CN108959327B - Service processing method, device and computer readable storage medium - Google Patents
Service processing method, device and computer readable storage medium Download PDFInfo
- Publication number
- CN108959327B CN108959327B CN201710391804.2A CN201710391804A CN108959327B CN 108959327 B CN108959327 B CN 108959327B CN 201710391804 A CN201710391804 A CN 201710391804A CN 108959327 B CN108959327 B CN 108959327B
- Authority
- CN
- China
- Prior art keywords
- service
- action
- user
- probability vector
- classification
- 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.)
- Active
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a service processing method, which comprises the following steps: acquiring a user question; forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier; determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector; searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services; and finishing the response processing of the user question according to the determined service logic. The invention also discloses a business processing device and a computer readable storage medium.
Description
Technical Field
The present invention relates to the field of human-computer intelligent interaction, and in particular, to a service processing method, device and computer-readable storage medium.
Background
The intelligent customer service system is a customer service system which takes a natural language understanding technology as a core and gives corresponding answers or guides and helps a user to finish a corresponding business process according to related consultation, inquiry, handling and other types of problems provided by a mobile user.
At present, in an intelligent customer service system, a text dialogue system based on a question and answer library is generally used. The system finds the standard question which is most matched with the user question through matching of the keywords and the rules, and feeds back the answer which corresponds to the standard question as the answer to the current user.
However, when the processing method is applied to the mobile customer service field, due to the characteristics of the mobile customer service field, the processing method is determined to cause inaccurate problem identification, inaccurate problem answer and processing, and the like.
Disclosure of Invention
In order to solve the existing technical problem, embodiments of the present invention provide a service processing method, a service processing device, and a computer-readable storage medium.
The technical scheme of the embodiment of the invention is realized as follows:
the embodiment of the invention provides a service processing method, which comprises the following steps:
acquiring a user question;
forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector;
searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services;
and finishing the response processing of the user question according to the determined service logic.
In the above scheme, the determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector includes:
taking the category with the highest probability in the action classification probability vector as an action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
In the above scheme, the method further comprises:
when the action classification label corresponding to the highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the second highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector;
searching an action classification label and a service classification label corresponding to the second highest confidence level of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; and according to the determined service logic, completing the response to the user question;
and when the action classification label corresponding to the second highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the third highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector, searching the corresponding business logic, and so on until the business logic corresponding to the user problem is found.
In the foregoing solution, when determining the confidence level of the user question, the method includes:
taking the product of each action classification probability in the action classification probability vector and the first value; the first value characterizes actions and business ticket adjustment parameters;
taking the product of each service classification probability in the service classification probability vector and the second value; the sum of the second value and the first value meets a preset condition;
summing the product of each action classification probability and the first value with the product of each service classification probability and the second value to form a set;
when the business logic corresponding to the user problem is determined, one summation result is sequentially selected from the set according to the sequence of the summation results from large to small, the category of the action classification probability corresponding to the selected summation result is the action classification label corresponding to the user problem, and the category of the business classification probability corresponding to the selected summation result is the business classification label corresponding to the user problem.
In the above solution, the completing, according to the determined action and the service logic corresponding to the service, the response processing of the user question includes:
and outputting answers to the user questions or guiding the user to complete the handling of the corresponding services according to the service logic.
In the foregoing solution, before forming the action classification probability vector and the service classification probability vector, the method further includes:
and performing word segmentation processing on the user question.
An embodiment of the present invention further provides a service processing apparatus, including:
an acquisition unit for acquiring a user question;
the classification unit is used for forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
a determining unit, configured to determine an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector;
the searching unit is used for searching the action classification label and the service classification label of the user problem in the information dimension table and determining the service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services;
and the processing unit is used for finishing response processing to the user question according to the determined service logic.
In the foregoing solution, the determining unit is specifically configured to:
taking the category with the highest probability in the action classification probability vector as an action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
In the above scheme, the determining unit is further configured to determine the second highest confidence level of the user problem by using an action classification probability vector and a service classification probability vector when the action classification label corresponding to the highest confidence level and the service logic corresponding to the service classification label are not found in the information dimension table;
the searching unit is also used for searching an action classification label and a service classification label corresponding to the second highest confidence level of the user problem in the information dimension table and determining the service logic corresponding to the user problem; and according to the determined service logic, completing the response to the user question;
when the action classification label corresponding to the second highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, the determining unit determines the third highest confidence level of the user problem by using the action classification probability vector and the business classification probability vector, and the searching unit searches the corresponding business logic, and so on until the business logic corresponding to the user problem is found.
In the foregoing scheme, the processing unit is specifically configured to:
and outputting answers to the user questions or guiding the user to complete the handling of the corresponding services according to the service logic.
In the above scheme, the apparatus further comprises:
and the word segmentation unit is used for carrying out word segmentation processing on the user question.
An embodiment of the present invention further provides a service processing apparatus, including: a processor and a memory for storing a computer program capable of running on the processor,
wherein the processor is configured to perform the steps of the above method when running the computer program.
Embodiments of the present invention also provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the above-mentioned method.
The business processing method, the business processing device and the computer readable storage medium provided by the embodiment of the invention are used for acquiring user problems; forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier; determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector; searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services; and finishing response processing on the user questions according to the determined service logic, and classifying and identifying the user questions by adopting a supervised machine learning algorithm in two dimensions of action and service, so that the requirements of the user can be accurately identified, and the problems can be accurately answered and processed.
Drawings
In the drawings, which are not necessarily drawn to scale, like reference numerals may describe similar components in different views. The drawings illustrate generally, by way of example, but not by way of limitation, various embodiments discussed herein.
Fig. 1 is a schematic flow chart of a method for processing a service according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating an information dimension table structure according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating a process of identifying and responding to a user question according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a service processing apparatus according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a hardware structure of a service processing apparatus according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following embodiments.
The intention of the user is accurately understood to become the basic and core technical requirements for the success of the intelligent customer service system. The current intelligent customer service system is generally a text dialogue system based on a question-answer library, namely, a user inputs a text, the intelligent customer service system finds a standard question most matched with the user question through matching of keywords and rules, and an answer corresponding to the standard question is fed back to the user as an answer. When the answers are matched, similarity matching and rearrangement after matching are carried out on the user questions in the massive chatting question-answer library, so that the question which is most matched with the user questions is found. Therefore, the current intelligent customer service system is suitable for dealing with the robot scene of general chatting.
The inventor finds that: for mobile services, there are often multiple expressions to the same basic problem, since some customers are not well aware of it and natural language (speech) itself has great flexibility. Literally expressing similar customer problems may require different business actions or different business transaction objects.
Meanwhile, due to the diversity of mobile services, dozens of handling actions and thousands of mobile company services are involved. For each action and business, a plurality of expressions are used, and the combination of two dimensions is rather inexhaustible. Therefore, if a question answer library is prepared and similar question matching is carried out, the workload is huge, the problem of data sparseness cannot be solved, and the effect is not good. When the action and the service intention required by the user cannot be accurately identified, the user cannot be helped to complete corresponding service handling.
In summary, the existing intelligent customer service system mainly has the following problems:
1. the semantic understanding of the user questions is to search the closest questions in the existing question-answering library and then perform secondary sorting or comprehensive sorting. For the inaccurate identification of the approximation problem, several related approximation problems can be found, but it is difficult to rank the correct problem first.
2. The mobile customer service field is professional, and relates to dozens of business processes such as inquiry, consultation, handling, sending and the like and two thousand specific businesses. If the existing technical scheme is used, a large number of standard questions and answers need to be prepared to cover the combination situation of two dimensions, so that a large amount of manpower and resources are consumed, and the cost of later maintenance and data updating is high. In addition, even if a lot of data is stored in the database, the data is sparse, and the retrieval result is inaccurate.
3. This solution of preparing a library of answers to questions is not particularly concerned with differentiating the nuances of user expressions in both the action and business dimensions. In addition, each word in the user question may have multiple descriptions or synonyms, the number of synonym combinations of each word is multiplied, and greater difficulty is brought to the preparation of the standard question bank.
In other words, when the method based on the question-answer library is applied to the field of mobile customer service, the following technical problems may exist:
1. the answer questions of the intelligent customer service of the operator are inaccurate, and similar related questions are difficult to distinguish. The synonym recognition capability is poor, synonyms of each word in the problem need to be exhausted in a problem library, and the algorithm generalization capability is poor.
2. In order to fully cover the user questions, a massive question answer library needs to be prepared, and the manual workload is large. And the coverage is not uniform and comprehensive enough. In the subsequent operation process, the knowledge updating cost is high, the manual manufacturing problem is caused, errors are easy to occur, and the management is difficult.
Meanwhile, the inventors have also found that: the problems posed by users in the field of mobile services are all of definite action and traffic classification. Such as: "I want to make a strange recharge", the action is: handling, the business is: and (4) mobile recharging. For another example, "the mobile phone reports that me does not need, and cancels the bar in the next month", the actions are: cancellation, the service is: and (5) mobile phone newspaper.
Based on this, in various embodiments of the present invention, supervised machine learning algorithms are adopted for user questions in two dimensions of action and service to perform classification and identification processing, so that the requirements of the user can be accurately identified, and accurate system response is realized.
An embodiment of the present invention provides a service processing method, which is applied to a mobile intelligent customer service system, and as shown in fig. 1, the method includes:
step 101: acquiring a user question;
here, in actual application, the presentation form of the acquired user question may be a voice form, a text form, or the like.
Step 102: forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
in order to subsequently and quickly find the service logic corresponding to the user question so as to perform corresponding response processing, after the user question is obtained, word segmentation processing may be performed on the user question first, which specifically includes: cleaning and word segmentation, wherein the cleaning means: stop words, illegal symbols and the like are removed.
Wherein the purpose of the word segmentation is to divide the user question into a plurality of words.
Forming an action classification probability vector by the user problem after word segmentation processing through an action classifier; and the user problem after word segmentation processing is processed through a service classifier to form a service classification probability vector.
The classifier is actually a classification model, and specifically, a classification model established based on training samples.
In practical application, the action classifier and the service classifier are supervised classification classifiers, and can be trained by selecting various classification models which can output classification result probability distribution, such as logic, Deep Neural Network (DNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and the like.
Step 103: determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector;
specifically, the category with the highest probability in the action classification probability vector is used as the action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
Step 104: searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem;
here, the information dimension table is a two-dimensional table formed by actions and services.
In actual application, as shown in fig. 2, a two-dimensional information dimension table is formed according to actions (rows) and services (columns) of a user question, and answers corresponding to the questions, namely service logic, are filled in the middle part. Such as: the introduction of the line shall be the introduction of the call details, the introduction of the network access rules, the introduction of the passwords, and the like. To handle this line, the handling method should be filled out or the user should be guided directly to handle the corresponding service.
It should be noted that: the two-dimensional table does not have knowledge of every grid corresponding to (n, m) coordinates. For example, assume that the action is a transaction and the service is a package, so there are: the + package is handled. But no action is transacted and the service is detailed, i.e. there is no: the + details are handled because this is not a legitimate business logic. Only the query + details, or the introduction + details are legal business logic.
Step 105: and finishing the response processing of the user question according to the determined service logic.
Specifically, according to the service logic, an answer to the user question is output, or the user is guided to complete the handling of the corresponding service.
Here, in the scheme of the embodiment of the present invention, when determining the action classification tag and the service classification tag corresponding to the user problem, the action classification tag and the service classification tag corresponding to the highest confidence level are used first to search the corresponding service logic in the information dimension table, and when finding the corresponding service logic, the response processing on the user problem is completed according to the determined service logic; when the corresponding service logic is not found, the action classification label and the service classification label corresponding to the second highest confidence level are used for searching the corresponding service logic in the information dimension table, and when the corresponding service logic is found, the response processing to the user problem is completed according to the determined service logic; and when the corresponding service logic is not found, searching the corresponding service logic in the information dimension table by using the action classification label and the service classification label corresponding to the third high confidence coefficient, and repeating the steps until the service logic corresponding to the user problem is found.
And determining the confidence of the user problem by utilizing the action classification probability vector and the service classification probability vector.
Specifically, the product of each action classification probability in the action classification probability vector and a first value is obtained; the first value characterizes actions and business ticket adjustment parameters;
taking the product of each service classification probability in the service classification probability vector and the second value; the sum of the second value and the first value meets a preset condition;
summing the product of each action classification probability and the first value with the product of each service classification probability and the second value to form a set;
when the business logic corresponding to the user problem is determined, one summation result is sequentially selected from the set according to the sequence of the summation results from large to small, the category of the action classification probability corresponding to the selected summation result is the action classification label corresponding to the user problem, and the category of the business classification probability corresponding to the selected summation result is the business classification label corresponding to the user problem.
Wherein the preset conditions are as follows: the sum of the first value and the second value is equal to 1.
In the above process, the formula for determining the second highest confidence level includes: MAX (rate active + (1-rate) attribute) is maximized on the condition that the action classification probability is not equal to the maximum action classification probability or the traffic classification probability is not equal to the maximum traffic classification probability. Accordingly, similar formulas may be employed for determining other confidence levels. That is, in the searching process, the action classification probability and the corresponding service classification probability corresponding to the maximum value of the summation result of the action classification probability and the corresponding service classification probability except the used action classification probability and the corresponding service classification probability are taken each time, and the corresponding service logic is searched in the information dimension table.
Here, rate denotes a first value, 1-rate denotes a second value; ACTIONn denotes an action classification probability and endityn denotes a traffic classification probability.
The used action classification probability and the corresponding service classification probability refer to: and determining the action classification probability and the corresponding service classification probability of the corresponding service logic which is not found in the information dimension table.
The business processing method provided by the embodiment of the invention obtains the user problem; forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier; determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector; searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services; and finishing response processing on the user questions according to the determined service logic, and classifying and identifying the user questions by adopting a supervised machine learning algorithm in two dimensions of action and service, so that the requirements of the user can be accurately identified, and the problems can be accurately answered and processed.
In addition, the action classification label and the business classification label corresponding to the highest confidence coefficient are used for searching the corresponding business logic in the information dimension table, and when the corresponding business logic is searched, the response processing to the user problem is completed according to the determined business logic; when the corresponding service logic is not found, the action classification label and the service classification label corresponding to the second highest confidence level are used for searching the corresponding service logic in the information dimension table, and when the corresponding service logic is found, the response processing to the user problem is completed according to the determined service logic; and when the corresponding business logic is not found, searching the corresponding business logic in the information dimension table by using the action classification label and the business classification label corresponding to the third high confidence coefficient, and repeating the steps until the business logic corresponding to the user problem is found, wherein when the two classification probabilities are close, ambiguity can be removed through the process, so that the requirements of the user can be further identified, and the problem can be accurately solved and processed.
In conjunction with the above description, the process of identifying and responding to a user question, as shown in fig. 3, includes the following steps:
step 301: performing word segmentation processing on the user problem in the voice form;
specifically, the user problems are cleaned, and stop words and illegal symbols are removed; and then word segmentation processing is carried out.
Step 302: forming a motion classification probability vector by the processed user problem through a motion classifier;
here, the ACTION classification probability vector may be expressed as ACTION ═ ACTION1, ACTION2 … ACTION ], where ACTION is the probability that the question is the nth ACTION class.
Here, the category with the highest probability is the action classification result of the sentence (i.e., the user question), i.e., the action classification label as the user question.
Step 303: forming a service classification probability vector by the processed user problem through a service classifier;
here, the traffic classification probability vector may be expressed as ENTITY ═ ENTITY1, ENTITY2 … ENTITY ], where ENTITY is the probability that the problem is the nth traffic class.
Here, the category with the highest probability is the traffic classification result of the sentence (i.e., the user question), i.e., the traffic classification label as the user question.
It should be noted that: in actual application, the execution sequence of steps 302 and 303 may not be sequential, for example, step 302 may be executed first, and then step 303 may be executed; step 303 may be performed first, and then step 302 may be performed; steps 302, 303 may also be performed simultaneously.
Step 304: inquiring in the information dimension table according to the action classification label of the user question and the service classification label thereof;
step 305: checking whether the corresponding position (such as m rows and n columns) has corresponding knowledge (namely business logic); if not, go to step 306, otherwise go to step 307;
step 306: if the corresponding knowledge is not found, selecting the user problem with the lower level of confidence coefficient, and then executing step 304;
here, the corresponding knowledge is not found, which indicates that there is no corresponding problem. For example, according to the semantic classification of two dimensions, the corresponding semantic is found to be cancel + detail. However, no such business logic exists. Indicating that the classification result is wrong. Possibly because the classification of the user's problem is inaccurate.
The second highest confidence user semantic understanding may be selected according to the following formula:
and searching according to the probability distribution in the ACTION vector and the ENTITY vector:
searching (n, m) in the dimension information base, and maximizing MAX (rate ACTIONN + (1-rate) ENTITYN) under the condition that the action classification is not equal to the maximum action label or the service is not equal to the original maximum service classification label.
Where rate is the action and service ticket adjustment parameter. In practical application, the rate may be taken empirically, for example, it may be taken as 0.5, which indicates that activity and ENTITY are equally important. Of course, other values may be used, for example, a value greater than 0.5 may be used when ACTION is more important than ENTITY, i.e., when ACTION takes precedence.
And sequentially selecting the user semantic understanding with the second highest confidence level, and when the step 304 is executed again and the corresponding knowledge is not found, selecting the user semantic understanding with the third highest confidence level again, executing the step 304, and repeating the steps until the corresponding knowledge is found, so as to output an answer.
Wherein, when the corresponding knowledge is not found finally, preset words are output to the user, such as: not good meaning, not knowing what you are saying, etc.
Step 307: finding out corresponding knowledge, wherein the two classification results are correct, positioning the action and service corresponding to the user question, and outputting a corresponding answer according to the corresponding knowledge or guiding the user to complete a corresponding handling process, namely outputting the answer.
As can be seen from the above description, the solution provided by the embodiment of the present invention is an intelligent customer service system that semantically understands a user question by combining two dimensional classification methods of an action and a service, and a solution for implementing a user question answer.
In summary, the solution of the embodiment of the present invention has the following advantages:
firstly, the semantic understanding of the current user questions is to search the closest questions in the existing question-answering database and then perform secondary sorting or comprehensive sorting. The method and the device have the advantages that the approximate problems are inaccurately identified, and the classifier and the labeled linguistic data are utilized, so that the accurate classification of two dimensions of actions and services can be realized in the approximate problems, and the user problems can be accurately understood.
Secondly, the field of mobile customer service is professional, and dozens of business processes such as inquiry, consultation, handling and sending and two thousand specific businesses are involved. If the current technical scheme is used, a large number of standard standby questions and answers need to be prepared to cover the combination situation of two dimensions. A large amount of manpower and resources are consumed, the scheme of the embodiment of the invention can decompose a knowledge result from two dimensions of action and business, a large amount of linguistic data can be reused, and the problem of sparse training linguistic data is solved. Meanwhile, the cost of subsequent knowledge maintenance and modification can be greatly reduced.
Thirdly, the problem of inaccurate classifier result is corrected through the form of a two-dimensional knowledge base, ambiguity elimination is carried out through legal business logic, and business logic knowledge can be conveniently introduced into a problem semantic understanding system.
In order to implement the method according to the embodiment of the present invention, an embodiment of the present invention further provides a service processing apparatus, as shown in fig. 4, where the apparatus includes:
an obtaining unit 41, configured to obtain a user question;
the classification unit is used for forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
a determining unit 42, configured to determine an action classification label and a service classification label corresponding to the user question by using the action classification probability vector and the service classification probability vector;
the searching unit 43 is configured to search the action classification tag and the service classification tag of the user problem in the information dimension table, and determine a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services;
and the processing unit 44 is configured to complete response processing on the user question according to the determined service logic.
Here, in actual application, the presentation form of the acquired user question may be a voice form, a text form, or the like.
In order to subsequently and quickly find the service logic corresponding to the user question so as to perform corresponding response processing, after the user question is obtained, word segmentation processing may be performed on the user question first, which specifically includes: cleaning and word segmentation, wherein the cleaning means: stop words, illegal symbols and the like are removed.
Wherein the purpose of the word segmentation is to divide the user question into a plurality of words.
Forming an action classification probability vector by the user problem after word segmentation processing through an action classifier; and the user problem after word segmentation processing is processed through a service classifier to form a service classification probability vector.
The classifier is actually a classification model, and specifically, a classification model established based on training samples.
In practical application, the action classifier and the service classifier are classifiers with supervised classification, and can be trained by selecting various classification models which can output classification result probability distribution, such as logic, DNN, CNN, RNN and the like.
In an embodiment, the determining unit 42 is specifically configured to:
taking the category with the highest probability in the action classification probability vector as an action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
In actual application, as shown in fig. 2, a two-dimensional information dimension table is formed according to actions (rows) and services (columns) of a user question, and answers corresponding to the questions, namely service logic, are filled in the middle part. Such as: the introduction of the line shall be the introduction of the call details, the introduction of the network access rules, the introduction of the passwords, and the like. To handle this line, the handling method should be filled out or the user should be guided directly to handle the corresponding service.
It should be noted that: the two-dimensional table does not have knowledge of every grid corresponding to (n, m) coordinates. For example, assume that the action is a transaction and the service is a package, so there are: the + package is handled. But no action is transacted and the service is detailed, i.e. there is no: the + details are handled because this is not a legitimate business logic. Only the query + details, or the introduction + details are legal business logic.
Here, in the scheme of the embodiment of the present invention, when determining the action classification tag and the service classification tag corresponding to the user problem, the action classification tag and the service classification tag corresponding to the highest confidence level are used first to search the corresponding service logic in the information dimension table, and when finding the corresponding service logic, the response processing on the user problem is completed according to the determined service logic; when the corresponding service logic is not found, the action classification label and the service classification label corresponding to the second highest confidence level are used for searching the corresponding service logic in the information dimension table, and when the corresponding service logic is found, the response processing to the user problem is completed according to the determined service logic; and when the corresponding service logic is not found, searching the corresponding service logic in the information dimension table by using the action classification label and the service classification label corresponding to the third high confidence coefficient, and repeating the steps until the service logic corresponding to the user problem is found.
Based on this, in an embodiment, the determining unit 42 is further configured to determine the second highest confidence level of the user problem by using an action classification probability vector and a service classification probability vector when the action classification label corresponding to the highest confidence level and the service logic corresponding to the service classification label are not found in the information dimension table;
the searching unit 43 is further configured to search the action classification label and the service classification label corresponding to the second highest confidence level of the user question in the information dimension table, and determine a service logic corresponding to the user question; and according to the determined service logic, completing the response to the user question;
when the action classification label corresponding to the second highest confidence level and the service logic corresponding to the service classification label are not found in the information dimension table, the determining unit 42 determines the third highest confidence level of the user problem by using the action classification probability vector and the service classification probability vector, and the searching unit 43 searches the corresponding service logic, and so on, until the service logic corresponding to the user problem is found.
The determining unit 42 determines the confidence of the user problem by using the motion classification probability vector and the traffic classification probability vector.
Specifically, the determining unit 42 multiplies each motion classification probability in the motion classification probability vector by a first value; the first value characterizes actions and business ticket adjustment parameters;
the determining unit 42 multiplies each traffic classification probability in the traffic classification probability vector by the second value; the sum of the second value and the first value meets a preset condition;
the determining unit 42 sums the product of each action classification probability and the first value with the product of each service classification probability and the second value to form a set;
when determining the service logic corresponding to the user question, the determining unit 42 sequentially selects a summation result from the set according to the sequence of the summation result from large to small, the category of the action classification probability corresponding to the selected summation result is the action classification label corresponding to the user question, and the category of the service classification probability corresponding to the selected summation result is the service classification label corresponding to the user question.
Wherein the preset conditions are as follows: the sum of the first value and the second value is equal to 1.
In the above process, the formula for determining the second highest confidence level includes: MAX (rate active + (1-rate) attribute) is maximized on the condition that the action classification probability is not equal to the maximum action classification probability or the traffic classification probability is not equal to the maximum traffic classification probability. Accordingly, similar formulas may be employed for determining other confidence levels. That is, in the searching process, the action classification probability and the corresponding service classification probability corresponding to the maximum value of the summation result of the action classification probability and the corresponding service classification probability except the used action classification probability and the corresponding service classification probability are taken each time, and the corresponding service logic is searched in the information dimension table.
Here, rate denotes a first value, 1-rate denotes a second value; ACTIONn denotes an action classification probability and endityn denotes a traffic classification probability.
The used action classification probability and the corresponding service classification probability refer to: and determining the action classification probability and the corresponding service classification probability of the corresponding service logic which is not found in the information dimension table.
The processing unit 44 is specifically configured to:
and outputting answers to the user questions or guiding the user to complete the handling of the corresponding services according to the service logic.
It should be noted that: in the service processing apparatus provided in the foregoing embodiment, when performing service processing, only the division of each program module is illustrated, and in practical applications, the processing allocation may be completed by different program modules according to needs, that is, the internal structure of the apparatus is divided into different program modules to complete all or part of the processing described above.
Correspondingly, an embodiment of the present invention further provides a service processing apparatus, as shown in fig. 5, where the service processing apparatus 50 includes: a processor 51 and a memory 52 for storing computer programs capable of running on the processor,
wherein, the processor 51 is configured to execute, when running the computer program, the following steps:
acquiring a user question;
forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector;
searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services;
and finishing the response processing of the user question according to the determined service logic.
In an embodiment, the processor 51 is configured to, when running the computer program, perform:
taking the category with the highest probability in the action classification probability vector as an action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
Here, the processor 51 is configured to execute, when running the computer program, the following steps:
when the action classification label corresponding to the highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the second highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector;
searching an action classification label and a service classification label corresponding to the second highest confidence level of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; and according to the determined service logic, completing the response to the user question;
and when the action classification label corresponding to the second highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the third highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector, searching the corresponding business logic, and so on until the business logic corresponding to the user problem is found.
In an embodiment, the processor 51 is further configured to execute, when running the computer program, the following:
when the confidence of the user problem is determined, taking the product of each action classification probability in the action classification probability vector and a first value; the first value characterizes actions and business ticket adjustment parameters;
taking the product of each service classification probability in the service classification probability vector and the second value; the sum of the second value and the first value meets a preset condition;
summing the product of each action classification probability and the first value with the product of each service classification probability and the second value to form a set;
when the business logic corresponding to the user problem is determined, one summation result is sequentially selected from the set according to the sequence of the summation results from large to small, the category of the action classification probability corresponding to the selected summation result is the action classification label corresponding to the user problem, and the category of the business classification probability corresponding to the selected summation result is the business classification label corresponding to the user problem.
In an embodiment, the processor 51 is configured to execute, when running the computer program, the following steps:
and outputting answers to the user questions or guiding the user to complete the handling of the corresponding services according to the service logic.
In an embodiment, the processor 51 is further configured to execute, when running the computer program, the following:
and performing word segmentation processing on the user problem before forming an action classification probability vector and a service classification probability vector.
Of course, in practical applications, as shown in fig. 5, the apparatus 50 may further include: at least one network interface 53. The various components in the business processing apparatus 50 are coupled together by a bus system 54. It will be appreciated that the bus system 54 is used to enable communications among the components. The bus system 54 includes a power bus, a control bus, and a status signal bus in addition to the data bus. For clarity of illustration, however, the various buses are labeled as bus system 54 in fig. 5.
It will be appreciated, however, that the memory 52 can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. Among them, the nonvolatile Memory may be a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a magnetic random access Memory (FRAM), a Flash Memory (Flash Memory), a magnetic surface Memory, an optical disk, or a Compact Disc Read-Only Memory (CD-ROM); the magnetic surface storage may be disk storage or tape storage. Volatile Memory can be Random Access Memory (RAM), which acts as external cache Memory. By way of illustration and not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), Enhanced Synchronous Dynamic Random Access Memory (Enhanced DRAM), Synchronous Dynamic Random Access Memory (SLDRAM), Direct Memory (DRmb Access), and Random Access Memory (DRAM). The memory 52 described in connection with the embodiments of the invention is intended to comprise, without being limited to, these and any other suitable types of memory.
The memory 52 in the embodiment of the present invention is used to store various types of data to support the operation of the service processing device 50. Examples of such data include: any computer program for operating on the business processing apparatus 50, such as an operating system 521 and application programs 522. The operating system 521 includes various system programs, such as a framework layer, a core library layer, a driver layer, and the like, for implementing various basic services and processing hardware-based tasks. The application 522 may include various applications for implementing various application services. A program implementing the method of an embodiment of the present invention may be included in application 522.
The method disclosed in the above embodiments of the present invention may be applied to the processor 51, or implemented by the processor 51. The processor 51 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 51. The Processor 51 may be a general purpose Processor, a Digital Signal Processor (DSP), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. The processor 51 may implement or perform the methods, steps, and logic blocks disclosed in embodiments of the present invention. A general purpose processor may be a microprocessor or any conventional processor or the like. The steps of the method disclosed by the embodiment of the invention can be directly implemented by a hardware decoding processor, or can be implemented by combining hardware and software modules in the decoding processor. The software modules may be located in a storage medium located in the memory 52, and the processor 51 reads the information in the memory 52 and performs the steps of the aforementioned method in conjunction with its hardware.
In an exemplary embodiment, the service processing Device 50 may be implemented by one or more Application Specific Integrated Circuits (ASICs), DSPs, Programmable Logic Devices (PLDs), Complex Programmable Logic Devices (CPLDs), Field Programmable Gate Arrays (FPGAs), general purpose processors, controllers, Micro Controllers (MCUs), microprocessors (microprocessors), or other electronic components for performing the foregoing methods.
In an exemplary embodiment, the present invention further provides a computer readable storage medium, such as a memory 52, including a computer program, which can be executed by the processor 51 of the service processing apparatus to implement the steps of the foregoing method. The computer readable storage medium may be Memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface Memory, optical disk, or CD-ROM.
Specifically, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the computer program performs:
acquiring a user question;
forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector;
searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services;
and finishing the response processing of the user question according to the determined service logic.
In one embodiment, the computer program, when executed by the processor, performs:
the determining the action classification label and the service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector includes:
taking the category with the highest probability in the action classification probability vector as an action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
Wherein, in an embodiment, when executed by the processor, the computer program further performs:
when the action classification label corresponding to the highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the second highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector;
searching an action classification label and a service classification label corresponding to the second highest confidence level of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; and according to the determined service logic, completing the response to the user question;
and when the action classification label corresponding to the second highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the third highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector, searching the corresponding business logic, and so on until the business logic corresponding to the user problem is found.
In one embodiment, the computer program, when executed by the processor, performs:
when the confidence of the user problem is determined, taking the product of each action classification probability in the action classification probability vector and a first value; the first value characterizes actions and business ticket adjustment parameters;
taking the product of each service classification probability in the service classification probability vector and the second value; the sum of the second value and the first value meets a preset condition;
summing the product of each action classification probability and the first value with the product of each service classification probability and the second value to form a set;
when the business logic corresponding to the user problem is determined, one summation result is sequentially selected from the set according to the sequence of the summation results from large to small, the category of the action classification probability corresponding to the selected summation result is the action classification label corresponding to the user problem, and the category of the business classification probability corresponding to the selected summation result is the business classification label corresponding to the user problem.
In one embodiment, the computer program, when executed by the processor, performs:
and outputting answers to the user questions or guiding the user to complete the handling of the corresponding services according to the service logic.
In one embodiment, the computer program, when executed by the processor, further performs:
and performing word segmentation processing on the user problem before forming an action classification probability vector and a service classification probability vector.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention.
Claims (13)
1. A method for processing a service, the method comprising:
acquiring a user question;
forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
determining an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector;
searching an action classification label and a service classification label of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services;
according to the determined service logic, completing response processing to the user question; wherein the content of the first and second substances,
the forming of the action classification probability vector by the user question through an action classifier and the forming of the service classification probability vector by the user question through a service classifier comprise one of the following steps:
firstly, the user problem passes through an action classifier to form an action classification probability vector; then, the user problem passes through a service classifier to form a service classification probability vector;
firstly, the user problem passes through a service classifier to form a service classification probability vector; then, the user question is processed by an action classifier to form an action classification probability vector;
and simultaneously, forming an action classification probability vector and a service classification probability vector by the user problem through an action classifier and a service classifier.
2. The method of claim 1, wherein determining the action class label and the traffic class label corresponding to the user question by using the action class probability vector and the traffic class probability vector comprises:
taking the category with the highest probability in the action classification probability vector as an action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
3. The method of claim 2, further comprising:
when the action classification label corresponding to the highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the second highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector;
searching an action classification label and a service classification label corresponding to the second highest confidence level of the user problem in an information dimension table, and determining a service logic corresponding to the user problem; and according to the determined service logic, completing the response to the user question;
and when the action classification label corresponding to the second highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, determining the third highest confidence level of the user problem by utilizing the action classification probability vector and the business classification probability vector, searching the corresponding business logic, and so on until the business logic corresponding to the user problem is found.
4. The method of claim 3, wherein the determining the confidence level of the user question comprises:
taking the product of each action classification probability in the action classification probability vector and the first value; the first value characterizes actions and business ticket adjustment parameters;
taking the product of each service classification probability in the service classification probability vector and the second value; the sum of the second value and the first value meets a preset condition;
summing the product of each action classification probability and the first value with the product of each service classification probability and the second value to form a set;
when the business logic corresponding to the user problem is determined, one summation result is sequentially selected from the set according to the sequence of the summation results from large to small, the category of the action classification probability corresponding to the selected summation result is the action classification label corresponding to the user problem, and the category of the business classification probability corresponding to the selected summation result is the business classification label corresponding to the user problem.
5. The method according to claim 1, wherein the completing the response processing to the user question according to the determined action and the service logic corresponding to the service comprises:
and outputting answers to the user questions or guiding the user to complete the handling of the corresponding services according to the service logic.
6. The method of any of claims 1 to 5, wherein prior to forming the action classification probability vector and the traffic classification probability vector, the method further comprises:
and performing word segmentation processing on the user question.
7. A traffic processing apparatus, characterized in that the apparatus comprises:
an acquisition unit for acquiring a user question;
the classification unit is used for forming an action classification probability vector by the user question through an action classifier; forming a service classification probability vector by the user problem through a service classifier;
a determining unit, configured to determine an action classification label and a service classification label corresponding to the user problem by using the action classification probability vector and the service classification probability vector;
the searching unit is used for searching the action classification label and the service classification label of the user problem in the information dimension table and determining the service logic corresponding to the user problem; the information dimension table is a two-dimensional table formed by actions and services;
the processing unit is used for finishing response processing to the user question according to the determined service logic; wherein the content of the first and second substances,
the classification unit is specifically configured to perform one of:
firstly, the user problem passes through an action classifier to form an action classification probability vector; then, the user problem passes through a service classifier to form a service classification probability vector;
firstly, the user problem passes through a service classifier to form a service classification probability vector; then, the user question is processed by an action classifier to form an action classification probability vector;
and simultaneously, forming an action classification probability vector and a service classification probability vector by the user problem through an action classifier and a service classifier.
8. The apparatus according to claim 7, wherein the determining unit is specifically configured to:
taking the category with the highest probability in the action classification probability vector as an action classification label of the user question; and the class with the highest probability in the service classification probability vector is used as a service classification label of the user problem; the maximum probability in the action classification probability vector and the maximum probability in the service classification probability vector form the highest confidence of the user problem.
9. The apparatus of claim 8,
the determining unit is further configured to determine the next highest confidence level of the user problem by using the action classification probability vector and the service classification probability vector when the action classification label corresponding to the highest confidence level and the service logic corresponding to the service classification label are not found in the information dimension table;
the searching unit is also used for searching an action classification label and a service classification label corresponding to the second highest confidence level of the user problem in the information dimension table and determining the service logic corresponding to the user problem; and according to the determined service logic, completing the response to the user question;
when the action classification label corresponding to the second highest confidence level and the business logic corresponding to the business classification label are not found in the information dimension table, the determining unit determines the third highest confidence level of the user problem by using the action classification probability vector and the business classification probability vector, and the searching unit searches the corresponding business logic, and so on until the business logic corresponding to the user problem is found.
10. The apparatus according to claim 7, wherein the processing unit is specifically configured to:
and outputting answers to the user questions or guiding the user to complete the handling of the corresponding services according to the service logic.
11. The apparatus of any one of claims 7 to 10, further comprising:
and the word segmentation unit is used for carrying out word segmentation processing on the user question.
12. A traffic processing apparatus, characterized in that the apparatus comprises: a processor and a memory for storing a computer program capable of running on the processor,
wherein the processor is adapted to perform the steps of the method of any one of claims 1 to 6 when running the computer program.
13. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710391804.2A CN108959327B (en) | 2017-05-27 | 2017-05-27 | Service processing method, device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710391804.2A CN108959327B (en) | 2017-05-27 | 2017-05-27 | Service processing method, device and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108959327A CN108959327A (en) | 2018-12-07 |
| CN108959327B true CN108959327B (en) | 2021-03-05 |
Family
ID=64494830
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710391804.2A Active CN108959327B (en) | 2017-05-27 | 2017-05-27 | Service processing method, device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108959327B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109840274B (en) * | 2018-12-28 | 2021-11-30 | 北京百度网讯科技有限公司 | Data processing method and device and storage medium |
| CN111008268B (en) * | 2019-10-31 | 2021-05-18 | 支付宝(杭州)信息技术有限公司 | Method and device for acquiring question reversing sentence corresponding to user question based on dialog system |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101320374A (en) * | 2008-07-10 | 2008-12-10 | 昆明理工大学 | Field question classification method combining syntax structural relationship and field characteristic |
| CN103533186A (en) * | 2013-09-23 | 2014-01-22 | 安徽科大讯飞信息科技股份有限公司 | Service flow process realization method and system based on voice calling |
| CN104050256A (en) * | 2014-06-13 | 2014-09-17 | 西安蒜泥电子科技有限责任公司 | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method |
| CN104050224A (en) * | 2013-03-15 | 2014-09-17 | 国际商业机器公司 | Combining different type coercion components for deferred type evaluation |
| CN104598445A (en) * | 2013-11-01 | 2015-05-06 | 腾讯科技（深圳）有限公司 | Automatic question-answering system and method |
| CN104679826A (en) * | 2015-01-09 | 2015-06-03 | 北京京东尚科信息技术有限公司 | Classification model-based context recognition method and system |
| CN105630827A (en) * | 2014-11-05 | 2016-06-01 | 阿里巴巴集团控股有限公司 | Information processing method and system, and auxiliary system |
| CN106294341A (en) * | 2015-05-12 | 2017-01-04 | 阿里巴巴集团控股有限公司 | A kind of Intelligent Answer System and theme method of discrimination thereof and device |
-
2017
- 2017-05-27 CN CN201710391804.2A patent/CN108959327B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101320374A (en) * | 2008-07-10 | 2008-12-10 | 昆明理工大学 | Field question classification method combining syntax structural relationship and field characteristic |
| CN104050224A (en) * | 2013-03-15 | 2014-09-17 | 国际商业机器公司 | Combining different type coercion components for deferred type evaluation |
| CN103533186A (en) * | 2013-09-23 | 2014-01-22 | 安徽科大讯飞信息科技股份有限公司 | Service flow process realization method and system based on voice calling |
| CN104598445A (en) * | 2013-11-01 | 2015-05-06 | 腾讯科技（深圳）有限公司 | Automatic question-answering system and method |
| CN104050256A (en) * | 2014-06-13 | 2014-09-17 | 西安蒜泥电子科技有限责任公司 | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method |
| CN105630827A (en) * | 2014-11-05 | 2016-06-01 | 阿里巴巴集团控股有限公司 | Information processing method and system, and auxiliary system |
| CN104679826A (en) * | 2015-01-09 | 2015-06-03 | 北京京东尚科信息技术有限公司 | Classification model-based context recognition method and system |
| CN106294341A (en) * | 2015-05-12 | 2017-01-04 | 阿里巴巴集团控股有限公司 | A kind of Intelligent Answer System and theme method of discrimination thereof and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108959327A (en) | 2018-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111125334B (en) | Search question-answering system based on pre-training | |
| KR102288249B1 (en) | Information processing method, terminal, and computer storage medium | |
| US9606984B2 (en) | Unsupervised clustering of dialogs extracted from released application logs | |
| US11238050B2 (en) | Method and apparatus for determining response for user input data, and medium | |
| CN111339277A (en) | Question-answer interaction method and device based on machine learning | |
| CN111046147A (en) | Question answering method and device and terminal equipment | |
| CN112035599A (en) | Query method and device based on vertical search, computer equipment and storage medium | |
| CN113590778A (en) | Intelligent customer service intention understanding method, device, equipment and storage medium | |
| US11120214B2 (en) | Corpus generating method and apparatus, and human-machine interaction processing method and apparatus | |
| CN115470338B (en) | Multi-scenario intelligent question answering method and system based on multi-path recall | |
| CN112632239A (en) | Brain-like question-answering system based on artificial intelligence technology | |
| CN108959327B (en) | Service processing method, device and computer readable storage medium | |
| CN106250366B (en) | A kind of data processing method and system for question answering system | |
| CN114282513A (en) | Text semantic similarity matching method and system, intelligent terminal and storage medium | |
| TW202034207A (en) | Dialogue system using intention detection ensemble learning and method thereof | |
| CN116680368B (en) | Water conservancy knowledge question-answering method, device and medium based on Bayesian classifier | |
| CN117493524A (en) | Intelligent question-answering control method, device, equipment and storage medium | |
| CN109684357B (en) | Information processing method and device, storage medium and terminal | |
| CN112148939A (en) | Data processing method and device and electronic equipment | |
| CN116186219A (en) | Man-machine dialogue interaction method, system and storage medium | |
| CN113095073B (en) | Corpus tag generation method and device, computer equipment and storage medium | |
| CN114492452A (en) | Method, device and equipment for training and appealing switching of pre-training language model | |
| CN112579774A (en) | Model training method, model training device and terminal equipment | |
| CN111708862A (en) | Text matching method and device and electronic equipment | |
| CN117874210B (en) | Autonomous interactive question-answering method, system, equipment and 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 |