CN113988190B - Customer intention analysis method, device, equipment and storage medium - Google Patents
Customer intention analysis method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN113988190B CN113988190B CN202111273716.5A CN202111273716A CN113988190B CN 113988190 B CN113988190 B CN 113988190B CN 202111273716 A CN202111273716 A CN 202111273716A CN 113988190 B CN113988190 B CN 113988190B
- Authority
- CN
- China
- Prior art keywords
- customer
- intention
- data
- preset
- voice
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to the field of artificial intelligence, and discloses a method, a device, equipment and a storage medium for analyzing customer intention, which are used for improving the efficiency of customer intention analysis. The customer intention analysis method comprises the steps of preprocessing customer history handling data, classifying the preprocessed history data based on a preset decision tree algorithm to obtain a intention classification standard, configuring intention rules based on a flow speech template to obtain customer intention rules, receiving voice interaction data returned by an artificial intelligent voice robot, comparing the voice interaction data with the customer intention rules to obtain a voice comparison result, extracting offline interaction data in the voice comparison result, matching the offline interaction data with preset customer intention labels to obtain a customer intention matching result, and classifying the voice comparison result and the customer intention matching result to obtain a customer intention analysis result. In addition, the invention also relates to a blockchain technology, and the client intention analysis result can be stored in the blockchain.
Description
Technical Field
The present invention relates to the field of similarity matching, and in particular, to a method, apparatus, device, and storage medium for analyzing intent of a client.
Background
The client intention analysis is mainly used for training a client intention analysis model through learning all voice data after the current artificial intelligence AI calls outwards, corresponding calling conditions, basic information, credit card service conditions and the like of the client, so that clients with handling intention are mined out, and the handling rate is improved.
The traditional client intent analysis method comprises the steps of firstly, manually analyzing interactive text of the AI and the client by hearing the AI outbound record, extracting keywords and intent in client interaction, and then, carrying out follow-up on the intent client in an offline data acquisition mode, and secondly, setting intent and nodes of the intent client in advance when designing an AI dialogue flow, and defining the clients as the intent client if the client triggers preset intent nodes in actual interaction. However, the analysis process in the above two ways is complex and time-consuming, requiring analysis of the full volume of interaction data, resulting in inefficiency in customer intent analysis.
Disclosure of Invention
The invention provides a client intention analysis method, device, equipment and storage medium, which are used for classifying preprocessing historical data based on a preset decision tree algorithm to obtain a intention classification standard, comparing voice interaction data with client intention rules to obtain a voice comparison result, extracting offline interaction data in the voice comparison result, matching the offline interaction data with a preset client intention degree label to obtain a client intention matching result, classifying the voice comparison result and the client intention degree matching result based on the intention classification standard to obtain a client intention analysis result, and improving the efficiency of client intention analysis.
The invention provides a client intention analysis method, which comprises the steps of obtaining client history handling data, preprocessing the client history handling data to obtain preprocessing history data, classifying the preprocessing history data based on a preset decision tree algorithm to obtain a intention grading standard, configuring intention rules based on a preset flow speech template to obtain client intention rules, receiving voice interaction data returned by an artificial intelligent voice robot, comparing the voice interaction data with the client intention rules to obtain a voice comparison result, extracting offline interaction data in the voice comparison result, matching the offline interaction data with preset client intention labels to obtain a client intention matching result, and classifying the voice comparison result and the client intention matching result based on the intention grading standard to obtain a client intention analysis result.
Optionally, in a first implementation manner of the first aspect of the present invention, the obtaining client history handling data, preprocessing the client history handling data to obtain preprocessing history data, classifying the preprocessing history data based on a preset decision tree algorithm to obtain a willingness grading standard includes obtaining client history handling data, performing deficiency value completion, outlier filtering and repeated value filtering on the client history handling data to obtain preprocessing history data, invoking a preset decision tree algorithm, performing traversal processing on the preprocessing history data to obtain a target decision tree, wherein the target decision tree includes a plurality of leaf nodes, obtaining a business handling rate corresponding to each leaf node in the target decision tree, sorting each leaf node according to a sequence of the business handling rate from large to small to obtain a leaf node sorting result, and classifying the leaf node sorting result according to a preset client quantity grading standard to obtain the willingness grading standard.
Optionally, in a second implementation manner of the first aspect of the present invention, the invoking the preset decision tree algorithm performs traversal processing on the preprocessing history data to obtain a target decision tree, where the target decision tree includes a plurality of leaf nodes, and the traversal processing on the preprocessing history data is performed to obtain a traversal result, and the traversal result is classified according to preset guest group features to obtain an initial decision tree, and pruning processing is performed on the initial decision tree to obtain a target decision tree, where the target decision tree includes a plurality of leaf nodes, and each leaf node corresponds to one guest group feature.
Optionally, in a third implementation manner of the first aspect of the present invention, configuring intent rules based on the preset flow-phone template to obtain customer intent rules, receiving voice interaction data returned by the artificial intelligent voice robot, and comparing the voice interaction data with the customer intent rules to obtain a voice comparison result includes obtaining the flow-phone template, configuring intent rules based on a plurality of rule types in the flow-phone template to obtain customer intent rules, receiving voice interaction data returned by the artificial intelligent voice robot to obtain a first matching degree between the voice interaction data and the customer intent rules, and calling a preset comparison algorithm to judge whether the first matching degree is greater than a preset first matching threshold to obtain a voice comparison result.
Optionally, in a fourth implementation manner of the first aspect of the present invention, extracting the offline interaction data in the voice comparison result, and matching the offline interaction data with a preset client intent label to obtain a client intent matching result includes extracting the voice interaction data in the voice comparison result, where the first matching degree is less than or equal to a preset first matching threshold, to obtain the offline interaction data, the first matching degree is a matching degree between the voice interaction data and the client intent rule, obtaining a second matching degree, calling a preset similarity matching algorithm, and judging whether the second matching degree is greater than the preset second matching threshold, to obtain a client intent matching result, where the second matching degree is a matching degree between the offline interaction data and the preset client intent label.
Optionally, in a fifth implementation manner of the first aspect of the present invention, the classifying the voice comparison result and the client intent matching result based on the intent grading criterion, to obtain the client intent analysis result includes extracting first client data from the voice comparison result, classifying the first client data into a preset first echelon intent grading based on the intent grading criterion to obtain a first classification result, wherein the first client data is voice interaction data with a first matching degree greater than a first matching threshold, extracting second client data from the client intent matching result, classifying the second client data into a preset second echelon intent grading based on the intent grading criterion to obtain a second classification result, wherein the second client data is offline interaction data with a second matching degree greater than a second matching threshold in the client intent matching result, and determining the first classification result and the second classification result as the client intent analysis result.
Optionally, in a sixth implementation manner of the first aspect of the present invention, when the obtaining of the client history handling data, preprocessing the client history handling data to obtain preprocessed history data, classifying the preprocessed history data based on a preset decision tree algorithm, and before obtaining the willingness classification standard, the client intent analysis method further includes obtaining client to-be-detected data, and sending the client to-be-detected data to an artificial intelligent voice robot, so that the artificial intelligent voice robot performs voice interaction to obtain voice interaction data.
The invention provides a client intention analysis device which comprises an acquisition module, a comparison module, a matching module and a classification module, wherein the acquisition module is used for acquiring client history handling data, preprocessing the client history handling data to obtain preprocessing history data, classifying the preprocessing history data based on a preset decision tree algorithm to obtain a intention classification standard, the comparison module is used for configuring intention rules based on a preset flow speech template to obtain client intention rules, receiving voice interaction data returned by an artificial intelligent voice robot, comparing the voice interaction data with the client intention rules to obtain a voice comparison result, the matching module is used for extracting offline interaction data in the voice comparison result, matching the offline interaction data with a preset client intention label to obtain a client intention matching result, and classifying the voice comparison result and the client intention matching result based on the intention classification standard to obtain a client intention analysis result.
Optionally, in a first implementation manner of the second aspect of the present invention, the obtaining module includes a preprocessing unit, a traversing unit, and a sorting unit, where the preprocessing unit is used to obtain a target decision tree, the target decision tree includes a plurality of leaf nodes, the sorting unit is used to obtain a business handling rate corresponding to each leaf node in the target decision tree, sort each leaf node according to a business handling rate from large to small, obtain a leaf node sorting result, and the sorting unit is used to sort the leaf node sorting result according to a preset client quantity sorting standard, so as to obtain a willingness sorting standard.
Optionally, in a second implementation manner of the second aspect of the present invention, the traversing unit is specifically configured to perform traversing processing on the preprocessing history data to obtain a traversing result, classify the traversing result according to a preset guest group feature to obtain an initial decision tree, and perform pruning processing on the initial decision tree to obtain a target decision tree, where the target decision tree includes a plurality of leaf nodes, and each leaf node corresponds to one guest group feature.
Optionally, in a third implementation manner of the second aspect of the present invention, the comparison module includes a configuration unit, a receiving unit, a first judging unit and a second judging unit, wherein the configuration unit is used for obtaining a procedure call template, configuring intent rules based on a plurality of rule categories in the procedure call template to obtain customer intent rules, the receiving unit is used for receiving voice interaction data returned by an artificial intelligent voice robot, obtaining first matching degree between the voice interaction data and the customer intent rules, and the first judging unit is used for calling a preset comparison algorithm to judge whether the first matching degree is greater than a preset first matching threshold value or not to obtain a voice comparison result.
Optionally, in a fourth implementation manner of the second aspect of the present invention, the matching module includes an extracting unit, configured to extract voice interaction data with a first matching degree smaller than or equal to a preset first matching threshold in the voice comparison result to obtain offline interaction data, where the first matching degree is a matching degree between the voice interaction data and the client intent rule, and a second judging unit, configured to obtain a second matching degree, invoke a preset similarity matching algorithm, and judge whether the second matching degree is greater than the preset second matching threshold to obtain a client intent matching result, where the second matching degree is a matching degree between the offline interaction data and a preset client intent label.
Optionally, in a fifth implementation manner of the second aspect of the present invention, the classification module includes a first division unit, configured to extract first customer data from the voice comparison result, divide the first customer data into a preset first echelon intent classification based on the intent classification standard to obtain a first classification result, where the first customer data is voice interaction data with a first matching degree greater than a first matching threshold, and a second division unit, configured to extract second customer data from the customer intent matching result, divide the second customer data into a preset second echelon intent classification based on the intent classification standard to obtain a second classification result, where the second customer data is offline interaction data with a second matching degree greater than a second matching threshold in the customer intent matching result, and determine the first classification result and the second classification result as customer intent analysis results.
Optionally, in a sixth implementation manner of the second aspect of the present invention, before the obtaining module, the client intent analysis device further includes a voice interaction module, including obtaining client to-be-tested data, and sending the client to-be-tested data to an artificial intelligent voice robot, so that the artificial intelligent voice robot performs voice interaction to obtain voice interaction data.
The third aspect of the invention provides a customer intent analysis device comprising a memory and at least one processor, wherein the memory stores a computer program, and the at least one processor invokes the computer program in the memory to cause the customer intent analysis device to execute the customer intent analysis method.
A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored therein, which when run on a computer, causes the computer to perform the above-described client intent analysis method.
According to the technical scheme, client history handling data are obtained, preprocessing is conducted on the client history handling data to obtain preprocessing history data, the preprocessing history data are classified based on a preset decision tree algorithm to obtain willingness grading standards, intention rules are configured based on a preset flow speech template to obtain client intention rules, voice interaction data returned by an artificial intelligent voice robot are received, the voice interaction data are compared with the client intention rules to obtain voice comparison results, offline interaction data in the voice comparison results are extracted, the offline interaction data are matched with preset client intention labels to obtain client intention matching results, and the voice comparison results and the client intention matching results are classified based on the willingness grading standards to obtain client intention analysis results. In the embodiment of the invention, based on a preset decision tree algorithm, preprocessing historical data is classified to obtain a willingness grading standard, voice interaction data is compared with customer intention rules to obtain a voice comparison result, offline interaction data in the voice comparison result is extracted, the offline interaction data is matched with a preset customer intention degree label to obtain a customer intention matching result, the voice comparison result and the customer intention degree matching result are respectively classified based on the willingness grading standard to obtain a customer intention analysis result, and communication nodes and keywords can be extracted effectively aiming at customers which are leaked in the interaction process to analyze the purchase intention of the customer on products, so that the efficiency of customer intention analysis is improved.
Drawings
FIG. 1 is a schematic diagram of an embodiment of a method for analyzing intent of a user according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of another embodiment of a method for analyzing intent of a user according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of an embodiment of a client intent analysis device;
FIG. 4 is a schematic diagram of another embodiment of a client intent analysis device in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of an embodiment of a client intent analysis device.
Detailed Description
The embodiment of the invention provides a client intention analysis method, device, equipment and storage medium, which are used for classifying preprocessing historical data based on a preset decision tree algorithm to obtain a intention classification standard, comparing voice interaction data with client intention rules to obtain a voice comparison result, extracting offline interaction data in the voice comparison result, matching the offline interaction data with preset client intention labels to obtain a client intention matching result, classifying the voice comparison result and the client intention matching result based on the intention classification standard to obtain a client intention analysis result, and improving the efficiency of client intention analysis.
The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For ease of understanding, a specific flow of an embodiment of the present invention is described below with reference to fig. 1, where an embodiment of a method for analyzing intent of a user in an embodiment of the present invention includes:
101. The method comprises the steps of obtaining client history handling data, preprocessing the client history handling data to obtain preprocessing history data, and classifying the preprocessing history data based on a preset decision tree algorithm to obtain willingness grading standards.
It will be appreciated that the execution subject of the present invention may be a client intent analysis device, or may be a terminal or a server, which is not limited herein. The embodiment of the invention is described by taking a server as an execution main body as an example. The server may be an independent server, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.
The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Wherein artificial intelligence (ARTIFICIAL INTELLIGENCE, AI) is the theory, method, technique, and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend, and expand human intelligence, sense the environment, acquire knowledge, and use knowledge to obtain optimal results.
Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.
The server acquires client history handling data, preprocesses the client history handling data to obtain preprocessing history data, classifies the preprocessing history data based on a preset decision tree algorithm, and obtains willingness grading standards. The server acquires the client history handling data, the client history handling data is acquired through a crawler, after the client history handling data is acquired through user authorization, the client history handling data is preprocessed firstly, and the preprocessing execution process can be that the server sequentially carries out missing value filling, outlier filtering and repeated value filtering on the client history handling data to obtain preprocessing history data. The method comprises the steps that a server calls a preset decision tree algorithm to traverse the preprocessing history data to obtain a traversing result, the traversing process can be any one or a combination of a plurality of front traversing, middle traversing and rear traversing, the traversing result is classified according to preset guest group characteristics to obtain an initial decision tree, the preset guest group characteristics comprise guest group quality, guest group level, guest group age and the like, after the initial decision tree is generated, pruning processing is needed to be carried out on the initial decision tree to obtain a target decision tree, the target decision tree comprises a plurality of leaf nodes, the server obtains service handling rate corresponding to each leaf node, sorts each leaf node according to the sequence of the service handling rate from large to small to obtain a leaf node sorting result, sorts according to preset client quantity grading standards to obtain willingness grading standards.
102. And configuring intent rules based on a preset flow speech operation template, obtaining customer intent rules, receiving voice interaction data returned by the artificial intelligent voice robot, and comparing the voice interaction data with the customer intent rules to obtain a voice comparison result.
The server configures intent rules based on a preset flow phone operation template, obtains client intent rules, receives voice interaction data returned by the artificial intelligent voice robot, compares the voice interaction data with the client intent rules, and obtains a voice comparison result. The flow speaking template comprises a plurality of rule categories, the rule categories comprise tail nodes, tail intentions, client speaking, stay time of robot nodes, hang-up types, passing times of the robot nodes, intermediate nodes, conversation time and interaction times, the server configures intent rules based on the rule categories to obtain client intent rules, the client intent rules in the embodiment comprise different standards corresponding to each rule category, for example, stay time of the robot nodes is less than 1 minute, conversation time is longer than 10 minutes and the like, voice interaction data returned after the artificial intelligent voice robot calls outwards are received in real time, real-time judgment is carried out on the voice interaction data, and whether the matching degree of the voice interaction data and the client intent rules is greater than a preset first matching threshold value or not is judged through a preset comparison algorithm, so that a voice comparison result is obtained.
103. And extracting offline interaction data in the voice comparison result, and matching the offline interaction data with a preset client intention label to obtain a client intention matching result.
The server extracts offline interaction data in the voice comparison result, and matches the offline interaction data with a preset client intention label to obtain a client intention matching result. The server extracts voice interaction data which does not meet the intention client rule from the voice comparison result to obtain offline interaction data, and performs secondary matching on the offline interaction data and the client intention label by calling a preset similarity matching algorithm to obtain a client intention matching result, wherein a second matching threshold is set in advance by manpower, and the preset similarity algorithm can be an Euclidean metric (Euclidean) algorithm, a Pearson correlation coefficient algorithm or a cosine similarity algorithm, and the client intention label is specially set in advance for the client offline data.
104. Based on willingness grading standards, the voice comparison result and the client intention degree matching result are respectively classified, and a client intention analysis result is obtained.
The server classifies the voice comparison result and the client intention degree matching result based on the willingness grading standard respectively to obtain a client intention analysis result. The classification algorithm applied in the classification process can be a K nearest neighbor algorithm, the server screens first customer data from the voice comparison result, performs secondary screening based on the customer intention degree label, screens second customer data from the customer intention degree matching result, and can extract communication nodes and keywords according to communication contents between the artificial intelligent voice robot and the customer for the lost customer in the interaction process so as to analyze the purchase intention of the customer for the product, and the server sequentially divides the first customer data and the second customer data into intention classification of different teams based on intention classification standards to obtain a customer intention analysis result, wherein the preset first teams and second teams are determined by manual concrete according to the intention classification standards.
In the embodiment of the invention, based on a preset decision tree algorithm, preprocessing historical data is classified to obtain a willingness grading standard, voice interaction data is compared with client intention rules to obtain a voice comparison result, offline interaction data in the voice comparison result is extracted, the offline interaction data is matched with a preset client intention degree label to obtain a client intention matching result, and the voice comparison result and the client intention degree matching result are respectively classified to obtain a client intention analysis result based on the willingness grading standard, so that the efficiency of client intention analysis is improved.
Referring to fig. 2, another embodiment of a method for analyzing intent of a user according to an embodiment of the present invention includes:
201. and acquiring client history handling data, and carrying out missing value completion, outlier filtering and repeated value filtering on the client history handling data to obtain preprocessing history data.
The server acquires the client history handling data, and performs missing value completion, outlier filtering and repeated value filtering on the client history handling data to obtain preprocessing history data. The server acquires client history handling data, the client history handling data is acquired through a crawler, after the client history handling data is acquired through user authorization, the client history handling data is preprocessed firstly, and the preprocessing can be performed by sequentially performing missing value filling, outlier filtering and repeated value filtering on the client history handling data to obtain preprocessed history data, wherein the missing value filling can be multiple interpolation, the outlier filtering mainly adopts an outlier detection algorithm z-score to identify and delete outliers, and the repeated value filtering is repeated value de-duplication processing.
202. And calling a preset decision tree algorithm, and performing traversal processing on the preprocessing history data to obtain a target decision tree, wherein the target decision tree comprises a plurality of leaf nodes.
The server calls a preset decision tree algorithm, and traverses the preprocessing history data to obtain a target decision tree, wherein the target decision tree comprises a plurality of leaf nodes. Specifically, the server performs traversal processing on the preprocessing history data to obtain traversal results, classifies the traversal results according to preset guest group characteristics to obtain an initial decision tree, performs pruning processing on the initial decision tree to obtain a target decision tree, wherein the target decision tree comprises a plurality of leaf nodes, and each leaf node corresponds to one guest group characteristic.
The traversing process may be any one or a combination of a plurality of pre-traversing, middle-traversing and post-traversing, classifying the traversing result according to preset guest group characteristics to obtain an initial decision tree, wherein the preset guest group characteristics include guest group quality, guest group level, guest group age and the like, after the initial decision tree is generated, pruning is required to be performed on the initial decision tree to obtain a target decision tree, the pruning process used in the embodiment is mainly post pruning, the post pruning is from bottom to top, estimating a non-leaf node from bottom to top for a complete decision tree which has been generated, if replacing a subtree corresponding to the node with a leaf node can bring about improvement of the generalization performance of the decision tree, replacing the subtree with the leaf node, the post pruning mainly comprises error rate reduction pruning (REP), pessimistic pruning (pesimistic-pruner), cost complexity pruning (counter-complexity pruning) and error pruning (PEP-base pruning).
203. And acquiring the service handling rate corresponding to each leaf node in the target decision tree, and sequencing each leaf node according to the sequence from the big service handling rate to the small service handling rate to obtain a leaf node sequencing result.
The server obtains the service handling rate corresponding to each leaf node in the target decision tree, and sorts each leaf node according to the sequence from the big service handling rate to the small service handling rate, so as to obtain a leaf node sorting result. The target decision tree comprises a plurality of leaf nodes, the server acquires the service handling rate corresponding to each leaf node, and sequences each leaf node according to the sequence from the higher service handling rate to the lower service handling rate to obtain a leaf node sequencing result.
204. And classifying the leaf node sequencing result according to a preset client quantity grading standard to obtain a willingness grading standard.
And classifying the leaf node sequencing result by the server according to a preset client quantity grading standard to obtain a willingness grading standard. The server classifies according to preset client quantity grading standards to obtain willingness grading standards, wherein the client quantity grading standards are not limited, a classification algorithm applied in the classification process can be a K nearest neighbor algorithm, for example, the client quantity can be classified into one grade according to 20%, 5 grades are shared by the corresponding willingness grading standards, each grade in the willingness grading standards has a corresponding standard grading value, the server classifies leaf node sequencing results to obtain the willingness grading standards by calling the K nearest neighbor algorithm, the standard grading values are used for matching and dividing data subsequently, and the setting rule of the standard grading values is that the lowest business handling rate in each grade willingness grading is extracted, and the standard grading value corresponding to each grade is determined.
205. And configuring intent rules based on a preset flow speech operation template, obtaining customer intent rules, receiving voice interaction data returned by the artificial intelligent voice robot, and comparing the voice interaction data with the customer intent rules to obtain a voice comparison result.
The server configures intent rules based on a preset flow phone operation template, obtains client intent rules, receives voice interaction data returned by the artificial intelligent voice robot, compares the voice interaction data with the client intent rules, and obtains a voice comparison result. The method comprises the steps of obtaining a flow phone operation template by a server, configuring intent rules based on a plurality of rule categories in the flow phone operation template to obtain client intent rules, receiving voice interaction data returned by an artificial intelligent voice robot by the server to obtain first matching degree between the voice interaction data and the client intent rules, and calling a preset comparison algorithm by the server to judge whether the first matching degree is larger than a preset first matching threshold value or not to obtain a voice comparison result.
The flow speech operation template comprises a plurality of rule categories, wherein the rule categories comprise tail nodes, namely a last node for communicating with the artificial intelligent voice robot, tail intention, intention of a client in the last process of communicating, client skill, node stay time of the robot, namely node stay time of a single robot, hang-up type, namely system hang-up and user active hang-up, the number of times of passing through the robot nodes is an integer which is larger than 0, the upper limit value is required to be checked to be larger than the lower limit value, the intermediate node is a plurality of nodes which are passed through in the voice communicating process, the relation among the nodes is that the intention rule of the client is judged when the nodes are simultaneously satisfied, the communication time is the total time of the whole communication, the interaction time of the artificial intelligent voice robot and the client in the communicating process, the server configures the intention rule based on the rule of the plurality of the rule categories, receives the voice interaction data returned after the artificial intelligent voice robot calls in real time, the voice interaction data is compared with the preset intention rule of the client in the voice algorithm, and whether the preset intention rule is matched with the first voice algorithm is matched with the voice algorithm in advance is set to be compared with the first threshold value or not.
206. And extracting offline interaction data in the voice comparison result, and matching the offline interaction data with a preset client intention label to obtain a client intention matching result.
The server extracts offline interaction data in the voice comparison result, and matches the offline interaction data with a preset client intention label to obtain a client intention matching result. The server extracts voice interaction data with the first matching degree smaller than or equal to a preset first matching threshold value in the voice comparison result to obtain offline interaction data, wherein the first matching degree is the matching degree of the voice interaction data and a client intention rule, and calculates the second matching degree and the preset second matching threshold value according to a preset similarity matching algorithm to obtain a client intention matching result, and the second matching degree is the matching degree of the offline interaction data and a preset client intention label.
In the process of communication between a client and an artificial intelligent voice robot, the problems of on-hook in the middle of the client, on-hook in the waiting process, non-connection, communication faults caused by signal reasons and the like may exist, so that voice interaction data do not hit a client intention rule, a server extracts voice interaction data which do not meet the intention client rule from a voice comparison result, off-line interaction data are obtained, the off-line interaction data and a client intention label are subjected to secondary matching by calling a preset similarity matching algorithm, the client intention matching result is obtained, the preset similarity algorithm can be an Euclidean metric algorithm, a Pearson correlation coefficient algorithm or a cosine similarity algorithm, the client intention label is specially set for the client off-line data in advance, for example, the on-hook in the client waiting seat is set to be a high intention label, the non-connection is defined to be a low intention label, and accordingly the client intention label is obtained, and specifically, the matching degree of the off-line interaction data hitting the high intention label and the preset client intention label is higher.
207. Based on willingness grading standards, the voice comparison result and the client intention degree matching result are respectively classified, and a client intention analysis result is obtained.
The server classifies the voice comparison result and the client intention degree matching result based on the willingness grading standard respectively to obtain a client intention analysis result. The method comprises the steps of extracting first client data from a voice comparison result by a server, dividing the first client data into preset first echelon intent classification based on intent classification standards to obtain a first classification result, extracting second client data from the client intent matching result by the server, dividing the second client data into preset second echelon intent classification based on intent classification standards to obtain a second classification result, wherein the second client data is offline interaction data with the second matching degree larger than a second matching threshold in the client intent matching result, and determining the first classification result and the second classification result as client intent analysis results by the server.
The first client data is screened through a client intention rule, secondary screening is carried out based on a client intention degree label to obtain second client data, communication nodes and keywords can be extracted according to communication contents of an artificial intelligent voice robot and the client for the lost client in the interaction process, the purchase intention of the client for products is analyzed, the first client data and the second client data are sequentially divided into intention steps of different echelons based on a intention grading standard, a client intention analysis result is obtained, for example, if the intention grading standard has 5 steps, the first echelon can be a first step and a second step which are in front of the intention grading standard, the second step can be a third step, each step in the intention grading standard has a corresponding standard grading value, the server can also read a first matching degree after dividing the extracted first client data into preset first echelons according to a preset standard grading value-first matching degree corresponding table, and the first client data are respectively divided into different first echelons, for example, and the first client data are divided into different steps according to the preset standard grading value-first matching degree corresponding table: the willingness-to-grade standard has 5 grades, the standard grade value (i.e. the lowest business handling rate) corresponding to each grade is 90%, 75%, 60%, 40%, 15%, the first echelon willingness grade manually set is the first two grades (i.e. the grade corresponding to the standard grade value is 90% and 75%), the server divides the extracted first customer data into the first echelon willingness grade, reads the first matching degree to be 80%, the first matching degree corresponding to the standard grade value 90% is 75% in the standard grade value-first matching degree corresponding table, 80% >75% represents a corresponding ranking with a first customer data hit standard ranking value of 90%. After the server divides the second client data into preset second echelon willingness to be classified, the second matching degree is read, and the second client data are respectively divided into different classification in the second echelon according to a preset standard classification value-second matching degree corresponding table. The offline interactive data (namely the second client data) with the second matching degree larger than the second matching threshold value in the client intention degree matching result is screened for the second time, so that the leaked data in interaction can be subjected to the second follow-up, the client intention is analyzed more accurately, the client handling rate is improved, the first client data and the second client data hitting the client intention rule in the voice interactive result are divided into different echelons of the intention grading standard and different wish grading in different echelons, the follow-up processing can be performed on the intention clients corresponding to the different client data according to the different grading in the client intention analysis result in a targeted manner, for example, manual second follow-up is performed, short messages are sent to the clients or intention confirming mails are sent to the clients, and the follow-up frequency can be specifically adjusted for the different echelons in the intention grading standard, for example, the follow-up frequency of the client corresponding to the intention grading of the first echelon is higher.
In the embodiment of the invention, based on a preset decision tree algorithm, preprocessing historical data is classified to obtain a willingness grading standard, voice interaction data is compared with client intention rules to obtain a voice comparison result, offline interaction data in the voice comparison result is extracted, the offline interaction data is matched with a preset client intention degree label to obtain a client intention matching result, and the voice comparison result and the client intention degree matching result are respectively classified to obtain a client intention analysis result based on the willingness grading standard, so that the efficiency of client intention analysis is improved.
The method for analyzing the intention of the user in the embodiment of the present invention is described above, and the apparatus for analyzing the intention of the user in the embodiment of the present invention is described below, referring to fig. 3, an embodiment of the apparatus for analyzing the intention of the user in the embodiment of the present invention includes:
The acquiring module 301 is configured to acquire client history handling data, preprocess the client history handling data to obtain preprocessed history data, and classify the preprocessed history data based on a preset decision tree algorithm to obtain a willingness classification standard;
The comparison module 302 is configured to configure intent rules based on a preset flow speech template, obtain customer intent rules, receive voice interaction data returned by the artificial intelligent voice robot, and compare the voice interaction data with the customer intent rules to obtain a voice comparison result;
the matching module 303 is configured to extract offline interaction data in the voice comparison result, and match the offline interaction data with a preset client intent label to obtain a client intent matching result;
the classification module 304 is configured to classify the voice comparison result and the customer intent matching result based on the willingness classification standard, respectively, to obtain a customer intent analysis result.
In the embodiment of the invention, based on a preset decision tree algorithm, preprocessing historical data is classified to obtain a willingness grading standard, voice interaction data is compared with client intention rules to obtain a voice comparison result, offline interaction data in the voice comparison result is extracted, the offline interaction data is matched with a preset client intention degree label to obtain a client intention matching result, and the voice comparison result and the client intention degree matching result are respectively classified to obtain a client intention analysis result based on the willingness grading standard, so that the efficiency of client intention analysis is improved.
Referring to fig. 4, another embodiment of the apparatus for analyzing intent of a user according to an embodiment of the present invention includes:
The acquiring module 301 is configured to acquire client history handling data, preprocess the client history handling data to obtain preprocessed history data, and classify the preprocessed history data based on a preset decision tree algorithm to obtain a willingness classification standard;
Specifically, the acquiring module 301 includes:
The preprocessing unit 3011 is configured to obtain client history handling data, and perform missing value completion, outlier filtering and repeated value filtering on the client history handling data to obtain preprocessing history data;
the traversing unit 3012 is used for calling a preset decision tree algorithm, traversing the preprocessing history data to obtain a target decision tree, wherein the target decision tree comprises a plurality of leaf nodes;
The sorting unit 3013 is configured to obtain a service handling rate corresponding to each leaf node in the target decision tree, and sort each leaf node according to a sequence from the higher service handling rate to the lower service handling rate, so as to obtain a leaf node sorting result;
The classification unit 3014 is configured to classify the leaf node sorting result according to a preset client quantity classification standard, so as to obtain a willingness classification standard;
The comparison module 302 is configured to configure intent rules based on a preset flow speech template, obtain customer intent rules, receive voice interaction data returned by the artificial intelligent voice robot, and compare the voice interaction data with the customer intent rules to obtain a voice comparison result;
the matching module 303 is configured to extract offline interaction data in the voice comparison result, and match the offline interaction data with a preset client intent label to obtain a client intent matching result;
the classification module 304 is configured to classify the voice comparison result and the customer intent matching result based on the willingness classification standard, respectively, to obtain a customer intent analysis result.
Optionally, the traversing unit 3012 may be specifically configured to:
The method comprises the steps of carrying out traversal processing on preprocessing historical data to obtain traversal results, classifying the traversal results according to preset guest group characteristics to obtain an initial decision tree, carrying out pruning processing on the initial decision tree to obtain a target decision tree, wherein the target decision tree comprises a plurality of leaf nodes, and each leaf node corresponds to one guest group characteristic.
Optionally, the comparing module 302 includes:
A configuration unit 3021, configured to obtain a flow phone operation template, configure intent rules based on a plurality of rule categories in the flow phone operation template, and obtain customer intent rules;
A receiving unit 3022, configured to receive voice interaction data returned by the artificial intelligent voice robot, and obtain a first matching degree between the voice interaction data and the client intent rule;
The first determining unit 3023 is configured to invoke a preset comparison algorithm to determine whether the first matching degree is greater than a preset first matching threshold value, so as to obtain a voice comparison result.
Optionally, the matching module 303 includes:
the extracting unit 3031 is configured to extract voice interaction data with a first matching degree smaller than or equal to a preset first matching threshold in the voice comparison result to obtain offline interaction data, where the first matching degree is a matching degree between the voice interaction data and a client intention rule;
And the second judging unit 3032 is configured to obtain a second matching degree, call a preset similarity matching algorithm, and judge whether the second matching degree is greater than a preset second matching threshold value, so as to obtain a client intention matching result, where the second matching degree is the matching degree of the offline interaction data and a preset client intention label.
Optionally, the classification module 304 includes:
the first dividing unit 3041 is configured to extract first customer data from the voice comparison result, divide the first customer data into a preset first echelon intent classification based on an intent classification standard, and obtain a first classification result, where the first customer data is voice interaction data with a first matching degree greater than a first matching threshold;
The second dividing unit 3042 is configured to extract second client data from the client intent matching result, divide the second client data into a preset second echelon intent classification based on the intent classification standard, and obtain a second classification result, where the second client data is offline interaction data in which a second matching degree in the client intent matching result is greater than a second matching threshold;
a determining unit 3043 for determining the first classification result and the second classification result as the customer intention analysis result.
Optionally, before the obtaining module 301, the client intent analysis device further includes a voice interaction module 305, including:
and acquiring the customer to-be-detected data, and sending the customer to-be-detected data to the artificial intelligent voice robot so that the artificial intelligent voice robot performs voice interaction to obtain voice interaction data.
In the embodiment of the invention, based on a preset decision tree algorithm, preprocessing historical data is classified to obtain a willingness grading standard, voice interaction data is compared with client intention rules to obtain a voice comparison result, offline interaction data in the voice comparison result is extracted, the offline interaction data is matched with a preset client intention degree label to obtain a client intention matching result, and the voice comparison result and the client intention degree matching result are respectively classified to obtain a client intention analysis result based on the willingness grading standard, so that the efficiency of client intention analysis is improved.
The above fig. 3 and fig. 4 describe the customer intent analysis device in the embodiment of the present invention in detail from the point of view of the modularized functional entity, and the following describes the customer intent analysis device in the embodiment of the present invention in detail from the point of view of hardware processing.
Fig. 5 is a schematic diagram of a client intent analysis device 500 according to an embodiment of the present invention, where the client intent analysis device 500 may vary widely according to configuration or performance, and may include one or more processors (central processing units, CPU) 510 (e.g., one or more processors) and memory 520, one or more storage media 530 (e.g., one or more mass storage devices) storing applications 533 or data 532. Wherein memory 520 and storage medium 530 may be transitory or persistent storage. The program stored on the storage medium 530 may include one or more modules (not shown), each of which may include a series of computer program operations in the customer intent analysis device 500. Still further, the processor 510 may be arranged to communicate with a storage medium 530 to execute a series of computer program operations in the storage medium 530 on the customer intent analysis device 500.
The customer intent analysis device 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input/output interfaces 560, and/or one or more operating systems 531, such as Windows Serve, mac OS X, unix, linux, freeBSD, and the like. It will be appreciated by those skilled in the art that the customer intent analysis device architecture shown in FIG. 5 is not limiting of the customer intent analysis device, and may include more or fewer components than shown, or may combine certain components, or a different arrangement of components.
The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, or a volatile computer readable storage medium, in which a computer program is stored which, when run on a computer, causes the computer to perform the steps of the customer intent analysis method.
Further, the computer readable storage medium may mainly include a storage program area, which may store an operating system, an application program required for at least one function, and the like, and a storage data area, which may store data created according to the use of the blockchain node, and the like.
The blockchain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm and the like. The blockchain (Blockchain), essentially a de-centralized database, is a string of data blocks that are generated in association using cryptographic methods, each of which contains information from a batch of network transactions for verifying the validity (anti-counterfeit) of its information and generating the next block. The blockchain may include a blockchain underlying platform, a platform product services layer, an application services layer, and the like.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in whole or in part in the form of a software product stored in a storage medium, comprising a number of computer programs for causing a computer device (which may be a personal computer, a server, a network device, etc.) to execute all or part of the steps of the method according to the embodiments of the present invention. The storage medium includes a U disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk, an optical disk, or other various media capable of storing program codes.
While the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that the foregoing embodiments may be modified or equivalents may be substituted for some of the features thereof, and that the modifications or substitutions do not depart from the spirit and scope of the embodiments of the invention.
Claims (10)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111273716.5A CN113988190B (en) | 2021-10-29 | 2021-10-29 | Customer intention analysis method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111273716.5A CN113988190B (en) | 2021-10-29 | 2021-10-29 | Customer intention analysis method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113988190A CN113988190A (en) | 2022-01-28 |
| CN113988190B true CN113988190B (en) | 2025-05-23 |
Family
ID=79744511
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111273716.5A Active CN113988190B (en) | 2021-10-29 | 2021-10-29 | Customer intention analysis method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113988190B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114706966A (en) * | 2022-03-23 | 2022-07-05 | 平安普惠企业管理有限公司 | Voice interaction method, device and equipment based on artificial intelligence and storage medium |
| CN115455164A (en) * | 2022-09-21 | 2022-12-09 | 中国工商银行股份有限公司 | Method and device for identifying will level, computer equipment and storage medium |
| CN115309737A (en) * | 2022-10-11 | 2022-11-08 | 深圳市明源云客电子商务有限公司 | Visitor intention analysis method, system, terminal device and readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1347618A (en) * | 1999-12-17 | 2002-05-01 | 皇家菲利浦电子有限公司 | Method and apparatus for recommending television programming using decision trees |
| CN110298682A (en) * | 2019-05-22 | 2019-10-01 | 深圳壹账通智能科技有限公司 | Intelligent Decision-making Method, device, equipment and medium based on user information analysis |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107992609B (en) * | 2017-12-15 | 2021-05-18 | 广东电网有限责任公司信息中心 | Complaint tendency judgment method based on text classification technology and decision tree |
| CN110990545B (en) * | 2019-11-28 | 2023-05-09 | 重庆锐云科技有限公司 | Artificial intelligent telephone customer-rubbing marketing management system and method |
| CN112884083A (en) * | 2021-03-31 | 2021-06-01 | 中国工商银行股份有限公司 | Intelligent outbound call processing method and device |
| CN113139059B (en) * | 2021-05-13 | 2022-07-15 | 八维(杭州)科技有限公司 | Intention grading method based on man-machine conversation |
-
2021
- 2021-10-29 CN CN202111273716.5A patent/CN113988190B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1347618A (en) * | 1999-12-17 | 2002-05-01 | 皇家菲利浦电子有限公司 | Method and apparatus for recommending television programming using decision trees |
| CN110298682A (en) * | 2019-05-22 | 2019-10-01 | 深圳壹账通智能科技有限公司 | Intelligent Decision-making Method, device, equipment and medium based on user information analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113988190A (en) | 2022-01-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113988190B (en) | Customer intention analysis method, device, equipment and storage medium | |
| CN113657545B (en) | User service data processing method, device, equipment and storage medium | |
| CN110147726B (en) | Service quality inspection method and device, storage medium and electronic device | |
| CN114389834B (en) | Method, device, equipment and product for identifying abnormal call of API gateway | |
| CN113051291A (en) | Work order information processing method, device, equipment and storage medium | |
| CN110399490A (en) | A kind of barrage file classification method, device, equipment and storage medium | |
| CN112733146B (en) | Penetration testing method, device and equipment based on machine learning and storage medium | |
| CN108334895A (en) | Sorting technique, device, storage medium and the electronic device of target data | |
| CN108734159A (en) | The detection method and system of sensitive information in a kind of image | |
| CN117332066B (en) | Intelligent agent text processing method based on large model | |
| CN112036323A (en) | Signature handwriting identification method, client and server | |
| CN115577172A (en) | Article recommendation method, device, equipment and medium | |
| CN113903357B (en) | Testing method, device, equipment and storage medium of voice robot | |
| CN112148919A (en) | A kind of music click rate prediction method and device based on gradient boosting tree algorithm | |
| CN111814909B (en) | Information processing method based on network live broadcast and online e-commerce delivery and cloud server | |
| CN113807898A (en) | Prediction method, device, equipment and storage medium for agent issuing probability | |
| CN115455184A (en) | Complaint work order classification method and device and related products | |
| CN118247006A (en) | Product information pushing processing method and device | |
| CN113569879A (en) | Training method of abnormal recognition model, abnormal account recognition method and related device | |
| CN116452212B (en) | Intelligent customer service commodity knowledge base information management method and system | |
| CN117172795A (en) | Intelligent technical service fee online consultation system | |
| Varun et al. | An efficient technique for feature selection to predict customer churn in telecom industry | |
| CN116151916A (en) | Intelligent marketing method based on XGBoost model | |
| CN113191711A (en) | Express delivery sending strategy determining method, device, equipment and storage medium | |
| CN114444460A (en) | Method, system and readable storage medium for generating a complete case from an element table |
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 |