CN113743124A - Intelligent question-answer exception processing method and device and electronic equipment - Google Patents
Intelligent question-answer exception processing method and device and electronic equipment Download PDFInfo
- Publication number
- CN113743124A CN113743124A CN202110985077.9A CN202110985077A CN113743124A CN 113743124 A CN113743124 A CN 113743124A CN 202110985077 A CN202110985077 A CN 202110985077A CN 113743124 A CN113743124 A CN 113743124A
- Authority
- CN
- China
- Prior art keywords
- question
- sentences
- intelligent
- answering
- abnormal
- 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.)
- Granted
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 9
- 230000002159 abnormal effect Effects 0.000 claims abstract description 60
- 238000000034 method Methods 0.000 claims abstract description 45
- 238000012545 processing Methods 0.000 claims abstract description 30
- 238000007781 pre-processing Methods 0.000 claims abstract description 26
- 238000010606 normalization Methods 0.000 claims description 15
- 238000001914 filtration Methods 0.000 claims description 11
- 238000012937 correction Methods 0.000 claims description 10
- 230000005856 abnormality Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 8
- 238000003058 natural language processing Methods 0.000 abstract description 2
- 238000004458 analytical method Methods 0.000 description 7
- 238000005070 sampling Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000005065 mining Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 241001122315 Polites Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000007943 implant Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- 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
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/242—Dictionaries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
Abstract
The invention discloses a method and a device for processing an intelligent question-answering exception and electronic equipment, and relates to the technical field of natural language processing and intelligent question-answering. The method comprises the steps of obtaining a conversation record of an intelligent customer service and a user, wherein the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by the user and answer sentences sent by the intelligent customer service; preprocessing the question sentences according to a preset preprocessing method; judging whether the preprocessed question sentences are abnormal problems or not according to a preset abnormal problem judgment method; setting a statement pair corresponding to the abnormal problem according to a preset abnormal problem processing rule; and updating the knowledge base of the intelligent customer service according to the statement pair. The intelligent question-answering abnormity processing method, the intelligent question-answering abnormity processing device and the electronic equipment can monitor and analyze abnormal problems and efficiently generate intelligent question-answering sentences.
Description
Technical Field
The invention relates to the technical field of natural language processing and intelligent question answering, in particular to a method and a device for processing an intelligent question answering exception and electronic equipment.
Background
In the traditional service industry, manual customer service is a labor-intensive and highly repetitive job for a full period of time. Therefore, in order to reduce labor cost and improve efficiency, more and more enterprises introduce intelligent customer service capable of automatically replying corresponding reply sentences according to the problems of the users, so that the working pressure of the artificial customer service is relieved to a certain extent, and the accuracy, the normalization and the stability of enterprise service are improved.
In order to ensure that the intelligent customer service can accurately answer the user, a huge knowledge base system needs to be maintained for the intelligent customer service. The knowledge base comprises a large number of standard questions and corresponding answers, the question-answering process of the intelligent customer service of the question-answering system mainly matches the questions of the user with the standard questions in the knowledge base, and if the matching is successful, the answers corresponding to the standard questions are returned. However, the user's questions are never the same, and usually the user will ask a new abnormal question for various reasons, which is not included in the knowledge base and cannot be accurately identified by the intelligent customer service due to the question and answer operation, so that the problem proposed by the user cannot be solved in the first time, which not only reduces the information experience of the user, but also increases the workload of the manual customer service due to the manual operation. Therefore, it is necessary to monitor and push abnormal problems which cannot be solved by intelligent customer service in time and maintain and update the knowledge base in real time.
Therefore, a processing method, an apparatus and an electronic device for monitoring and analyzing abnormal problems to efficiently generate intelligent question-answering abnormalities are needed to solve the above technical problems in the prior art.
Disclosure of Invention
In order to solve at least one problem mentioned in the background art, the invention provides a method and a device for processing an intelligent question-answering exception and an electronic device, which can monitor and analyze the exception problem and efficiently generate an intelligent question-answering sentence.
The embodiment of the invention provides the following specific technical scheme:
a method for processing intelligent question-answering exception, comprising the following steps:
s1: acquiring a conversation record of an intelligent customer service and a user, wherein the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by the user and answer sentences sent by the intelligent customer service;
s3: preprocessing the question sentences according to a preset preprocessing method;
s4: judging whether the preprocessed question sentence is an abnormal question or not according to a preset abnormal question judging method, if so, executing step S5, and if not, ending the processing;
s5: setting a statement pair corresponding to the abnormal problem according to a preset abnormal problem processing rule;
s6: and updating the knowledge base of the intelligent customer service according to the statement pair.
Further, before the step S1, the method further includes the following steps:
s0: triggering a preset log code.
Further, the following steps are also included between the step S1 and the step S3:
s2: storing the acquired session record to a storage medium;
the storage medium comprises a file system, a system memory and Kafka.
Further, the preprocessing of the session record comprises:
and normalizing the session records, and converting the acquired session records into character coded data.
Further, the preprocessing of the session record further comprises:
correcting errors of the session records, namely correcting errors of wrongly written words included in the session records by using a preset error correction rule;
filtering the session records, and filtering the session records;
the method comprises the following steps of (1) removing duplication of question sentences, namely comparing the obtained question sentences with question sentences stored in a knowledge base of the intelligent customer service to remove duplication;
and normalizing the question questions, and normalizing the obtained question questions with similar semantics into standard questions based on a semantic similarity algorithm.
Further, the question normalization includes the following steps:
s31: initialization, including by preset initialization<key，value>Dictionary dit of key-value pairstCounting the frequency of occurrence of question questions, wherein: key is the problem, value is the frequency of occurrence of the problem;
s32: calculating the similarity; semantic similarity calculation is carried out on the newly input question query and each question key in the dictionary by utilizing a semantic similarity calculation method, and the score of the maximum similarity is determinedmaxCorresponding question and is noted as keymax；
S33: judging the attribution of the newly added question; when scoremaxWhen the similarity is larger than or equal to the set similarity threshold, the newly input question query and the question key corresponding to the maximum similarity score are judgedmaxExpressing the same question and keymaxThe corresponding value + 1; when scoremaxIf the similarity is smaller than the set similarity threshold, the newly input query is judged to be a new problem, and the dictionary fact is usedtAdding a corresponding number<key，value>A key-value pair;
s34: determining normalized dictionary ditt。
Further, the step S4 includes the following steps:
s41: obtaining normalized dictionary dit in period ttAnd T-T problem normalization dictionary fact of previous periodt-T；
S42: obtaining the normalized dictionary dit in the period ttNumber qc of any question iitNumber fraction qcritSmooth ring ratio growth rate qsrit；
S43: according to the data information obtained in the step S42, calculating the abnormal growth rate qr of any question i in the period tit：
Wherein: problem normalization dictionary fact of previous period T-Tt-T，qritIs the abnormal growth rate, qcr, of any question i within the period titIs the normalized dictionary dit within the period ttThe ratio of the number of any question i, qsritIs the normalized dictionary dit within the period ttSmooth loop ratio growth rate, qc, of any question iitIs the normalized dictionary dit within the period ttNumber of any question questions i, qci(t-T)Is the normalized dictionary fact in the previous period T-TtThe number of any question i;
s44: based on the preset abnormality determination threshold, the abnormal increase rate qr of any question i obtained in step S43itIf the value exceeds the preset abnormality determination threshold, step S5 is executed, otherwise, the process ends.
Further, the semantic similarity algorithm is a Bert-based text similarity algorithm and/or a WMD-based text similarity algorithm.
The invention also provides an intelligent question-answering abnormity device, which comprises: the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring a conversation record, the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by a user and answer sentences sent by an intelligent customer service;
the storage module is used for storing the acquired session records;
the preprocessing module is used for preprocessing the acquired session records;
the judging module is used for judging whether the obtained question is an abnormal question or not;
the processing module is used for formulating a corresponding answer sentence for the question sentence determined as the abnormal problem;
and the updating module is used for updating the knowledge base of the question-answering system according to the session record.
The present invention also provides an electronic device, including: one or more processors; and a memory associated with the one or more processors for storing program instructions which, when read and executed by the one or more processors, perform the method of handling smart question-and-answer exceptions of any of claims 1-8.
The embodiment of the invention has the following beneficial effects:
the method for processing the intelligent question-answer exception comprises the steps of obtaining a conversation record of an intelligent customer service and a user, wherein the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by the user and answer sentences sent by the intelligent customer service; preprocessing the question sentences according to a preset preprocessing method; judging whether the preprocessed question sentences are abnormal problems or not according to a preset abnormal problem judgment method; setting a statement pair corresponding to the abnormal problem according to a preset abnormal problem processing rule; and updating the knowledge base of the intelligent customer service according to the statement pair. The problems that analysis and mining speed of question sentences and answer sentences between a user and the intelligent customer service is low, and analysis and mining speed of question sentences and answer sentences between the user and the intelligent customer service is low, so that the knowledge base updating speed is low, the answer success rate is low, and the working pressure of manual customer service is high are solved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart (I) of a processing method of an intelligent question-answering exception according to an embodiment of the present application;
fig. 2 shows a flowchart of a processing method of an intelligent question and answer exception according to an embodiment of the present application (ii).
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
Specifically, as shown in fig. 1, the process of analyzing and mining the intelligent question-answer exception of the intelligent customer service and the user according to the processing method of the intelligent question-answer exception provided by the embodiment of the present application includes:
in the design and development stage of a product, a person skilled in the art can summarize problems that can occur according to the working requirements and working scenes of different systems, and then implant a code identifier of a problem to be collected into a code in advance in a code embedding manner, where the problem to be collected may be the following situations: the question (including the standard question or the similar question corresponding to the question) does not exist in the knowledge base, the intelligent customer service cannot give an answer, and after the user obtains the answer, the manual operation is selected; when the intelligent customer service question-answering flow triggers the point, namely when a conversation occurs between the user and the question-answering system, a conversation record between the intelligent customer service and the client is collected, wherein the conversation record at least comprises two sentences, the sentences comprise question sentences sent by the user and answer sentences sent by the intelligent customer service, and as shown in table 1, the example is a part of question coding example:
TABLE 1 problem code example table
Storing the collected session records into a specified storage medium, where the storage medium may be: the file system, the system memory, the Kafka and the like can specifically select the corresponding storage mode according to different conditions. For example: when the analysis period is long and the total amount of problems is large, the problems can be stored in a file system; when the period is short and the total amount of problems is moderate, the problems can be stored in a system memory; when analysis is required in real time, Kafka can be used for problem propagation.
After obtaining a session record between the intelligent customer service and the user, preprocessing the question sentences according to a preset preprocessing method;
specifically, the process of preprocessing the session record includes:
text error correction, namely correcting errors of wrongly written words contained in the text by using a preset error correction rule;
the text may include a speech sentence and a text sentence. When the conversation record is a text statement, the main wrongly written characters are homophones caused by a user input method; when the conversation record is a speech sentence, the speech sentence needs to be converted into a text sentence through a speech recognition technology, and in the sentence conversion process, not only homophones but also any pronunciation or words with the same pronunciation are included. Therefore, the embodiment of the application combines the language model and the word frequency characteristic, sets corresponding error correction rules for the voice statement and the text statement respectively, and can correct the wrongly written words according to the corresponding error correction rules;
filtering forbidden words, and carrying out filtering operation on the forbidden words in the text;
specifically, the forbidden words include removing dirty words and other unsatisfactory sentences that do not conform to polite terms, and performing a filtering operation thereon.
Denoising the data, namely removing 'noise' contained in the obtained text;
specifically, the noise includes irrelevant characters including preset useless punctuations and preset stop words, and irrelevant information contained in any text, such as trade names, place names and the like;
data duplication removal, namely comparing the acquired text with the text stored in a knowledge base of the intelligent customer service to remove duplication; the acquired text can be compared with the text in the knowledge base of the intelligent customer service according to a semantic similarity algorithm; wherein the semantic similarity algorithm may be a Bert-based text similarity algorithm and/or a WMD-based text similarity algorithm.
Problem normalization and determination of problem dictionary dittAnd normalizing the acquired question questions into standard questions based on semantic similarity calculation of a text similarity algorithm of Bert and/or a text similarity algorithm of WMD.
Specifically, the problem normalization described above may include the steps of:
initialization, including by preset initialization<key，value>Key-value pair question dictionary dittCounting the frequency of occurrence of question questions, wherein: key is the problem, value is the frequency of occurrence of the problem;
calculating the similarity; semantic similarity calculation is carried out on the newly input question query and each question key in the dictionary by utilizing a semantic similarity calculation method, and the score of the maximum similarity is determinedmaxCorresponding question and is noted as keymax；
Judging the attribution of the newly added question; when scoremaxWhen the similarity is larger than or equal to the set similarity threshold, the newly input question query and the question key corresponding to the maximum similarity score are judgedmaxExpressing the same question and keymaxThe corresponding value + 1; when scoremaxIf the similarity is less than the set similarity threshold, the newly input query is judged to be a new question, and a corresponding number of queries are added to the question dictionary dictt<key，value>A key-value pair;
determining a problem dictionary ditt。
Judging an abnormal problem, namely judging whether the preprocessed question sentence is an abnormal problem or not;
specifically, the above abnormal problem determination may include the steps of: obtaining normalized dictionary dit in period ttAnd T-T problem normalization dictionary fact of previous periodt-T；
Obtaining the normalized dictionary dit in the period ttNumber qc of any question iitNumber fraction qcritSmooth ring ratio growth rate qsrit(ii) a Expressing normalized dictionary dit as queryitA plurality of questions asked by the user and collected in the system are described by taking tables 2 to 4 as examples:
TABLE 2 summary of number of questions asked during period t
| query (question) | Count (number of times) |
| query1 | 200 |
| query2 | 122 |
| query3 | 800 |
| query4 | 900 |
| query5 | 78 |
| query6 | 500 |
| query7 | 56 |
| query8 | 99 |
| query9 | 111 |
| query10 | 89 |
| query11 | 10 |
TABLE 3 summary of times of questions asked during T-T period
TABLE 4 summary of growth rates
As can be seen from fig. 2-table 4: query6The abnormal growth rate of (1) is the highest, and the abnormal growth rate is increased from 78 to 500, and the abnormal problem determination requirements are met no matter from the absolute ratio of the quantity or from the growth rate of the return-to-positive smooth ring ratio; query11The loop ratio increase rate of the correction smoothing of (2) is high, but the absolute ratio of the data is low, so that the data is not suitable for processing as an abnormal problem, and therefore, the necessity of adopting the absolute ratio of the data as a judgment factor and smoothing the loop ratio increase rate can be reflected; query4The absolute percentage of the data is 30.35%, the data is still the maximum percentage problem, but because the ring ratio growth rate is negative, the consulting amount of the problem is reflected to be descending, the data can be pushed as an abnormal problem in the last period, and in the current period, the abnormal growth rate is ranked at 4, the importance of the data is not ignored because of the negative growth of the data, and the descending trend can be reflected; therefore, the number of data to be considered and the ratio increase rate correction of the smooth ring ratio can also be explainedAccording to the necessity of absolute proportion;
calculating to obtain the abnormal growth rate qr of any question i in the period t according to the obtained data informationit：
Wherein: problem normalization dictionary fact of previous period T-Tt-T，qritIs the abnormal growth rate, qcr, of any question i within the period titIs the normalized dictionary dit within the period ttThe ratio of the number of any question i, qsritIs the normalized dictionary dit within the period ttSmooth loop ratio growth rate, qc, of any question iitIs the normalized dictionary dit within the period ttNumber of any question questions i, qci(t-T)Is the normalized dictionary fact in the previous period T-TtThe number of any question i;
according to a preset abnormity judgment threshold value, if the abnormal growth rate qr of any question i is obtaineditIf the question exceeds a preset abnormal judgment threshold value, determining the question as an abnormal question; the setting of the abnormal judgment threshold is not limited, and a person skilled in the art can normalize the dictionary dit according to the service scene requirements, that is, the length of the sampling period t and the requirements of the sampling period ttAnd selecting a suitable anomaly determination threshold according to the sampling quantity and the service requirement.
And pushing the question determined as the abnormal question to a pushing system, wherein the pushing system can be used as a pushing channel in various ways such as letter, IM, mail and the like.
Preferably, after the question questions judged as abnormal questions are pushed to the pushing system, the abnormal questions can be artificially judged for the second time by operation and maintenance personnel, if the operation and maintenance personnel secondarily determine that the question questions are abnormal questions, a page is configured by means of a dynamic rule base, the questions and corresponding questions are added to the question-answering system through the dynamic rule base, the questions take effect in real time after completion, and similar questions of subsequent users can be automatically answered by the intelligent customer service system.
Example two
In response to the foregoing embodiments, the present application provides a method for processing an intelligent question-answering exception, with reference to fig. 2, where the method includes the following steps:
s1: acquiring a conversation record of an intelligent customer service and a user, wherein the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by the user and answer sentences sent by the intelligent customer service;
s3: preprocessing the question sentences according to a preset preprocessing method;
s4: judging whether the preprocessed question sentence is an abnormal question or not according to a preset abnormal question judging method, if so, executing step S5, and if not, ending the processing;
s5: setting a statement pair corresponding to the abnormal problem according to a preset abnormal problem processing rule;
s6: and updating the knowledge base of the intelligent customer service according to the statement pair.
Preferably, before the step S1, the method further includes the following steps:
s0: triggering a preset log code.
Preferably, the following steps are further included between the step S1 and the step S3:
s2: storing the acquired session record to a storage medium;
the storage medium comprises a file system, a system memory and Kafka.
In this embodiment, the preprocessing of the session record includes:
and normalizing the session records, and converting the acquired session records into character coded data.
In this embodiment, the preprocessing of the session record further includes:
correcting errors of the session records, namely correcting errors of wrongly written words included in the session records by using a preset error correction rule;
filtering the session records, and filtering the session records;
the method comprises the following steps of (1) removing duplication of question sentences, namely comparing the obtained question sentences with question sentences stored in a knowledge base of the intelligent customer service to remove duplication;
and normalizing the question questions, and normalizing the obtained question questions with similar semantics into standard questions based on a semantic similarity algorithm.
In this embodiment, the normalization of the question includes the following steps:
s31: initialization, including by preset initialization<key，value>Dictionary dit of key-value pairstCounting the frequency of occurrence of question questions, wherein: key is the problem, value is the frequency of occurrence of the problem;
s32: calculating the similarity; semantic similarity calculation is carried out on the newly input question query and each question key in the dictionary by utilizing a semantic similarity calculation method, and the score of the maximum similarity is determinedmaxCorresponding question and is noted as keymax；
S33: judging the attribution of the newly added question; when scoremaxWhen the similarity is larger than or equal to the set similarity threshold, the newly input question query and the question key corresponding to the maximum similarity score are judgedmaxExpressing the same question and keymaxThe corresponding value + 1; when scoremaxIf the similarity is smaller than the set similarity threshold, the newly input query is judged to be a new problem, and the dictionary fact is usedtAdding a corresponding number<key，value>A key-value pair;
s34: determining normalized dictionary ditt。
In this embodiment, the step S4 includes the following steps:
s41: obtaining normalized dictionary dit in period ttAnd T-T problem normalization dictionary fact of previous periodt-T；
S42: obtaining the normalized dictionary dit in the period ttNumber qc of any question iitNumber fraction qcritSmooth ring ratio growth rate qsrit；
S43: according to the data information obtained in the step S42, calculating the abnormal growth rate qr of any question i in the period tit：
Wherein: problem normalization dictionary fact of previous period T-Tt-T；
S44: based on the preset abnormality determination threshold, the abnormal increase rate qr of any question i obtained in step S43itIf the value exceeds the preset abnormality determination threshold, step S5 is executed, otherwise, the process ends.
In this embodiment, the semantic similarity algorithm includes a text similarity algorithm based on Bert and a text similarity algorithm based on WMD.
EXAMPLE III
Corresponding to the first embodiment and the second embodiment, the application provides an intelligent question-answering exception device, which comprises:
the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring a conversation record, the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by a user and answer sentences sent by an intelligent customer service;
the storage module is used for storing the acquired session records;
the preprocessing module is used for preprocessing the acquired session records;
the judging module is used for judging whether the obtained question is an abnormal question or not;
the processing module is used for formulating a corresponding answer sentence for the question sentence determined as the abnormal problem;
and the updating module is used for updating the knowledge base of the question-answering system according to the session record.
Preferably, the storage module stores the acquired session record to a storage medium, where the storage medium includes a file system, a system memory, and Kafka; the storage medium for session recording is not required, and those skilled in the art can select a storage medium and a data storage scheme that meet the storage requirement according to the actual service scenario requirement.
Preferably, in this embodiment, when the analysis period is long and the total number of problems is large, the analysis period may be stored in a file system; when the period is short and the total amount of problems is moderate, the problems can be stored in a system memory; when analysis is required in real time, Kafka can be used for problem propagation.
Preferably, the pre-processing module comprises the steps of: converting the acquired session record into character encoding (uTF-8 encoding) data; correcting errors of wrongly written words included in the session record by using a preset error correction rule; the session record is filtered, for example: when the session length of the intelligent customer service and the user is less than two rounds, the session record is deleted, the above example only illustrates a method for filtering the session record, and a person skilled in the art can set different filtering rule algorithms according to the actual needs of a service scene, and filter information which does not meet requirements; comparing the obtained question sentences with the question sentences stored in a knowledge base of the intelligent customer service to remove duplication; and normalizing the obtained question questions with similar semantics into standard questions by using a text similarity algorithm based on Bert and/or a text similarity algorithm based on WMD.
Preferably, the judging module comprises the following steps: s41: obtaining normalized dictionary dit in period ttAnd T-T problem normalization dictionary fact of previous periodt-T(ii) a S42: obtaining the normalized dictionary dit in the period ttNumber qc of any question iitNumber fraction qcritSmooth ring ratio growth rate qsrit(ii) a S43: according to the data information obtained in the step S42, calculating the abnormal growth rate qr of any question i in the period tit：
Wherein: problem normalization dictionary fact of previous period T-Tt-T(ii) a S44: based on the preset abnormality determination threshold, the abnormal increase rate qr of any question i obtained in step S43itIf the value exceeds the preset abnormality determination threshold, step S5 is executed, otherwise, step S is executedEnding the processing, setting the abnormal judgment threshold value is not limited, and a person skilled in the art can normalize the dictionary dit according to the service scene requirements, namely the length of the sampling period t and the requirements of the sampling period ttAnd selecting a suitable anomaly determination threshold according to the sampling quantity and the service requirement.
Example four
Corresponding to all the above embodiments, an electronic device provided in an embodiment of the present application includes:
one or more processors;
and a memory associated with the one or more processors for storing program instructions that, when read and executed by the one or more processors, perform the method of handling smart question-and-answer exceptions described in embodiments 1 and 2 above.
Preferably, an electronic device provided in the embodiments of the present application may specifically include a processor, a processor communicatively connected through a communication bus, an input/output interface, a network interface, and a memory.
The processor can be implemented by a general-purpose CPU, a microprocessor, an application specific integrated circuit, or one or more integrated circuits, and the like, and is configured to execute a related program to implement the technical solution provided by the present application;
the memory may be implemented in the form of ROM, RAM, static storage devices, dynamic storage devices, etc. The memory can store a processing method for controlling the electronic equipment to run and realizing the intelligent question-answering exception, which is described in the application, through software and/or firmware;
the input/output interface is used for connecting the input/output module to realize the input and output of information. The input/output module may be configured in the electronic device as a component, or may be external to the device to provide a corresponding function. The input device may include various sensing elements such as a keyboard, a mouse, a touch screen, and the like, and may also be similar to wireless communication such as a mobile network, wifi, bluetooth, and the like.
It should be noted that although the electronic device is described above only with reference to the processor, the input/output interface, the network interface and the memory, in a specific implementation process, the electronic device may further include other components necessary for achieving normal operation of the electronic device.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. An intelligent question-answering exception processing method is characterized by comprising the following steps:
s1: acquiring a conversation record of an intelligent customer service and a user, wherein the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by the user and answer sentences sent by the intelligent customer service;
s3: preprocessing the question sentences according to a preset preprocessing method;
s4: judging whether the preprocessed question sentence is an abnormal question or not according to a preset abnormal question judging method, if so, executing step S5, and if not, ending the processing;
s5: setting a statement pair corresponding to the abnormal problem according to a preset abnormal problem processing rule;
s6: and updating the knowledge base of the intelligent customer service according to the statement pair.
2. The method for processing the intelligent question answering exception according to claim 1, wherein before the step S1, the method further comprises the following steps:
s0: triggering a preset log code.
3. The method for processing the intelligent question answering exception according to the claim 1 or 2, wherein the following steps are further included between the step S1 and the step S3:
s2: storing the acquired session record to a storage medium;
the storage medium comprises a file system, a system memory and Kafka.
4. The method for processing intelligent question-answering exceptions according to claim 3, wherein the preprocessing of the session record comprises:
and normalizing the session records, and converting the acquired session records into character coded data.
5. The method for processing intelligent question-answering exceptions according to claim 4, wherein the preprocessing of the session record further comprises:
correcting errors of the session records, namely correcting errors of wrongly written words included in the session records by using a preset error correction rule;
filtering the session records, and filtering the session records;
the method comprises the following steps of (1) removing duplication of question sentences, namely comparing the obtained question sentences with question sentences stored in a knowledge base of the intelligent customer service to remove duplication;
and normalizing the question questions, and normalizing the obtained question questions with similar semantics into standard questions based on a semantic similarity algorithm.
6. The method for processing the intelligent question-answering exception according to claim 5, wherein the question normalization comprises the following steps:
s31: initialization, including by preset initialization<key，value>Dictionary dit of key-value pairstCounting the frequency of occurrence of question questions, wherein: key is the problem, value is the frequency of occurrence of the problem;
s32: calculating the similarity; using semantic similarity algorithm to query the newly input question and the new one in dictionarySemantic similarity calculation is carried out on each question key, and the score of the maximum similarity is determinedmaxCorresponding question and is noted as keymax；
S33: judging the attribution of the newly added question; when scoremaxWhen the similarity is larger than or equal to the set similarity threshold, the newly input question query and the question key corresponding to the maximum similarity score are judgedmaxExpressing the same question and keymaxThe corresponding value + 1; when scoremaxIf the similarity is smaller than the set similarity threshold, the newly input query is judged to be a new problem, and the dictionary fact is usedtAdding a corresponding number<key，value>A key-value pair;
s34: determining normalized dictionary ditt。
7. The method for processing the intelligent question-answering exception according to any one of claims 1-2 and 4-6, wherein the step S4 includes the following steps:
s41: obtaining normalized dictionary dit in period ttAnd normalizing dictionary fact for problem in previous period T-Tt-T；
S42: obtaining the normalized dictionary dit in the period ttNumber qc of any question iitNumber fraction qcritSmooth ring ratio growth rate qsrit；
S43: according to the data information obtained in the step S42, calculating the abnormal growth rate qr of any question i in the period tit：
Wherein: problem normalization dictionary fact of previous period T-Tt-T，qritIs the abnormal growth rate, qcr, of any question i within the period titIs the normalized dictionary dit within the period ttThe ratio of the number of any question i, qsritIs the normalized dictionary dit within the period ttSmooth loop ratio growth rate, qc, of any question iitIs the normalized dictionary dit within the period ttNumber of any question questions i, qci(t-T)Is the normalized dictionary fact in the previous period T-TtThe number of any question i;
s44: based on the preset abnormality determination threshold, the abnormal increase rate qr of any question i obtained in step S43itIf the value exceeds the preset abnormality determination threshold, step S5 is executed, otherwise, the process ends.
8. The method for processing the intelligent question-answering exception according to claim 6, wherein the semantic similarity algorithm is a Bert-based text similarity algorithm and/or a WMD-based text similarity algorithm.
9. An apparatus for intelligently answering exceptions, the apparatus comprising:
the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring a conversation record, the conversation record at least comprises two sentences, and the sentences comprise question sentences sent by a user and answer sentences sent by an intelligent customer service;
the storage module is used for storing the acquired session records;
the preprocessing module is used for preprocessing the acquired session records;
the judging module is used for judging whether the obtained question is an abnormal question or not;
the processing module is used for formulating a corresponding answer sentence for the question sentence determined as the abnormal problem;
and the updating module is used for updating the knowledge base of the question-answering system according to the session record.
10. An electronic device, characterized in that the electronic device comprises:
one or more processors;
and a memory associated with the one or more processors for storing program instructions which, when read and executed by the one or more processors, perform the method of handling smart question-and-answer exceptions of any of claims 1-8.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110985077.9A CN113743124B (en) | 2021-08-25 | 2021-08-25 | Intelligent question-answering exception processing method and device and electronic equipment |
| CA3170622A CA3170622A1 (en) | 2021-08-25 | 2022-08-17 | Method of intelligently processing q&a abnormality, device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110985077.9A CN113743124B (en) | 2021-08-25 | 2021-08-25 | Intelligent question-answering exception processing method and device and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113743124A true CN113743124A (en) | 2021-12-03 |
| CN113743124B CN113743124B (en) | 2024-03-29 |
Family
ID=78732952
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110985077.9A Active CN113743124B (en) | 2021-08-25 | 2021-08-25 | Intelligent question-answering exception processing method and device and electronic equipment |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN113743124B (en) |
| CA (1) | CA3170622A1 (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103377239A (en) * | 2012-04-26 | 2013-10-30 | 腾讯科技（深圳）有限公司 | Method and device for calculating inter-textual similarity |
| CN104679910A (en) * | 2015-03-25 | 2015-06-03 | 北京智齿博创科技有限公司 | Intelligent answering method and system |
| CN107315766A (en) * | 2017-05-16 | 2017-11-03 | 广东电网有限责任公司江门供电局 | A kind of voice response method and its device for gathering intelligence and artificial question and answer |
| CN107609101A (en) * | 2017-09-11 | 2018-01-19 | 远光软件股份有限公司 | Intelligent interactive method, equipment and storage medium |
| WO2019153613A1 (en) * | 2018-02-09 | 2019-08-15 | 平安科技（深圳）有限公司 | Chat response method, electronic device and storage medium |
| CN110162611A (en) * | 2019-04-23 | 2019-08-23 | 苏宁易购集团股份有限公司 | A kind of intelligent customer service answer method and system |
| WO2020019686A1 (en) * | 2018-07-27 | 2020-01-30 | 众安信息技术服务有限公司 | Session interaction method and apparatus |
| CN110737763A (en) * | 2019-10-18 | 2020-01-31 | 成都华律网络服务有限公司 | Chinese intelligent question-answering system and method integrating knowledge map and deep learning |
| CN112148743A (en) * | 2020-09-18 | 2020-12-29 | 北京达佳互联信息技术有限公司 | Method, device, equipment and storage medium for updating intelligent customer service knowledge base |
| CN113282733A (en) * | 2021-06-11 | 2021-08-20 | 上海寻梦信息技术有限公司 | Customer service problem matching method, system, device and storage medium |
-
2021
- 2021-08-25 CN CN202110985077.9A patent/CN113743124B/en active Active
-
2022
- 2022-08-17 CA CA3170622A patent/CA3170622A1/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103377239A (en) * | 2012-04-26 | 2013-10-30 | 腾讯科技（深圳）有限公司 | Method and device for calculating inter-textual similarity |
| CN104679910A (en) * | 2015-03-25 | 2015-06-03 | 北京智齿博创科技有限公司 | Intelligent answering method and system |
| CN107315766A (en) * | 2017-05-16 | 2017-11-03 | 广东电网有限责任公司江门供电局 | A kind of voice response method and its device for gathering intelligence and artificial question and answer |
| CN107609101A (en) * | 2017-09-11 | 2018-01-19 | 远光软件股份有限公司 | Intelligent interactive method, equipment and storage medium |
| WO2019153613A1 (en) * | 2018-02-09 | 2019-08-15 | 平安科技（深圳）有限公司 | Chat response method, electronic device and storage medium |
| WO2020019686A1 (en) * | 2018-07-27 | 2020-01-30 | 众安信息技术服务有限公司 | Session interaction method and apparatus |
| CN110162611A (en) * | 2019-04-23 | 2019-08-23 | 苏宁易购集团股份有限公司 | A kind of intelligent customer service answer method and system |
| CN110737763A (en) * | 2019-10-18 | 2020-01-31 | 成都华律网络服务有限公司 | Chinese intelligent question-answering system and method integrating knowledge map and deep learning |
| CN112148743A (en) * | 2020-09-18 | 2020-12-29 | 北京达佳互联信息技术有限公司 | Method, device, equipment and storage medium for updating intelligent customer service knowledge base |
| CN113282733A (en) * | 2021-06-11 | 2021-08-20 | 上海寻梦信息技术有限公司 | Customer service problem matching method, system, device and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 高森: "农业问答系统中问题分类和相似度计算的研究", 《中国优秀硕士学位论文全文数据库信息科技辑》, no. 1 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CA3170622A1 (en) | 2023-02-25 |
| CN113743124B (en) | 2024-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110555101A (en) | customer service knowledge base updating method, device, equipment and storage medium | |
| CN115168562A (en) | Method, device, equipment and medium for constructing intelligent question-answering system | |
| CN115454706A (en) | System abnormity determining method and device, electronic equipment and storage medium | |
| CN113342955A (en) | Question and answer sentence processing method and device and electronic equipment | |
| CN112951233A (en) | Voice question and answer method and device, electronic equipment and readable storage medium | |
| CN112579781B (en) | Text classification method, device, electronic equipment and medium | |
| CN111625858B (en) | Intelligent multi-mode data desensitization method and device in vertical field | |
| CN111950267A (en) | Method and device for extracting text triples, electronic equipment and storage medium | |
| CN113743124B (en) | Intelligent question-answering exception processing method and device and electronic equipment | |
| CN116340777A (en) | Training method of log classification model, log classification method and device | |
| CN114138743A (en) | ETL task automatic configuration method and device based on machine learning | |
| CN112002306B (en) | Speech class recognition method and device, electronic equipment and readable storage medium | |
| CN114840507A (en) | Data governance method and device, electronic equipment and storage medium | |
| CN114141235A (en) | Voice corpus generation method and device, computer equipment and storage medium | |
| CN113468176A (en) | Information input method and device, electronic equipment and computer readable storage medium | |
| CN112906650B (en) | Intelligent processing method, device, equipment and storage medium for teaching video | |
| CN112836529B (en) | Method and device for generating target corpus sample | |
| CN113836921B (en) | Electronic method and device for paper case data and electronic equipment | |
| CN114519357B (en) | Natural language processing method and system based on machine learning | |
| CN113779237B (en) | Method, system, mobile terminal and readable storage medium for constructing social behavior sequence diagram | |
| CN111191095A (en) | Webpage data acquisition method, device, equipment and medium | |
| CN114912445A (en) | Method and device for identifying case source line text data, storage medium and electronic equipment | |
| CN117763099A (en) | Interaction method and device of intelligent customer service system | |
| CN116821345A (en) | Customer service session anomaly detection method, system, electronic equipment and storage medium | |
| CN114648019A (en) | Event relation recognition method and device, computer equipment and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |