CN111859958A - High-complaint-risk user identification method, complaint early warning method and related equipment - Google Patents

High-complaint-risk user identification method, complaint early warning method and related equipment Download PDF

Info

Publication number
CN111859958A
CN111859958A CN202010714915.4A CN202010714915A CN111859958A CN 111859958 A CN111859958 A CN 111859958A CN 202010714915 A CN202010714915 A CN 202010714915A CN 111859958 A CN111859958 A CN 111859958A
Authority
CN
China
Prior art keywords
user
complaint
word
early warning
score value
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN202010714915.4A
Other languages
Chinese (zh)
Inventor
林峰
尹钏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Property and Casualty Insurance Company of China Ltd
Original Assignee
Ping An Property and Casualty Insurance Company of China Ltd
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 Ping An Property and Casualty Insurance Company of China Ltd filed Critical Ping An Property and Casualty Insurance Company of China Ltd
Priority to CN202010714915.4A priority Critical patent/CN111859958A/en
Publication of CN111859958A publication Critical patent/CN111859958A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • 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
    • G06Q30/00—Commerce
    • G06Q30/01—Customer relationship services

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Business, Economics & Management (AREA)
  • Computational Linguistics (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Finance (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Telephonic Communication Services (AREA)

Abstract

The invention relates to the technical field of information, and discloses a high complaint risk user identification method, a complaint early warning method and related equipment, wherein the high complaint risk user identification method comprises the following steps: preprocessing the chat corpus of the user to obtain effective phrases in the chat corpus, comparing the effective phrases with a preset complaint early warning text library, calculating a first score value, calculating a time interval of each time that the user speaks relative to the last speaking, counting the times that the time interval is lower than the preset time interval, and calculating a second score value based on the times; calculating a total score value according to the first score value and the second score value; judging whether the total score value is larger than a preset threshold value or not; and if so, determining that the user is a high complaint risk user. By the method, the emotion of the user can be monitored in real time, business personnel can conveniently adjust the dialect, the complaint risk of the user is controlled, the number of complaints is reduced, and the satisfaction degree of the user is improved. In addition, the invention also relates to a block chain technology, and the chat corpus can be stored in the block chain.

Description

High-complaint-risk user identification method, complaint early warning method and related equipment
Technical Field
The invention relates to the technical field of information, in particular to a high-complaint-risk user identification method, a complaint early warning method and related equipment.
Background
With continuous deepening of innovation, marketization degree is continuously improved, market competition is continuously intensified, a plurality of enterprises urgently need to change a traditional thinking mode and a working mode, market service awareness is established, trust of customers is gained, market share is guaranteed, the enterprises need to control product quality, the enterprises need to optimize after-sales service, independent awareness and right-of-maintenance awareness of users are continuously intensified, and the users can express dissatisfaction of the users through complaints under the condition that high-quality service cannot be obtained.
At present, for the problem of complaint of a user, the complaint number of service personnel, the customer satisfaction degree, the problem solving time effectiveness and the like are mostly analyzed through a system, or manual spot check and carding are mainly performed, because of lack of automatic monitoring and analysis, multiple sides are more important to follow responsibility afterwards, and the service personnel can not be warned by recognizing the risk that the user possibly complaints when the service personnel communicates with the user about a case, the occurrence of complaint events is avoided, and for partial complaint warning, such as the complaint warning of telephone call, the duration of calling the complaint call is short, the main purpose is to provide information related to the complaint, the customer rarely mentions own feeling and mood, so the complaint is judged through the recognition of voice and tone in the telephone recording, the data latitude is small, and the risk of complaint of the user is not enough to be recognized and warned. In addition, the scene of converting the voice into the text only can cover the better condition of the Mandarin, and the number of complaints can be accurately estimated is less.
Disclosure of Invention
The invention mainly aims to solve the technical problem that the complaint risk of a user cannot be identified in real time in the existing language material analysis of communication between business personnel and customers.
The invention provides a method for identifying a user with high complaint risk, which comprises the following steps:
obtaining a chat corpus of a user in a communication process between a service person and the user;
preprocessing the chat corpus to obtain effective phrases in the chat corpus, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing;
comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts;
calculating a time interval of the user speaking for the last time each time, counting the times that the time interval is lower than a preset time interval, and calculating a second score value of the user based on the times;
calculating a total score value according to the first score value and the second score value;
judging whether the total score value is larger than a preset threshold value or not;
and if so, determining that the user is a high complaint risk user.
Optionally, in a first implementation manner of the first aspect of the present invention, the comparing the effective phrase with a preset complaint early-warning text library, and calculating the first score value of the user includes:
acquiring the word weight of the complaint early warning word segmentation;
determining the word weight of the effective words in the effective word group according to the word weight of the complaint early warning word segmentation;
determining the similarity between the effective word group and each complaint early warning text in the complaint early warning text library according to the word weight of the complaint early warning segmentation word and the word weight of the effective word;
and taking the maximum value of the similarity in the similarities as the first scoring value.
Optionally, in a second implementation manner of the first aspect of the present invention, the determining, according to the word weight of the complaint early warning segmented word, the word weight of the effective word in the effective word group includes:
matching effective words in the effective word groups with complaint early warning participles in the complaint early warning text library;
when complaint early-warning participles consistent with effective words in the effective phrases exist in the complaint early-warning text base, word weights of the complaint early-warning participles consistent with the effective words in the complaint early-warning text base are obtained;
obtaining the word weight of the effective words in the effective word groups according to the word weight of the complaint early-warning word segmentation consistent in the complaint early-warning text library;
and when no effective word consistent with the effective word of the effective word group exists in the complaint early warning text library, the word weight of the effective word in the effective word group is zero.
The second aspect of the present invention provides a complaint early warning method, including:
obtaining a chat corpus of a user in a communication process between a service person and the user;
preprocessing the chat corpus to obtain effective phrases in the chat corpus, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing;
comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts;
calculating a time interval of the user speaking for the last time each time, counting the times that the time interval is lower than a preset time interval, and calculating a second score value of the user based on the times;
calculating a total score value according to the first score value and the second score value;
judging whether the total score value is larger than a preset threshold value or not;
if so, determining that the user is a high complaint risk user, and acquiring user data of the high complaint risk user;
and determining a solution aiming at the user with the high complaint risk according to the user data, and early warning the service personnel, wherein the solution is used for being referred by the service personnel after early warning.
Optionally, in a first implementation manner of the second aspect of the present invention, before the obtaining of the chat corpus of the user in the process of communication between the service personnel and the user, the method further includes:
acquiring user data of a complained user, and extracting user characteristics from the user data of the complained user;
forming a feature vector of each complained user according to the user features;
clustering the complained users based on the distance between the feature vectors to obtain more than one user cluster;
and respectively analyzing the principal component characteristics aiming at the user clusters to determine the solution of each user cluster.
A third aspect of the present invention provides a high complaint risk user identification device, including:
the first acquisition module is used for acquiring the chat linguistic data of the user in the communication process between the business personnel and the user;
the first processing module is used for preprocessing the chat corpus to obtain effective phrases in the chat corpus;
the first scoring module is used for comparing the effective word group with a preset complaint early warning text library, calculating a first scoring value of a user, calculating a time interval between the user and the last speech when the user speaks every time, counting the times of the time interval being lower than the preset time interval, and calculating a second scoring value of the user based on the times;
the first calculating module is used for calculating a total score value according to the first score value and the second score value;
the first judgment module is used for judging whether the total score value is larger than a preset threshold value or not;
and the first determining module is used for determining that the user is a high complaint risk user when the total score value is determined to be greater than the preset threshold value.
Optionally, in a first implementation manner of the third aspect of the present invention, the first scoring module includes:
the obtaining unit is used for obtaining the word weight of the complaint early warning word segmentation;
the determining unit is used for determining the word weight of the effective word in the effective word group according to the word weight of the complaint early warning word segmentation;
the similarity unit is used for determining the similarity between the effective word group and each complaint early warning text in the complaint early warning text library according to the word weight of the complaint early warning segmentation word and the word weight of the effective word;
and the scoring unit is used for taking the maximum value of the similarity in the similarities as the first scoring value.
Optionally, in a second implementation manner of the third aspect of the present invention, the determining unit is specifically configured to:
matching effective words in the effective word groups with complaint early warning participles in the complaint early warning text library;
when complaint early-warning participles consistent with effective words in the effective phrases exist in the complaint early-warning text base, word weights of the complaint early-warning participles consistent with the effective words in the complaint early-warning text base are obtained;
obtaining the word weight of the effective words in the effective word groups according to the word weight of the complaint early-warning word segmentation consistent in the complaint early-warning text library;
and when no effective word consistent with the effective word of the effective word group exists in the complaint early warning text library, the word weight of the effective word in the effective word group is zero.
Optionally, in a third implementation manner of the third aspect of the present invention, the first scoring module is further specifically configured to:
taking the current speaking time of the user as the starting time, and calculating the speaking times of the user in a preset time period before the starting time;
calculating the average speaking time interval of the user according to the preset time period and the speaking times;
and obtaining the second score value according to the average speaking time interval and a preset second score value scoring standard.
A fourth aspect of the present invention provides a complaint warning device, including:
the second acquisition module is used for acquiring the chat linguistic data of the user in the communication process between the business personnel and the user;
the second processing module is used for preprocessing the chat corpus to obtain effective phrases in the chat corpus;
the second scoring module is used for comparing the effective word group with a preset complaint early warning text library, calculating a first scoring value of a user, calculating a time interval between the effective word group and the previous speech when the user speaks each time, counting the times of the time interval being lower than the preset time interval, and calculating a second scoring value of the user based on the times;
the second calculation module is used for calculating a total score value according to the first score value and the second score value;
a second judging module for judging whether the total score value is larger than a preset threshold value,
the second determining module is used for determining that the user is a high complaint risk user and acquiring user data of the high complaint risk user when the total score value is determined to be larger than the preset threshold value;
and the early warning module is used for determining a solution aiming at the high complaint risk user according to the user data and early warning the service personnel, wherein the solution is used for being referred by the service personnel after early warning.
Optionally, in a first implementation manner of the fourth aspect of the present invention, the complaint early warning apparatus further includes a user analysis module, where the user analysis module is specifically configured to:
acquiring user data of a complained user, and extracting user characteristics from the user data of the complained user;
forming a feature vector of each complained user according to the user features;
clustering the complained users based on the distance between the feature vectors to obtain more than one user cluster;
and respectively analyzing the principal component characteristics aiming at the user clusters to determine the solution of each user cluster.
A fifth aspect of the present invention provides an electronic device, comprising: a memory having instructions stored therein and at least one processor, the memory and the at least one processor interconnected by a line; the at least one processor invoking the instructions in the memory to cause the electronic device to perform the high complaint risk user identification method described above;
alternatively, the first and second electrodes may be,
the at least one processor invokes the instructions in the memory to cause the electronic device to perform the complaint warning method described above.
A sixth aspect of the present invention provides a computer-readable storage medium having stored therein a computer program which, when run on a computer, causes the computer to execute the above-described high complaint risk user identification method;
alternatively, the first and second electrodes may be,
the computer-readable storage medium has stored therein a computer program which, when run on a computer, causes the computer to execute the complaint warning method described above.
The invention provides a high complaint risk user identification method, a complaint early warning method and related equipment, wherein the high complaint risk user identification method comprises the following steps: obtaining a chat corpus of a user in a communication process between a service person and the user; preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing; comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts; calculating a time interval of each speaking of the user relative to the last speaking, counting the times of the time interval being lower than a preset time interval, and calculating a second score value of the user based on the times; calculating a total score value according to the first score value and the second score value; judging whether the total score value is larger than a preset threshold value or not; and if so, determining that the user is a high complaint risk user. By the method, the emotion of the user can be monitored in real time, when the user is identified to be a high-risk complaint user at present, business personnel can control the risk of the complaint of the user in the process by adjusting the dialect, the number of the complaints is reduced from the result, and the satisfaction degree of the user is improved.
Drawings
FIG. 1 is a schematic diagram of a first embodiment of a high complaint risk user identification method in an embodiment of the invention;
FIG. 2 is a schematic diagram of a second embodiment of a high complaint risk user identification method according to an embodiment of the invention;
FIG. 3 is a diagram of a third embodiment of a high complaint risk user identification method according to an embodiment of the invention;
FIG. 4 is a schematic diagram of an embodiment of a complaint warning method according to an embodiment of the invention;
FIG. 5 is a schematic diagram of an embodiment of a high complaint risk user identification device in an embodiment of the invention;
FIG. 6 is a schematic diagram of another embodiment of a high complaint risk user identification device in an embodiment of the invention;
FIG. 7 is a schematic diagram of an embodiment of a complaint warning device according to an embodiment of the invention;
fig. 8 is a schematic diagram of an embodiment of an electronic device in an embodiment of the invention.
Detailed Description
The embodiment of the invention provides a method for identifying a user with high complaint risk, which comprises the following specific steps: obtaining a chat corpus of a user in a communication process between a service person and the user; preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing; comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts; calculating a time interval of each speaking of the user relative to the last speaking, counting the times of the time interval being lower than a preset time interval, and calculating a second score value of the user based on the times; calculating a total score value according to the first score value and the second score value; judging whether the total score value is larger than a preset threshold value or not; and if so, determining that the user is a high complaint risk user. By the method, the emotion of the user can be monitored in real time, when the user is identified to be a high-risk complaint user at present, business personnel can control the risk of the complaint of the user in the process by adjusting the dialect, the number of the complaints is reduced from the result, and the satisfaction degree of the user is improved.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," or "having," and any variations thereof, are intended to cover non-exclusive inclusions, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
For convenience of understanding, a specific flow of the embodiment of the present invention is described below, and referring to fig. 1, an embodiment of the method for identifying a user with high complaint risk in the embodiment of the present invention includes:
101. obtaining a chat corpus of a user in a communication process between a service person and the user;
it should be emphasized that, in order to ensure the privacy and security of the user and the service personnel during the communication process, the chat corpus can be stored in the nodes of a block chain.
At this step, the communication of the business person with the user may be through a client or through a web page provided by the business person to the user, wherein, through the web page, when the user logs in the web page, the user can start to monitor the page communication content to obtain the chat content between the user and the service personnel, when the user logs in the client, the chat content can be monitored and obtained by adopting a special acquisition tool of the Hook technology, after the chat contents are obtained by the two methods, the chat contents are divided into the chat contents sent by the user and the chat contents sent by the service personnel, the chat contents sent by the user are periodically and automatically generated into HTML files, and then, the file content is analyzed and stored in an Elasticissearch to be used as a chat corpus for subsequent early warning, and the chat content sent by service personnel is stored in a database for subsequent complaint and responsibility pursuit. The mode of automatically generating the HTML file by the chat content of the user is mainly to keep consistent with the format of a general website and facilitate the integration with the system.
In practical application, the chat corpus may also be a natural language text sent by a user when a service person performs communication interaction with the user through a communication tool, where the communication tool may be an instant communication tool such as a QQ, a wechat, an enterprise app for communication between an enterprise and the user, and the purpose of obtaining the chat corpus is to subsequently identify an emotion included in a natural language expressed by the user. Taking an application scene as an example, when two people chat with WeChat, both sides can identify the emotion of the other side from the natural language sent by the other side, and when the emotion of the user is identified as impatience and anger by text interaction, for example, for a customer service robot, the emotion of the user is sent out or manually processed.
102. Preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing;
in this embodiment, after obtaining the chat corpus, the chat corpus needs to be preprocessed, where the preprocessing mainly includes word segmentation processing and deletion of stop words in a segmentation group obtained after the word segmentation processing based on a preset stop word library, and the rest are effective words, and effective words are used to jointly construct an effective word group, which is intended to reduce the computation load when subsequently comparing with a preset emotion segmentation group.
In this embodiment, the chat corpus is mainly subjected to word segmentation processing through a crust segmentation method, the crust segmentation method is a crust segmentation module of Python, and the method supports three segmentation modes, namely an accurate mode, a full mode and a search engine mode. The invention adopts an accurate word segmentation mode with a part-of-speech tagging function, so that the deletion of stop words can be conveniently carried out subsequently, for example, for the analysis result that the user can contact your customers for a plurality of times today but still does not effectively respond to the user, i feel angry, namely the analysis result is formed through word segmentation processing and part-of-speech tagging, namely the analysis result comprises that the user can contact your for a plurality of times per n, but/c still/p does not contact your/n, the user can do/v effectively/ad, the user can respond to the user for a plurality of times per m, but/c still does not have/ad pair/pmy/n, the user can do/v effectively/ad, the user can feel/v/ad.
The stop word dictionary can be constructed according to parts of speech, such as data of digital words, quantifier words, pronouns, adverbs, prepositions, conjunctions, auxiliary words, vocabularies and punctuations, in the process of preceding word segmentation, part of speech tagging can be carried out on the chat corpus through an accurate mode of ending part of speech notation, when stop words in the chat corpus are cleared by using a stop word library, the removal of the parts of speech can be directly carried out on the basis of the preceding part of speech tagging, and meanwhile, the number of the stop words in the stop word dictionary can be increased according to different requirements.
103. Comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts;
in practical applications, the preset complaint early warning text base integrates some languages which represent high complaint risks of the user and are met by daily business personnel in the communication process of the user into a database as texts, for example, "i feel very bad for my claims", and also can express certain words of emotion in daily life, for example, "i are angry now", and the complaint early warning words can be added and deleted appropriately in the actual use process. Meanwhile, the complaint early-warning text library also preprocesses the complaint early-warning text in the same way as the chat corpus.
In this embodiment, effective words are paired with a preset complaint early warning text base, where the complaint early warning text base includes a complaint early warning text, the complaint early warning text base further includes complaint early warning participles left after performing word segmentation and stop word deletion on the complaint early warning text, similarity between the effective word groups and the complaint early warning text base is calculated through pairing, and when all effective words in the effective word groups are unsuccessfully paired with the complaint early warning text base, the similarity is zero.
104. Calculating a time interval between each speaking time of the user and the last speaking time, counting the times of the time interval being lower than a preset time interval, and calculating a second score value of the user based on the times;
in this embodiment, when the speaking interval of the user is calculated, it is intended that when the user frequently speaks, the current emotion of the user is not good, the problem cannot be solved, so that the user frequently speaks, and hopes that the user can pay attention to the problem, and when the problem is solved, the user does not pay attention to communication with service personnel, so that the communication interval is long or the communication is directly ended, which represents that the current complaint risk is low, so that by calculating the speaking interval of the user, whether the user is currently in a state with a high complaint risk can be determined from another dimension.
In this embodiment, the manner of calculating the user speaking time interval is to use the current time as the starting time, calculate the number of times that the user speaks in a preset time period, that is, calculate an average interval that the user speaks in a certain time period, for example, the preset time period is 1 minute, take the current speaking of the user as the starting time, take the last speaking time before one minute as the ending time, calculate the number of speaking times of the minute, find an average interval of speaking each time in one minute according to the number of speaking times, when the average interval is smaller than the preset time interval, count as one time, and calculate the second average value of the user according to the number of times that the average interval is greater than the preset time interval.
105. Calculating a total score value according to the first score value and the second score value;
106. judging whether the total score value is larger than a preset threshold value or not;
107. and if so, determining that the user is a high complaint risk user.
In this embodiment, the first score value and the second score value are taken as two dimensions, and the total score value is calculated by predetermining the weights of the two dimensions, wherein the calculation formula is as follows:
a×x+b×y
wherein a is the weight of the first scoring value, b is the weight of the second scoring value, x is the first scoring value, and y is the second scoring value.
In practical application, a preset threshold value can be directly and respectively set for the first score value and the second score value, and when both the score values are larger than the preset threshold value, the user is directly determined as a high complaint risk complaint user.
The embodiment of the invention provides a high complaint risk user identification method, which comprises the steps of obtaining chat linguistic data of a user in the communication process between business personnel and the user; preprocessing the chat corpus to obtain effective phrases in the chat corpus, comparing the effective phrases with a preset complaint early warning text library, calculating a first score value of a user, calculating a time interval of each time that the user speaks relative to the last speaking, counting the times that the time interval is lower than the preset time interval, and calculating a second score value of the user based on the times; calculating a total score value according to the first score value and the second score value; judging whether the total score value is larger than a preset threshold value or not; and if so, determining that the user is a high complaint risk user. By the method, the emotion of the user can be monitored in real time, when the user is identified to be a high-risk complaint user at present, business personnel can control the risk of the complaint of the user in the process by adjusting the dialect, the number of the complaints is reduced from the result, and the satisfaction degree of the user is improved.
Referring to fig. 2, a second embodiment of the method for identifying a user with high complaint risk according to the embodiment of the present invention includes:
201. obtaining chat linguistic data of a user in the communication process between a service person and the user, and preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data;
202. acquiring the word weight of the complaint early-warning word segmentation;
further, before the obtaining of the word weight of the complaint early warning segmented word, the method further includes:
determining the number of the complaint early warning texts and the number of the complaint early warning word segments;
determining the number of complaint early warning texts to which the complaint early warning participles belong and the frequency of occurrence of each complaint early warning participle in the complaint early warning words;
and determining the word weight of each complaint early-warning word according to the number of the complaint early-warning texts, the number of the complaint early-warning participles, the number of the complaint early-warning texts to which the complaint early-warning participles belong and the occurrence frequency of each complaint early-warning participle in the complaint early-warning words.
In this step, the manner of determining the number of the complaint early warning texts to which the complaint early warning participles belong and the number of times that each complaint early warning participle appears in the complaint early warning words is that the system extracts a certain complaint early warning participle from the complaint early warning text library, and then determines how many complaint early warning texts in the complaint early warning text library have the complaint early warning participle, so as to determine the number of the complaint early warning texts to which the complaint early warning participles belong. And when the system detects that repeated complaint early-warning word segmentation exists in the complaint early-warning text library, the number of texts to which the word belongs is determined by extracting the word once. Next, the system determines the number of times each complaint warning segmented word repeatedly appears in the complaint warning words to determine the number of times each complaint warning word appears in the words of the matching library.
In this step, the system calculates the word weight of a certain complaint early-warning word in the complaint early-warning text library according to the number of the complaint early-warning texts, the number of the complaint early-warning participles, the number of the complaint early-warning texts to which the complaint early-warning participles belong, and the occurrence frequency of each complaint early-warning participle in the complaint early-warning words. And calculating the word weights of other complaint early-warning participles in the complaint early-warning text base according to the same mode until the complaint early-warning participles with the unwatched word weights do not exist in the complaint early-warning text base, and determining to finish the word weight calculation operation of all the complaint early-warning participles in the complaint early-warning text base. And after the system calculates the word weight of one complaint early-warning word, the system can detect the same word as the complaint early-warning word in the matching library and endow the word weight of the complaint early-warning word to the same word in the matching library.
In this step, the calculation formula of the word weight of each complaint early-warning participle is as follows:
Figure BDA0002597815110000101
wherein Q isiIs the word weight of the ith complaint early-warning participle, i is the serial number of the complaint early-warning participle in the complaint early-warning text library, niIndicates the number of occurrences of the ith word in the complaint warning text corpus,
Figure BDA0002597815110000102
the method comprises the steps of obtaining a number of all complaint early warning participles in a complaint early warning text base, | D | the number of texts in the complaint early warning text base, | T is the complaint early warning participle in the complaint early warning text base, | j is the complaint early warning text in the complaint early warning text base, j: t is the complaint early warning text j with the complaint early warning participle t, j: t belongs to DjSet d of complaint early warning texts j representing complaint early warning participles t in complaint early warning text libraryj，|j:t∈djAnd | represents the number of the complaint early warning texts j with the complaint early warning participles t in the complaint early warning text library.
203. Matching effective words in the effective word groups with complaint early warning participles in a complaint early warning text library;
204. when complaint early-warning participles consistent with effective words in effective phrases exist in a complaint early-warning text base, word weights of the complaint early-warning participles consistent with the effective words in the complaint early-warning text base are obtained;
205. obtaining the word weight of effective words in effective phrases according to the word weight of the consistent complaint early warning word segmentation in the complaint early warning text library;
206. when no effective word consistent with the effective word of the effective word group exists in the complaint early warning text library, the word weight of the effective word in the effective word group is zero;
in this embodiment, when the system detects that a complaint early-warning word segment identical to an effective word extracted from an effective word group exists in the complaint early-warning text base, a word weight of the identical complaint early-warning word segment in the complaint early-warning text base is obtained, and the word weight of the complaint early-warning word segment is assigned to the effective word extracted from the effective word group, so as to obtain a word weight of the effective word extracted from the effective word group.
In this embodiment, when a word consistent with a word extracted from a text to be matched is not matched in the matching library, it is determined that the word does not exist in the matching library, and the word weight of the word is 0. By the method, the word weight of each effective word in the effective word group can be quickly and simply determined without calculating the word weight of the effective word in the effective word group when the chat corpus is obtained every time.
207. Determining the similarity between an effective word group and each complaint early warning text in a complaint early warning text library according to the word weight of the complaint early warning participle and the word weight of the effective word, and taking the maximum value of the similarity as a first score value;
in this embodiment, because the frequency of occurrence of a certain complaint early warning word in different complaint early warning texts is very high, for example, for two complaint early warning texts, "because you do not solve my problem, i feel angry now," "business personnel handling my case do not contact me in time, i feel angry," and in both complaint early warning texts, the complaint early warning word "angry" appears, so the word weight of "angry" is higher than that of other complaint early warning words, and the complaint early warning word with higher word weight represents that the complaint risk of the user is relatively higher.
In this embodiment, after determining that the valid phrase has the similarity of each complaint early warning text, the similarity of each complaint early warning text is ranked, and the highest value of the similarity is taken as the first score value.
In this embodiment, the calculation formula of the similarity is as follows:
Figure BDA0002597815110000111
wherein S isi,jAnd representing the similarity between the effective phrase i and the jth complaint early warning text in the complaint early warning text library. i means a valid phrase, NiRepresenting the number of effective words in an effective word group i, j representing the jth complaint early warning text in a complaint early warning text library, and NjAnd the word number after the jth complaint early warning text in the complaint early warning text library is participled is represented. N is a radical ofcAnd the number of the same words between the effective word group i and the jth complaint early warning text in the complaint early warning text library is represented.
Figure BDA0002597815110000112
Representing the sum of the word weights of the effective word groups and the same words in the complaint early warning text extracted from the complaint early warning text library,
Figure BDA0002597815110000113
representing the sum of the word weights of all words in the valid phrase,
Figure BDA0002597815110000114
and the sum of the word weights of the words of the complaint early warning texts extracted from the complaint early warning text library is represented.
208. Calculating a time interval of each speaking of the user relative to the last speaking, counting the times of the time interval being lower than a preset time interval, and calculating a second score value of the user based on the times;
209. calculating a total score value according to the first score value and the second score value;
210. judging whether the total score value is larger than a preset threshold value or not;
211. and if so, determining that the user is a high complaint risk user.
On the basis of the previous embodiment, the obtaining process of the first score value and the determining process of the word weight of the complaint early-warning segmented word are described in detail in this embodiment. The first scoring value is obtained by mainly obtaining the word weight of the effective word group, the similarity between each effective word and the complaint early warning text is determined, the emotion contained in the current chat corpus of the user can be accurately determined, and then the business personnel are warned.
Referring to fig. 3, a third embodiment of the method for identifying a user with high complaint risk according to the embodiment of the present invention includes:
301. obtaining a chat corpus of a user in a communication process between a service person and the user;
302. preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data;
303. comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user;
304. taking the current speaking time of the user as the starting time, and calculating the speaking times of the user in a preset time period before the starting time;
305. calculating the average speaking time interval of the user according to the preset time period and the speaking times;
306. obtaining a second score value according to the average speaking time interval and a preset second score value scoring standard;
in this embodiment, the preset second score value scoring criterion may be obtained by dividing a time interval, for example, 60 minutes for 10 to 12 seconds, 70 minutes for 10 to 8 seconds, and the specific scoring criterion may be adjusted according to actual requirements.
307. Calculating a total score value according to the first score value and the second score value;
further, after the step, in practical application, each time after the business personnel and the user communicate with each other, the highest first score value and the highest second score value of the chat corpus in the communication process are obtained, the magnitudes of the two highest score values are judged, the score value with the highest score value is taken as a high score value, the difference between the two highest score values is calculated, if the proportion of the difference in the high score value is larger than a preset proportion threshold, for example, the difference in the high score value occupies 40% of the high score value, the weight of the high score value is increased by 1%, and in practical application, the weight can be adjusted by setting a plurality of proportion threshold values, for example, when the difference in the high score value occupies 50% of the high score value, the weight of the high score value is increased by 2%.
308. Judging whether the total score value is larger than a preset threshold value or not;
309. and if so, determining that the user is a high complaint risk user.
This embodiment describes the calculation process of the second score value in detail based on the above embodiment. The frequency of speaking of the user can be accurately obtained through calculation of the second score, the higher the frequency is, the less full the emotion of the user is represented, the higher the probability of the customer complaint is, and the real-time calculation can be used for detecting the user in real time.
Referring to fig. 4, an embodiment of a complaint warning method according to an embodiment of the present invention includes:
401. acquiring user data of a complaint user, and extracting user characteristics from the user data of the complaint user;
402. forming a feature vector of each complained user according to the user features;
403. clustering the complained users based on the distance between the feature vectors to obtain more than one user cluster;
404. analyzing the principal component characteristics respectively aiming at the user clusters to determine a solution of each user cluster;
in this embodiment, the specific type of the extracted user feature may be preset, because the complaint performed by the user is often caused by that the complaint in some aspect is not satisfied, and the staff may have different solutions for different complaints of the user after being complained, so that the analysis of the user feature may be performed after the solution of the previous complaint user is used, for example, in the complaint of a claim case, different types may be divided according to different conditions by obtaining acceptance information, distribution information, processing information, and the like of the case included in the user data in the process of performing the claim by the user. After the user features are obtained, the feature vectors are obtained after the user features are subjected to normalization, space mapping and other processing.
In this embodiment, the complained users are clustered based on the distance between feature vectors, for example, the euclidean distance, where the obtained users in the same cluster all have similar solutions, the clustering method includes, but is not limited to, K-means clustering, clustering based on a classification model, and the like, so that the principal component feature may be understood as a main user feature in the user cluster, and a physical meaning corresponding to the main user feature in one user cluster may often represent a main solution corresponding to the user in the user cluster. Based on the theory, the principal component characteristics in each user cluster can be obtained through a characteristic analysis method.
405. Obtaining a chat corpus of a user in a communication process between a service person and the user;
406. preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data;
407. comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user;
408. calculating a time interval of each speaking of the user relative to the last speaking, counting the times of the time interval being lower than a preset time interval, and calculating a second score value of the user based on the times;
409. calculating a total score value according to the first score value and the second score value;
410. judging whether the total score value is larger than a preset threshold value or not;
411. if yes, determining that the user is a high complaint risk user;
412. acquiring user data of users with high complaint risks;
413. and determining a solution aiming at the user with high complaint risk according to the user data, and early warning business personnel.
In this embodiment, data analysis is performed on user data corresponding to a user who previously complained, principal component characteristics of various complained users are analyzed to determine main solutions of various complained users, user characteristics are extracted according to the user data of the user with a high complaint risk, a characteristic vector of the user characteristics is formed, a solution which is closest to the current user with the high complaint risk and a previous complained user is determined according to the characteristic vector of the principal component characteristics and the characteristic vector of the user characteristics, after a corresponding solution is obtained, the corresponding solution and an early warning signal are simultaneously sent to business personnel, and after the business personnel receive the early warning signal, the business personnel refer to the solution to soothe the user with the high complaint risk.
In this embodiment, on the basis of the previous embodiment, measures taken after the user is identified as a high complaint risk user are added, when a service person communicates with the user, and after the server identifies that the user is a high complaint risk user, the server acquires user data corresponding to the user, analyzes a user cluster where the user is located according to the user data, and provides a solution corresponding to the user cluster of the service person, so that the service person can conveniently adjust a speech technology and flexibly deal with the complaint risk.
In the above description of the method for identifying a user with high complaint risk in the embodiment of the present invention, referring to fig. 5, a device for identifying a user with high complaint risk in the embodiment of the present invention is described below, where one embodiment of the device for identifying a user with high complaint risk in the embodiment of the present invention includes:
a first obtaining module 501, configured to obtain a chat corpus of a user in a process of communication between a service person and the user;
a first processing module 502, configured to pre-process the chat corpus to obtain an effective phrase in the chat corpus;
the first scoring module 503 is configured to compare the effective phrase with a preset complaint early warning text library, calculate a first score of a user, calculate a time interval between each utterance of the user and a previous utterance, count the number of times that the time interval is lower than the preset time interval, and calculate a second score of the user based on the number of times;
a first calculating module 504, configured to calculate a total score value according to the first score value and the second score value;
a first determining module 505, configured to determine whether the total score value is greater than a preset threshold;
a first determining module 506, configured to determine that the user is a high complaint risk user when it is determined that the total score value is greater than the preset threshold.
It should be emphasized that, in order to ensure the privacy and security of the user and the service personnel during the communication process, the chat corpus can be stored in the nodes of a block chain.
The embodiment of the invention provides a method for identifying a user with high complaint risk, which comprises the following specific steps: the method comprises the steps of obtaining chat linguistic data of service personnel and users, preprocessing the chat linguistic data to obtain effective word phrases, matching the effective word phrases with a preset complaint early warning text library, calculating a current first score value of the users, calculating a second score value of the users according to the speaking time interval of the users, calculating a total score value according to the first score value, the second score value and corresponding weight, determining that the users are high complaint risk users, and determining a solution for the high complaint risk users for reference of the service personnel while early warning the service personnel according to the user data of the high complaint risk users. By the method, the emotion of the user can be monitored in real time, business personnel can adjust the dialect conveniently, the risk of complaints of the user is controlled in the process, the number of the complaints is reduced from the result, and the satisfaction degree of the user is improved.
Referring to fig. 6, another embodiment of the high complaint risk user identification device in the embodiment of the present invention includes:
a first obtaining module 501, configured to obtain a chat corpus of a user in a process of communication between a service person and the user;
a first processing module 502, configured to pre-process the chat corpus to obtain an effective phrase in the chat corpus;
the first scoring module 503 is configured to compare the effective phrase with a preset complaint early warning text library, calculate a first score of a user, calculate a time interval between each utterance of the user and a previous utterance, count the number of times that the time interval is lower than the preset time interval, and calculate a second score of the user based on the number of times;
a first calculating module 504, configured to calculate a total score value according to the first score value and the second score value;
a first determining module 505, configured to determine whether the total score value is greater than a preset threshold;
a first determining module 506, configured to determine that the user is a high complaint risk user when it is determined that the total score value is greater than the preset threshold.
Wherein the first scoring module 503 comprises:
an obtaining unit 5031, configured to obtain a word weight of the complaint early warning word segmentation;
a determining unit 5032, configured to determine word weights of effective words in the effective word groups according to the word weights of the complaint early warning segmented words;
a similarity unit 5033, configured to determine, according to the word weight of the complaint early-warning segmented word and the word weight of the effective word, a similarity between the effective word group and each complaint early-warning text in the complaint early-warning text base;
a scoring unit 5034 configured to take a maximum value of the similarities as the first score value.
Optionally, the determining unit 5033 is specifically configured to:
matching effective words in the effective word groups with complaint early warning participles in the complaint early warning text library;
when complaint early-warning participles consistent with effective words in the effective phrases exist in the complaint early-warning text base, word weights of the complaint early-warning participles consistent with the effective words in the complaint early-warning text base are obtained;
obtaining the word weight of the effective words in the effective word groups according to the word weight of the complaint early-warning word segmentation consistent in the complaint early-warning text library;
and when no effective word consistent with the effective word of the effective word group exists in the complaint early warning text library, the word weight of the effective word in the effective word group is zero.
Optionally, the first score value module 503 may be further specifically configured to:
taking the current speaking time of the user as the starting time, and calculating the speaking times of the user in a preset time period before the starting time;
calculating the average speaking time interval of the user according to the preset time period and the speaking times;
and obtaining the second score value according to the average speaking time interval and a preset second score value scoring standard.
The embodiment of the invention provides a device and a method for identifying a user with high complaint risk, which are operated, wherein the specific implementation process comprises the following steps: obtaining a chat corpus of a user in a communication process between a service person and the user; preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing; comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts; calculating a time interval of each speaking of the user relative to the last speaking, counting the times of the time interval being lower than a preset time interval, and calculating a second score value of the user based on the times; calculating a total score value according to the first score value and the second score value; judging whether the total score value is larger than a preset threshold value or not; and if so, determining that the user is a high complaint risk user. By the method, the emotion of the user can be monitored in real time, when the user is identified to be a high-risk complaint user at present, business personnel can control the risk of the complaint of the user in the process by adjusting the dialect, the number of the complaints is reduced from the result, and the satisfaction degree of the user is improved.
Referring to fig. 7, a complaint early warning apparatus in an embodiment of the present invention is described below, where an embodiment of the complaint early warning apparatus in the embodiment of the present invention includes:
a second obtaining module 701, configured to obtain a chat corpus of the user in a process of communication between a service person and the user;
a second processing module 702, configured to pre-process the chat corpus to obtain an effective phrase in the chat corpus;
the second scoring module 703 is configured to compare the effective phrase with a preset complaint early warning text library, calculate a first score of a user, calculate a time interval between each utterance of the user and the last utterance, count the number of times that the time interval is lower than the preset time interval, and calculate a second score of the user based on the number of times;
a second calculating module 704, configured to calculate a total score value according to the first score value and the second score value;
a second judging module 705 for judging whether the total score value is larger than a preset threshold value,
a second determining module 706, configured to determine that the user is a high complaint risk user and obtain user data of the high complaint risk user when it is determined that the total score value is greater than the preset threshold;
and an early warning module 707, configured to determine a solution for the user with the high complaint risk according to the user data, and perform early warning on the service staff, where the solution is used for the service staff to refer to after early warning.
The complaint early warning apparatus further includes a user analysis module 708, where the user analysis module 708 is specifically configured to:
acquiring user data of a complained user, and extracting user characteristics from the user data of the complained user;
forming a feature vector of each complained user according to the user features;
clustering the complained users based on the distance between the feature vectors to obtain more than one user cluster;
and respectively analyzing the principal component characteristics aiming at the user clusters to determine the solution of each user cluster.
The embodiment of the invention provides a complaint early warning device and method, which are operated, and the specific implementation process comprises the following steps: obtaining a chat corpus of a user in a communication process between a service person and the user; preprocessing the chat linguistic data to obtain effective phrases in the chat linguistic data, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing; comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts; calculating a time interval of each speaking of the user relative to the last speaking, counting the times of the time interval being lower than a preset time interval, and calculating a second score value of the user based on the times; calculating a total score value according to the first score value and the second score value; judging whether the total score value is larger than a preset threshold value or not; and if so, determining that the user is the user with high complaint risk, acquiring user data of the user with high complaint risk, and analyzing the solution based on the user data. By the method, the emotion of the user can be monitored in real time, when the user is identified to be a high-risk complaint user at present, business personnel can control the risk of the complaint of the user in the process by adjusting the dialect, the number of the complaints is reduced from the result, and the satisfaction degree of the user is improved.
Fig. 5, fig. 6, and fig. 7 describe the high complaint risk user identification device and the complaint early warning device in the embodiment of the present invention in detail from the perspective of the modular functional entity, and the electronic device in the embodiment of the present invention is described in detail from the perspective of hardware processing.
Fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, where the electronic device 800 may have a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 810 (e.g., one or more processors) and a memory 820, and one or more storage media 830 (e.g., one or more mass storage devices) storing an application 833 or data 832. Memory 820 and storage medium 830 may be, among other things, transient or persistent storage. The program stored in the storage medium 830 may include one or more modules (not shown), each of which may include a sequence of instructions operating on the electronic device 800. Still further, the processor 810 may be configured to communicate with the storage medium 830, and execute a series of instruction operations in the storage medium 830 on the electronic device 800 to implement the steps of the high complaint risk user identification method or the steps of the complaint warning method provided by the above embodiments.
The electronic device 800 may also include one or more power supplies 840, one or more wired or wireless network interfaces 850, one or more input-output interfaces 860, and/or one or more operating systems 831, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, and so forth. Those skilled in the art will appreciate that the electronic device configuration shown in fig. 8 does not constitute a limitation of the electronic device and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium, and which may also be a volatile computer-readable storage medium, having stored therein a computer program (i.e., instructions), for causing a computer to perform the steps of the high complaint risk user identification method when the computer program runs on a computer, or for causing a computer to perform the steps of the complaint warning method when the computer program runs on a computer, and optionally, the computer program is executed by a processor on a computer.
The block chain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which 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) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. A high complaint risk user identification method is characterized by comprising the following steps:
obtaining a chat corpus of a user in a communication process between a service person and the user;
preprocessing the chat corpus to obtain effective phrases in the chat corpus, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing;
comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts;
calculating a time interval of the user speaking for the last time each time, counting the times that the time interval is lower than a preset time interval, and calculating a second score value of the user based on the times;
calculating a total score value according to the first score value and the second score value;
judging whether the total score value is larger than a preset threshold value or not;
and if so, determining that the user is a high complaint risk user.
2. The method for identifying the user with high complaint risk according to claim 1, wherein the step of comparing the effective phrase with a preset complaint early warning text library and calculating the first score value of the user comprises:
acquiring the word weight of the complaint early warning word segmentation;
determining the word weight of the effective words in the effective word group according to the word weight of the complaint early warning word segmentation;
determining the similarity between the effective word group and each complaint early warning text in the complaint early warning text library according to the word weight of the complaint early warning segmentation word and the word weight of the effective word;
and taking the maximum value of the similarity in the similarities as the first scoring value.
3. The method for identifying the user with high complaint risk according to claim 2, wherein the determining the word weight of the effective word in the effective word group according to the word weight of the complaint early warning segmented word comprises:
matching effective words in the effective word groups with complaint early warning participles in the complaint early warning text library;
when complaint early-warning participles consistent with effective words in the effective phrases exist in the complaint early-warning text base, word weights of the complaint early-warning participles consistent with the effective words in the complaint early-warning text base are obtained;
obtaining the word weight of the effective words in the effective word groups according to the word weight of the complaint early-warning word segmentation consistent in the complaint early-warning text library;
and when no effective word consistent with the effective word of the effective word group exists in the complaint early warning text library, the word weight of the effective word in the effective word group is zero.
4. The method for identifying a user with a high complaint risk according to claim 1, wherein the calculating a time interval between each utterance made by the user and the last utterance, counting the number of times that the time interval is lower than a preset time interval, and calculating a second score value of the user based on the number of times comprises:
taking the current speaking time of the user as the starting time, and calculating the speaking times of the user in a preset time period before the starting time;
calculating the average speaking time interval of the user according to the preset time period and the speaking times;
and obtaining the second score value according to the average speaking time interval and a preset second score value scoring standard.
5. A complaint early warning method is characterized by comprising the following steps:
obtaining a chat corpus of a user in a communication process between a service person and the user;
preprocessing the chat corpus to obtain effective phrases in the chat corpus, wherein the preprocessing comprises word segmentation processing and word deletion processing, and the effective phrases are a set of residual words after the word segmentation processing and the word deletion processing;
comparing the effective phrases with a preset complaint early warning text library, and calculating a first score value of a user, wherein the complaint early warning text library comprises complaint early warning texts and complaint early warning word segments obtained by preprocessing the complaint early warning texts;
calculating a time interval of the user speaking for the last time each time, counting the times that the time interval is lower than a preset time interval, and calculating a second score value of the user based on the times;
calculating a total score value according to the first score value and the second score value;
judging whether the total score value is larger than a preset threshold value or not;
if so, determining that the user is a high complaint risk user, and acquiring user data of the high complaint risk user;
and determining a solution aiming at the user with the high complaint risk according to the user data, and early warning the service personnel, wherein the solution is used for being referred by the service personnel after early warning.
6. The complaint early warning method of claim 5, wherein before the obtaining of the chat corpus of the user in the process of communication between the service personnel and the user, the method further comprises:
acquiring user data of a complained user, and extracting user characteristics from the user data of the complained user;
forming a feature vector of each complained user according to the user features;
clustering the complained users based on the distance between the feature vectors to obtain more than one user cluster;
and respectively analyzing the principal component characteristics aiming at the user clusters to determine the solution of each user cluster.
7. A high complaint risk user identification device, characterized by comprising:
the first acquisition module is used for acquiring the chat linguistic data of the user in the communication process between the business personnel and the user;
the first processing module is used for preprocessing the chat corpus to obtain effective phrases in the chat corpus;
the first scoring module is used for comparing the effective word group with a preset complaint early warning text library, calculating a first scoring value of a user, calculating a time interval between the user and the last speech when the user speaks every time, counting the times of the time interval being lower than the preset time interval, and calculating a second scoring value of the user based on the times;
the first calculating module is used for calculating a total score value according to the first score value and the second score value;
the first judgment module is used for judging whether the total score value is larger than a preset threshold value or not;
and the first determining module is used for determining that the user is a high complaint risk user when the total score value is determined to be greater than the preset threshold value.
8. A complaint early warning device, characterized in that the complaint early warning device includes:
the second acquisition module is used for acquiring the chat linguistic data of the user in the communication process between the business personnel and the user;
the second processing module is used for preprocessing the chat corpus to obtain effective phrases in the chat corpus;
the second scoring module is used for comparing the effective word group with a preset complaint early warning text library, calculating a first scoring value of a user, calculating a time interval between the effective word group and the previous speech when the user speaks each time, counting the times of the time interval being lower than the preset time interval, and calculating a second scoring value of the user based on the times;
the second calculation module is used for calculating a total score value according to the first score value and the second score value;
a second judging module for judging whether the total score value is larger than a preset threshold value,
the second determining module is used for determining that the user is a high complaint risk user and acquiring user data of the high complaint risk user when the total score value is determined to be larger than the preset threshold value;
and the early warning module is used for determining a solution aiming at the high complaint risk user according to the user data and early warning the service personnel, wherein the solution is used for being referred by the service personnel after early warning.
9. An electronic device, characterized in that the electronic device comprises: a memory having instructions stored therein and at least one processor, the memory and the at least one processor interconnected by a line;
the at least one processor invoking the instructions in the memory to cause the electronic device to perform the high complaint risk user identification method of any of claims 1-4;
alternatively, the first and second electrodes may be,
the at least one processor invokes the instructions in the memory to cause the electronic device to perform the complaint warning method of claim 5 or 6.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a high complaint risk user identification method according to any one of claims 1-4;
alternatively, the first and second electrodes may be,
the computer program, when executed by a processor, implements a complaint warning method as claimed in claim 5 or 6.
CN202010714915.4A 2020-07-23 2020-07-23 High-complaint-risk user identification method, complaint early warning method and related equipment Pending CN111859958A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010714915.4A CN111859958A (en) 2020-07-23 2020-07-23 High-complaint-risk user identification method, complaint early warning method and related equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010714915.4A CN111859958A (en) 2020-07-23 2020-07-23 High-complaint-risk user identification method, complaint early warning method and related equipment

Publications (1)

Publication Number Publication Date
CN111859958A true CN111859958A (en) 2020-10-30

Family

ID=72949724

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010714915.4A Pending CN111859958A (en) 2020-07-23 2020-07-23 High-complaint-risk user identification method, complaint early warning method and related equipment

Country Status (1)

Country Link
CN (1) CN111859958A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112163585A (en) * 2020-11-10 2021-01-01 平安普惠企业管理有限公司 Text auditing method and device, computer equipment and storage medium
CN112860876A (en) * 2021-03-31 2021-05-28 中国工商银行股份有限公司 Session auxiliary processing method and device
CN113610399A (en) * 2021-08-09 2021-11-05 广州品唯软件有限公司 Risk monitoring method, system and device for customer service background

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112163585A (en) * 2020-11-10 2021-01-01 平安普惠企业管理有限公司 Text auditing method and device, computer equipment and storage medium
CN112163585B (en) * 2020-11-10 2023-11-10 上海七猫文化传媒有限公司 Text auditing method and device, computer equipment and storage medium
CN112860876A (en) * 2021-03-31 2021-05-28 中国工商银行股份有限公司 Session auxiliary processing method and device
CN113610399A (en) * 2021-08-09 2021-11-05 广州品唯软件有限公司 Risk monitoring method, system and device for customer service background

Similar Documents

Publication Publication Date Title
US10642889B2 (en) Unsupervised automated topic detection, segmentation and labeling of conversations
CN111859958A (en) High-complaint-risk user identification method, complaint early warning method and related equipment
CN109509470B (en) Voice interaction method and device, computer readable storage medium and terminal equipment
CN110069784A (en) A kind of voice quality inspection methods of marking, device, terminal and can storage medium
CN109584865B (en) Application program control method and device, readable storage medium and terminal equipment
CN110176117A (en) A kind of monitoring device and monitoring method of Behavior-based control identification technology
CN112735383A (en) Voice signal processing method, device, equipment and storage medium
WO2017091985A1 (en) Method and device for recognizing stop word
CN112231451B (en) Reference word recovery method and device, conversation robot and storage medium
CN113055537A (en) Voice quality inspection method, device, equipment and storage medium for customer service personnel
CN113990352A (en) User emotion recognition and prediction method, device, equipment and storage medium
CN111523317B (en) Voice quality inspection method and device, electronic equipment and medium
CN114722199A (en) Risk identification method and device based on call recording, computer equipment and medium
CN112562736B (en) Voice data set quality assessment method and device
CN111460114A (en) Retrieval method, device, equipment and computer readable storage medium
CN110796565A (en) Analysis method and analysis system for supervision logs
CN115935076A (en) Travel service information pushing method and system based on artificial intelligence
CN116127010A (en) Intention recognition method, device, electronic equipment and storage medium
CN115392220A (en) Satisfaction degree value determination method and device
CN113590768B (en) Training method and device for text relevance model, question answering method and device
CN111539834A (en) Pre-check method and device
CN117077632B (en) Automatic generation method for information theme
CN113345419B (en) Dialect accent based speech translation method, system and readable storage medium
CN111553600B (en) Big data-based smart city business distribution system
CN116757222A (en) Emotion recognition method, emotion recognition device, computer-readable storage medium and electronic device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination