相关属性数据中的长尾数据。
具体地，处理器802，根据各所述语音业务相关属性数据对应的属性值分布，判断各所述语音业务相关属性数据是否具有长尾效应。以及相应地，根据各所述语音业务相关属性数据对应的属性值分布，确定长尾数据阈值；根据所述长尾数据阈值，消除所述具有长尾效应的语音业务相关属性数据的长尾数据。
另外，所述语音业务相关属性数据包括下述任一项或其任意组合：响铃时间、接听时间、分配时间、鉴权请求时间、加密请求时间、业务请求接收时间、挂断时间、结束通信时间、释放时间、呼叫时间。
在本发明所提供的几个实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式 实现，也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元，可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本发明各个实施例所述方法的部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。

Claims (20)

  1. 一种语音业务的评价处理方法，其特征在于，包括：
    获取待评价语音业务相关记录数据；
    采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价，获取所述待评价语音业务相关记录数据的评价值；
    根据所述评价值对所述待评价语音业务相关记录数据对应的语音业务进行评价处理。
  2. 根据权利要求1所述的方法，其特征在于，所述采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价之前，还包括：
    获取语音业务相关记录数据；
    根据所述语音业务相关记录数据，获取所述语音业务评价模型。
  3. 根据权利要求2所述的方法，其特征在于，所述语音业务相关记录数据包括：语音业务相关属性数据，对应地，所述根据所述语音业务相关记录数据，获取所述语音业务评价模型具体包括：
    根据所述语音业务相关属性数据，获取所述语音业务评价模型。
  4. 根据权利要求3所述的方法，其特征在于，所述根据所述语音业务相关属性数据，获取所述语音业务评价模型，包括：
    采用聚类算法，对所述语音业务相关属性数据进行聚类，获取预设数目个属性数据分类；
    根据各分类中的属性数据，计算各分类对应的分类度量指标值；
    根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果；
    采用机器学习法，对所述优劣排序结果进行训练，获取所述语音业务评价模型。
  5. 根据权利要求4所述的方法，其特征在于，所述各分类对应 的分类度量指标值包括：所述各分类对应的平均值和标准差。
  6. 根据权利要求5所述的方法，其特征在于，所述根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果，包括：
    根据所述各分类对应的平均值和各分类中属性数据的属性特征，对所述各分类进行排序，获取第一排序结果；
    若所述第一排序结果中包括平均值相同的分类，则将各所述平均值相同的分类按照对应的标准差进行排序，获取第二排序结果；
    根据所述第一排序结果和所述第二排序结果，获取所述优劣排序结果。
  7. 根据权利要求3至6任一项所述的方法，其特征在于，所述根据所述语音业务相关属性数据，获取所述语音业务评价模型之前，还包括：
    若所述语音业务相关属性数据中包括两个属性值分布具有相关性的语音业务相关属性数据，则在所述两个语音业务相关属性数据中删除任一个语音业务相关属性数据。
  8. 根据权利要求3至6任一项所述的方法，其特征在于，所述根据所述语音业务相关属性数据，获取所述语音业务评价模型之前，还包括：
    判断各所述语音业务相关属性数据是否具有长尾效应；
    若所述语音业务相关属性数据中包括具有长尾效应的语音业务相关属性数据，则消除所述具有长尾效应的语音业务相关属性数据中的长尾数据。
  9. 根据权利要求8所述的方法，其特征在于，所述判断各所述语音业务相关属性数据是否具有长尾效应，包括：
    根据各所述语音业务相关属性数据对应的属性值分布，判断各所述语音业务相关属性数据是否具有长尾效应；
    所述消除所述具有长尾效应的语音业务相关属性数据中的长尾数据，包括：
    根据各所述语音业务相关属性数据对应的属性值分布，确定长 尾数据阈值；
    根据所述长尾数据阈值，消除所述具有长尾效应的语音业务相关属性数据的长尾数据。
  10. 根据权利要求3-9任一项所述的方法，其特征在于，所述语音业务相关属性数据包括下述任一项或其任意组合：响铃时间、接听时间、分配时间、鉴权请求时间、加密请求时间、业务请求接收时间、挂断时间、结束通信时间、释放时间、呼叫时间。
  11. 一种语音业务的评价处理装置，其特征在于，包括：
    获取模块，用于获取待评价语音业务相关记录数据；
    第一评价模块，用于采用根据语音业务相关记录数据获取的语音业务评价模型，对所述待评价语音业务相关记录数据进行评价，获取所述待评价语音业务相关记录数据的评价值；
    第二评价模块，用于根据所述评价值对所述待评价语音业务相关记录数据对应的语音业务进行评价处理。
  12. 根据权利要求11所述的装置，其特征在于，还包括：
    模型建立模块，用于获取语音业务相关记录数据；根据所述语音业务相关属性数据，获取所述语音业务评价模型。
  13. 根据权利要求12所述的装置，其特征在于，所述语音业务相关记录数据包括：语音业务相关属性数据，对应地，
    所述模型建立模块，具体用于根据所述语音业务相关属性数据，获取所述语音业务评价模型。
  14. 根据权利要求13所述的装置，其特征在于，所述模型建立模块包括：
    聚类单元，用于采用聚类算法，对所述语音业务相关属性数据进行聚类，获取预设数目个属性数据分类；
    计算单元，用于根据各分类中的属性数据，计算各分类对应的分类度量指标值；
    排序单元，用于根据所述各分类对应的所述分类度量指标值，对所述各分类进行排序，获取优劣排序结果；
    学习单元，用于采用机器学习法，对所述优劣排序结果进行训 练，获取所述语音业务评价模型。
  15. 根据权利要求13所述的装置，其特征在于，所述各分类对应的分类度量指标值包括：所述各分类对应的平均值和标准差。
  16. 根据权利要求15所述的装置，其特征在于，所述排序单元，具体用于根据所述各分类对应的平均值和各分类中属性数据的属性特征，对所述各分类进行排序，获取第一排序结果；若所述第一排序结果中包括平均值相同的分类，则将各所述平均值相同的分类按照对应的标准差进行排序，获取第二排序结果；根据所述第一排序结果和所述第二排序结果，获取所述优劣排序结果。
  17. 根据权利要求13至16任一项所述的装置，其特征在于，还包括：
    第一优化模块，用于若所述语音业务相关属性数据中包括两个属性值分布具有相关性的语音业务相关属性数据，则在所述两个语音业务相关属性数据中删除任一个语音业务相关属性数据。
  18. 根据权利要求13至16任一项所述的装置，其特征在于，还包括：
    第二优化模块，用于判断各所述语音业务相关属性数据是否具有长尾效应；若所述语音业务相关属性数据中包括具有长尾效应的语音业务相关属性数据，则消除所述具有长尾效应的语音业务相关属性数据中的长尾数据。
  19. 根据权利要求18所述的装置，其特征在于，所述第二优化模块，具体用于根据各所述语音业务相关属性数据对应的属性值分布，判断各所述语音业务相关属性数据是否具有长尾效应；
    根据各所述语音业务相关属性数据对应的属性值分布，确定长尾数据阈值；根据所述长尾数据阈值，消除所述具有长尾效应的语音业务相关属性数据的长尾数据。
  20. 根据权利要求13-19任一项所述的装置，其特征在于，所述语音业务相关属性数据包括下述任一项或其任意组合：响铃时间、接听时间、分配时间、鉴权请求时间、加密请求时间、业务请求接收时间、挂断时间、结束通信时间、释放时间、呼叫时间。
PCT/CN2015/072609 2014-06-16 2015-02-10 语音业务的评价处理方法及装置 WO2015192668A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP15807778.4A EP3002692A4 (en) 2014-06-16 2015-02-10 EVALUATION PROCESSING AND DEVICE FOR A LANGUAGE SERVICE
US14/971,103 US9813549B2 (en) 2014-06-16 2015-12-16 Voice service evaluation processing method and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201410266958.5A CN105224558A (zh) 2014-06-16 2014-06-16 语音业务的评价处理方法及装置
CN201410266958.5 2014-06-16

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US14/971,103 Continuation US9813549B2 (en) 2014-06-16 2015-12-16 Voice service evaluation processing method and apparatus

Publications (1)

Publication Number Publication Date
WO2015192668A1 true WO2015192668A1 (zh) 2015-12-23

Family

ID=54934838

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/072609 WO2015192668A1 (zh) 2014-06-16 2015-02-10 语音业务的评价处理方法及装置

Country Status (4)

Country Link
US (1) US9813549B2 (zh)
EP (1) EP3002692A4 (zh)
CN (1) CN105224558A (zh)
WO (1) WO2015192668A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106531190B (zh) * 2016-10-12 2020-05-05 科大讯飞股份有限公司 语音质量评价方法和装置
US10496396B2 (en) 2017-09-29 2019-12-03 Oracle International Corporation Scalable artificial intelligence driven configuration management
WO2019177481A1 (en) * 2018-03-12 2019-09-19 Ringcentral, Inc., (A Delaware Corporation) System and method for evaluating the quality of a communication session
US10789065B2 (en) * 2018-05-07 2020-09-29 Oracle lnternational Corporation Method for automatically selecting configuration clustering parameters
CN116156439B (zh) * 2023-04-18 2023-06-20 深圳市微克科技有限公司 智能穿戴电子对讲交互系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1905725A (zh) * 2006-07-31 2007-01-31 华为技术有限公司 语音业务测试系统和方法
CN102448083A (zh) * 2010-10-12 2012-05-09 中兴通讯股份有限公司 一种无线通信语音质量评价装置、系统及方法
CN102790996A (zh) * 2011-05-19 2012-11-21 中兴通讯股份有限公司 下行无线传输质量测量方法、装置和系统
CN103686833A (zh) * 2013-12-17 2014-03-26 中国联合网络通信集团有限公司 移动网络语音质量评估方法和装置

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6574216B1 (en) * 1997-03-11 2003-06-03 Verizon Services Corp. Packet data network voice call quality monitoring
US6785361B1 (en) * 2001-08-30 2004-08-31 Bellsouth Intellectual Property Corporation System and method for performance measurement quality assurance
US7076036B1 (en) * 2001-10-17 2006-07-11 Sprint Communications Company L.P. Traffic and capacity modeling process
US6914967B1 (en) * 2004-03-11 2005-07-05 Network Enhanced Telecom, Llp Method for measuring quality of service in a telecommunications network
US7843991B2 (en) * 2005-04-21 2010-11-30 Qualcomm Incorporated Methods and apparatus for monitoring voice quality on a wireless communication device
US8503313B1 (en) * 2006-12-31 2013-08-06 At&T Intellectual Property Ii, L.P. Method and apparatus for detecting a network impairment using call detail records
US7957731B2 (en) * 2008-02-26 2011-06-07 Metrico Wireless, Inc. System and method for determining mobile telephone voice quality in a live network
US9071677B2 (en) * 2013-02-12 2015-06-30 Unify Square, Inc. Enhanced data capture, analysis, and reporting for unified communications
CN103560919B (zh) 2013-11-15 2017-07-04 中国联合网络通信集团有限公司 一种基于语音业务的质量评估方法及设备
US9119086B1 (en) * 2014-05-08 2015-08-25 Cellco Partnership Evaluating 3G and voice over long term evolution voice quality
US11017330B2 (en) * 2014-05-20 2021-05-25 Elasticsearch B.V. Method and system for analysing data

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1905725A (zh) * 2006-07-31 2007-01-31 华为技术有限公司 语音业务测试系统和方法
CN102448083A (zh) * 2010-10-12 2012-05-09 中兴通讯股份有限公司 一种无线通信语音质量评价装置、系统及方法
CN102790996A (zh) * 2011-05-19 2012-11-21 中兴通讯股份有限公司 下行无线传输质量测量方法、装置和系统
CN103686833A (zh) * 2013-12-17 2014-03-26 中国联合网络通信集团有限公司 移动网络语音质量评估方法和装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3002692A4 *

Also Published As

Publication number Publication date
EP3002692A1 (en) 2016-04-06
CN105224558A (zh) 2016-01-06
US20160105544A1 (en) 2016-04-14
US9813549B2 (en) 2017-11-07
EP3002692A4 (en) 2016-08-17

Similar Documents

Publication Publication Date Title
WO2015192668A1 (zh) 语音业务的评价处理方法及装置
US10922206B2 (en) Systems and methods for determining performance metrics of remote relational databases
CN104579768B (zh) 客户端升级方法和装置
US20150135320A1 (en) Methods and apparatus to identify malicious activity in a network
CN105719104B (zh) 一种智能审批的方法及装置
CN106911780A (zh) 业务id生成方法、装置及系统
US9131341B2 (en) System and method for locating a mobile device
US9305458B2 (en) Proximity-detection systems and methods
CN111047425B (zh) 一种行为预测方法及装置
CN108985954B (zh) 一种建立各标识的关联关系的方法以及相关设备
US11120223B2 (en) Probabilistic word embeddings for text classification
US11431726B2 (en) Systems and methods for communication between a plurality of electronic devices based on subscriber access score
US20140341466A1 (en) Noise estimation for images
US20230379763A1 (en) Dynamic continuous quality of service adjustment system
CN110245684B (zh) 数据处理方法、电子设备和介质
CN111062634A (zh) 审批任务分配方法、装置、计算机设备及存储介质
CN109472312B (zh) 应用程序评估方法、装置、计算机设备及存储介质
CN111177363A (zh) 一种群组消息的写入方法、装置、服务器和存储介质
AU2014349181B2 (en) Techniques to rate-adjust data usage with a virtual private network
CN107018039B (zh) 测试服务器集群性能瓶颈的方法和装置
CN111368858B (zh) 用户满意度评估方法及装置
CN113313196B (zh) 标注数据处理方法、相关装置及计算机程序产品
CN104811365B (zh) 基于实时沟通社交群体的消息投放方法、系统及服务器
US9479579B2 (en) Grouping processing method and system
CN108880935B (zh) 网络节点重要度的获得方法和装置、设备、存储介质

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 2015807778

Country of ref document: EP

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15807778

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE