CN117952468A - Objection identification and objection response scoring method and equipment - Google Patents
Objection identification and objection response scoring method and equipment Download PDFInfo
- Publication number
- CN117952468A CN117952468A CN202410116300.XA CN202410116300A CN117952468A CN 117952468 A CN117952468 A CN 117952468A CN 202410116300 A CN202410116300 A CN 202410116300A CN 117952468 A CN117952468 A CN 117952468A
- Authority
- CN
- China
- Prior art keywords
- objection
- data
- response
- conversation
- category
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000004044 response Effects 0.000 title claims abstract description 377
- 238000013077 scoring method Methods 0.000 title claims description 5
- 238000000034 method Methods 0.000 claims abstract description 57
- 238000012549 training Methods 0.000 claims abstract description 31
- 239000013598 vector Substances 0.000 claims description 44
- 238000012545 processing Methods 0.000 claims description 23
- 238000002372 labelling Methods 0.000 claims description 15
- 238000012937 correction Methods 0.000 claims description 14
- 238000012216 screening Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 11
- 238000003058 natural language processing Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 239000000284 extract Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000002457 bidirectional effect Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012552 review Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/166—Editing, e.g. inserting or deleting
- G06F40/186—Templates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/01—Customer relationship services
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Human Resources & Organizations (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Evolutionary Biology (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Finance (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Accounting & Taxation (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Databases & Information Systems (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method and equipment for identifying objections and scoring objection responses, wherein the method comprises the following steps: acquiring sales session data; determining an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs by using the objection response identification model obtained through training; according to the objection type and time information of each objection data and the type and time information of each objection response data, matching each objection data with each objection response data to obtain at least one objection response pair; the objection reply pair comprises objection data and objection reply data; scoring the objection response data of each objection response according to the user portraits, the objection categories and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data. The scheme can improve the efficiency of sales managers, and the identification and scoring of the objection response are beneficial to the duplication and sales training and improve the sales order rate.
Description
Technical Field
The invention relates to the technical field of natural language processing, in particular to a method and equipment for identifying objections and scoring objections response.
Background
In a sales and customer dialogue scene, customers can not be prevented from making objections to products, and whether sales can reasonably solve the objections of customers has important influence on bill formation. For the sales management layer, grasping customer objection information and handling mode of sales for customer objection, and having important meaning for the sales process to multiplex and promote sales team performance, how to obtain objection information and evaluate handling mode of customer objection is a problem to be solved by those skilled in the art.
Disclosure of Invention
Aiming at the problems existing in the prior art, the embodiment of the invention provides a method and equipment for identifying objections and scoring objections response.
The invention provides a method for identifying objections and scoring objection responses, which comprises the following steps:
acquiring sales session data;
Determining an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs by using an objection response identification model obtained through training;
Matching each piece of objection data with each piece of objection response data according to the objection type and time information of each piece of objection data and the type and time information of each piece of objection response data to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data;
Scoring the objection response data of each objection response pair according to the user portrait, the objection category and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
According to the method for identifying objections and scoring objections responses provided by the invention, the method for determining an objection category of at least one piece of objection data and a category to which at least one piece of objection response data belongs in sales session data by using an objection response identification model obtained through training comprises the following steps:
performing feature conversion processing on the sales session data to obtain feature vectors;
Inputting the feature vector into the objection response identification model to obtain an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs; the objection response identification model is established based on the BERT model.
According to the method for identifying objections and scoring objections, the matching is performed on each objection data and each objection response data according to the objection category and time information of each objection data and the category and time information of each objection response data, so as to obtain at least one objection response pair, which comprises the following steps:
matching, for any one of the objection data, objection response data having the same category as the objection data according to an objection category of the objection data;
Screening the objection response data with the same category as the objection data according to the time information of the objection data and the time information of the objection response data with the same category as the objection data, and forming at least one objection response pair by the screened at least one objection response data and the objection data.
According to the method for identifying objections and scoring objection responses provided by the invention, each objection response is scored on the objection response data according to the user portrait and a preset conversation template library, and the method comprises the following steps:
Aiming at any objection response pair, obtaining a characteristic vector of objection response data in the objection response pair;
Obtaining all the conversation templates corresponding to the objection response pairs from the conversation template library according to the user portraits and the objection categories;
According to the feature vector of the objection response to the objection response data and the vector representation of each of the speaking templates corresponding to the objection response pair, determining the similarity between the objection response data and each of the speaking templates;
and determining the scores of the objection response to the objection response data according to the similarity between the objection response to the objection response data and each session template.
According to the method for identifying objection and scoring objection response provided by the invention, the step of determining the score of the objection response to the objection response data according to the similarity between the objection response data and each of the speaking templates comprises the following steps:
Taking the conversation template with the highest similarity as a conversation template matched with the objection response pair;
And determining the scores of the objection response pairs on the objection response data according to a conversation template matched with the objection response pairs.
According to the method for identifying objections and scoring objection responses provided by the invention, the method further comprises the following steps:
Acquiring an initial conversation corresponding to a sales service, and converting the initial conversation into a feature vector;
Determining objection response data matched with the initial conversation from a preset conversation database according to the feature vector corresponding to the initial conversation;
And generating a conversation template based on the initial conversation and the objection response data matched with the initial conversation, and storing the conversation template into the conversation template library.
According to the method for identifying objections and scoring the objections, the objection response identification model is obtained through training in the following mode:
screening a plurality of candidate objection data from the session data sample according to the keywords;
determining the objection category of each candidate objection data by using a natural language processing model obtained through training;
labeling each candidate objection data according to the objection category to obtain objection labeling data corresponding to each candidate objection data;
and training a pre-established objection response identification model by utilizing each objection labeling data.
According to the method for identifying objections and scoring objection responses provided by the invention, the sales session data only comprises objection data, and the method further comprises the following steps:
identifying an objection class of objection data in the session data containing objections by using the objection response identification model;
And determining a recommended conversation corresponding to the objection data according to the objection data, the objection category of the objection data, the user portrait and the conversation template library.
According to the method for identifying objections and scoring objections responses provided by the invention, after determining a recommended session corresponding to the objection data according to an objection category of the objection data, a user portrait and the session template library, the method further comprises:
acquiring a correction result of the recommended conversation and a correction result of the objection class;
and iterating the objection response identification model according to the correction result of the objection class.
The invention also provides a objection identification and objection response scoring device, which comprises:
The acquisition module is used for acquiring sales session data;
the processing module is used for determining the objection category of at least one piece of objection data in the sales session data and the category to which the at least one piece of objection response data belongs by utilizing the objection response identification model obtained through training;
The processing module is further configured to match each of the objection data with each of the objection response data according to the objection category and time information of each of the objection data and the category and time information of each of the objection response data, so as to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data;
The processing module is also used for scoring the objection response data of each objection response according to the user portrait, the objection category and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements any of the above methods for identifying objections and scoring objections responses when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method of objection recognition and objection response scoring as described in any one of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements a method of objection recognition and objection response scoring as described in any one of the above.
According to the method and the device for identifying objections and scoring objections response, provided by the invention, an objection type of at least one piece of objection data in sales session data and a type to which the at least one piece of objection response data belongs are determined by utilizing an objection response identification model obtained through training; matching each piece of objection data with each piece of objection response data according to the objection type and time information of each piece of objection data and the type and time information of each piece of objection response data to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data; scoring the objection response data in each objection response according to the user portrait and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data; according to the scheme, the objection response identification model obtained through training can automatically identify the objection data of the clients and the objection response data of the sales personnel, the efficiency is high, the objection response data is scored based on the user portrait and the conversation template library, the accuracy is high, the work efficiency of the sales manager can be effectively improved, support is provided for the sales manager to review and train, and further the objection resolution and the individualization rate of the clients are improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for identifying objections and scoring objections responses according to the present invention;
FIG. 2 is a second flow chart of the method for identifying objections and scoring objections responses according to the present invention;
FIG. 3 is a schematic diagram of the method for identifying objections and scoring objections responses provided by the present invention;
FIG. 4 is a schematic diagram of the structure of the device for identifying objections and scoring objections responses according to the present invention;
fig. 5 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The following describes the technical solution of the embodiment of the present invention in detail with reference to fig. 1 to 5. The following embodiments may be combined with each other, and some embodiments may not be repeated for the same or similar concepts or processes.
Fig. 1 is a flow chart of the method for identifying objections and scoring objections responses provided by the invention. As shown in fig. 1, the method provided in this embodiment includes:
Step 101, obtaining sales session data;
Specifically, when sales personnel and clients are in conversation, conversation contents are recorded, and sales conversation data, such as voice data and text data, are obtained;
optionally, acquiring sales session data according to different usage scenarios;
For example, if the manager needs to obtain a score for the objection response data, then obtaining sales session data including the objection data and the objection response data;
Or, if the salesperson needs to answer to the objection data, namely, recommending the conversation, acquiring sales session data only containing the objection data;
Sales session data is used to make objection answer scoring and conversation recommendations.
Optionally, user basic information, such as age, gender, occupation, income, liability, etc., of the user may also be obtained in advance, to extract user portraits, and to recommend words to the objection data and score the objection response data of the sales personnel according to the user portraits.
Step 102, determining an objection category of at least one piece of objection data in sales session data and a category to which the at least one piece of objection response data belongs by using an objection response recognition model obtained through training;
specifically, the objection response identification model may extract objection data and objection response data from sales session data, so as to identify an objection class of the objection data and a class to which the objection response data belongs;
Inputting sales session data into an objection response identification model, and acquiring an objection category of at least one piece of objection data and a category of at least one piece of objection response data in the output sales session data.
The objection response recognition model can be established based on a deep learning algorithm and trained based on training data samples.
Step 103, matching each piece of objection data with each piece of objection response data according to the objection category and time information of each piece of objection data and the category and time information of each piece of objection response data, so as to obtain at least one objection response pair; the objection reply pair comprises objection data and objection reply data;
Specifically, according to the objection type of the objection data and the type of the objection response data, screening the objection data and the objection response data, wherein the objection type is the same as the type of the objection response data;
According to the time information of the objection data and the time information of the objection response data, acquiring the objection response data matched with the objection data, namely the objection response data meeting a certain time sequence;
And taking the objection data and objection response data matched with the objection data as an objection response pair.
Step 104, scoring the objection response data of each objection response according to the user portrait, the objection category and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
Specifically, the corresponding conversation template in the conversation template library can be determined based on the objection response pair objection data and the user portrait, the matched conversation template can be determined based on the objection response pair objection response data and the user portrait, or the corresponding conversation template can be determined based on the objection response pair objection data, the objection response data and the user portrait.
Scoring is then performed according to the determined conversation template, for example, scoring is performed according to the degree of matching and similarity of the objection response data and the conversation template.
According to the method, the objection response identification model obtained through training can automatically identify the objection data of the clients and the objection response data of the sales personnel, the efficiency is high, the objection response data is scored based on the user portraits and the conversation template library, the accuracy is high, the work efficiency of the sales manager can be effectively improved, support is provided for the sales manager to review and train, and further the objection resolution and the individuality of the clients are improved.
Optionally, the objection response recognition model is trained by:
screening a plurality of candidate objection data from the session data sample according to the keywords;
determining the objection category of each candidate objection data by using a natural language processing model obtained through training;
labeling each candidate objection data according to the objection category to obtain objection labeling data corresponding to each candidate objection data;
and training a pre-established objection response identification model by utilizing each objection labeling data.
Specifically, candidate objection data is screened from session data through keywords, then an objection class is predicted through a deep learning model (for example, a natural language processing model, such as a generated pre-training transducer (GENERATIVE PRE-Trained Transformer, GPT) model) obtained through training, the candidate objection data is manually screened to obtain target objection data, and the objection class is marked on the target objection data to obtain objection marking data;
Training a pre-established objection response recognition model by using objection labeling data, namely, inputting target objection data into the pre-established objection response recognition model for training, wherein the target objection data needs to be subjected to feature conversion processing before being input into the objection response recognition model, so as to obtain feature vectors corresponding to the target objection data.
Furthermore, the objection answer identification model input objection annotation data and corresponding objection answer annotation data can be trained.
The pre-established objection response model is obtained based on the BERT model, which is known as Bidirectional Encoder Representation from Transformers, namely Encoder of the bidirectional transducer, and the BERT model is a multi-layer transducer model. The input of the BERT model is a word vector of each word/word (or token) in the text, for example, the input text is segmented to obtain a plurality tokens, and then each token is converted into a token ID.
Alternatively, step 102 may be specifically implemented as follows:
performing feature conversion processing on the sales session data to obtain feature vectors;
Inputting the feature vector into the objection response identification model to obtain an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs; the objection response identification model is established based on the BERT model.
Specifically, feature conversion processing is performed on the sales session data to obtain feature vectors, such as word segmentation is performed on the sales session data, and token after word segmentation is converted into token ID to form feature vectors;
And inputting the feature vector into the objection response identification model, and outputting an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs.
In the embodiment, the objection response recognition model is trained through data labeling, so that the recognition result of the model is more accurate.
Optionally, the method further comprises:
Acquiring an initial conversation corresponding to a sales service, and converting the initial conversation into a feature vector;
Determining objection response data matched with the initial conversation from a preset conversation database according to the feature vector corresponding to the initial conversation;
And generating a conversation template based on the initial conversation and the objection response data matched with the initial conversation, and storing the conversation template into the conversation template library.
Specifically, firstly, according to an initial conversation corresponding to a service, a natural language processing model is utilized to obtain a feature vector corresponding to the initial conversation (for example, vector conversion is carried out in an embedding embedding mode), then similar objection response data are matched from a preset session database through vector representation corresponding to the initial conversation, a conversation template is generated based on the initial conversation and the objection response data matched with the initial conversation, and the conversation template is stored in the conversation template library.
For example, the feature vector corresponding to the initial conversation is calculated, and the similarity of vector representation of the objection response data in the conversation database is calculated, so that the objection response data matched with the initial conversation is obtained, in order to improve the efficiency, the category of the initial conversation can be determined first, the objection response data with the same category can be extracted, and then the similarity matching can be performed.
In the embodiment, the voice operation template library is generated in advance, data support is provided for the subsequent objection response scoring and voice operation recommendation, so that the efficiency of the objection response scoring and voice operation recommendation is higher, and the accuracy is higher.
Alternatively, step 103 may be specifically implemented as follows:
matching, for any one of the objection data, objection response data having the same category as the objection data according to an objection category of the objection data;
Screening the objection response data with the same category as the objection data according to the time information of the objection data and the time information of the objection response data with the same category as the objection data, and forming at least one objection response pair by the screened at least one objection response data and the objection data.
Specifically, according to the objection type of the objection data and the type of the objection response data, screening the objection data and the objection response data, wherein the objection type is the same as the type of the objection response data;
According to the time information of the objection data and the time information of the objection response data, namely based on the conversation time sequence, screening the objection response data with the same category as the objection data, thereby forming an objection response pair with the screened objection response data.
For example, the starting time of the objection data is 10 points and 10 minutes, the objection response data of the same category comprises 3 pieces, the starting time is 10 points, 10 points and 15 minutes and 10 points and 09 minutes respectively, and then the objection response data of the starting time and 10 points and 15 minutes and the objection data form an objection response pair.
In the embodiment, the objection data and the objection response data are matched by screening the objection response data, so that the accuracy of the objection response scoring is improved.
Alternatively, step 104 may be specifically implemented as follows:
Aiming at any objection response pair, obtaining a characteristic vector of objection response data in the objection response pair;
Obtaining all the conversation templates corresponding to the objection response pairs from the conversation template library according to the user portraits and the objection categories;
According to the feature vector of the objection response to the objection response data and the vector representation of each of the speaking templates corresponding to the objection response pair, determining the similarity between the objection response data and each of the speaking templates;
and determining the scores of the objection response to the objection response data according to the similarity between the objection response to the objection response data and each session template.
Specifically, all the conversation templates corresponding to the objection response pairs are obtained from a conversation template library according to the user portraits and the objection categories; for example, the corresponding conversation template may be determined based on the objection response versus the objection data and the user profile, or a matching conversation template may be determined based on the objection response versus the objection data and the user profile, or a corresponding conversation template may be determined based on the objection response versus the objection data, the objection response data, and the user profile.
According to the feature vector of the objection response pair objection response data and the vector representation of each corresponding objection response pair, determining the similarity between the objection response data and each corresponding objection template;
and determining the scores of the objection response data in the objection response pair according to the similarity between the objection response data and each conversation template. The similarity may be represented by cosine similarity, for example.
Further, the similarity between the objection response data and each conversation template can be weighted and averaged to obtain a final score.
Optionally, the determining the score of the objection response to the objection response data in the objection response pair according to the similarity between the objection response data in the objection response pair and each of the speaking templates includes:
Taking the conversation template with the highest similarity as a conversation template matched with the objection response pair;
And determining the scores of the objection response pairs on the objection response data according to a conversation template matched with the objection response pairs.
The score of the objection response data may be determined specifically by the similarity corresponding to the objection response pair matched session template, for example, the similarity is used as the score of the objection response data.
Optionally, the method further comprises:
and extracting the user portrait according to the user basic information corresponding to the sales session data.
In the above embodiment, the feature vector of the objection response to the objection response data is obtained; acquiring at least one conversation template corresponding to the objection response pair from the conversation template library according to the user portrait; according to the feature vector of the objection response to the objection response data and the vector representation of each of the speaking templates corresponding to the objection response pair, determining the similarity between the objection response data and each of the speaking templates; and determining the scores of the objection response to the objection response data according to the similarity between the objection response data and each session template, so that the scores are more objective and accurate, conform to the actual scene, and have lower implementation complexity.
Further, in the case where sales session data contains only objection data, as shown in fig. 2, the method further includes the steps of:
step 201, identifying the objection category of the objection data in the sales session data by using an objection response identification model;
Step 202, determining recommended dialects corresponding to the objection data according to the objection data, the objection category of the objection data, the user image and the dialects template library.
Specifically, sales personnel can upload session data containing objections, and identify an objection category of the objection data in the session data containing objections by using an objection response identification model;
according to the objection category of objection data, the user portrait and the conversation template library, determining the recommended conversation corresponding to the objection data, for example, a pre-trained conversation recommendation model can be utilized to extract conversation templates matched with the objection data from the conversation template library based on the objection category of objection data and the user portrait, and the matched conversation templates are used as recommended conversation.
For example, the speech recommendation model may be built based on a deep learning algorithm and trained using training sample data including objection data and objection response data.
Optionally, the following operations may also be performed after step 203:
acquiring a correction result of the recommended conversation and a correction result of the objection class;
and iterating the objection response identification model according to the correction result of the objection class.
Specifically, after the sales personnel acquires the recommended call, the sales personnel can correct the recommended call, correct the objection data output by the objection response recognition model and the objection category of the objection data, iterate the objection response recognition model based on the correction result, and optimize the objection response recognition model, so that the recognition result of the model is more accurate.
Optionally, the corrected result of the recommended speaking operation can be used for iterating the speaking operation recommendation model, so that the recommended speaking operation result is more accurate.
In the above embodiment, for the sales management layer, the customer objection information and the processing mode of the sales personnel for the customer objection (i.e. objection response) are mastered, that is, the degree of the processing mode of the sales personnel for the customer objection is known through the objection accompanying score, so that the management efficiency is improved, the sales process can be duplicated, and the method has important significance for improving the performance of the sales team; for sales personnel, the user image and conversation template library can guide the sales personnel to select reasonable conversation coping aiming at customer objections, and the single rate can be effectively improved.
Illustratively, as shown in FIG. 3, the objection answer scoring and speaking recommendation system includes the following modules:
1. Standard conversation template library: the system comprises a plurality of conversation templates, a conversation database and a conversation database, wherein the conversation templates are used for responding to the objection data and can be used for recommending conversation to sales personnel or scoring the objection response data;
2. And a data input module: the part is an input module of the whole system, and inputs sales session data and user basic information according to the use scene. The sales session data is used for identifying the objection data of the user and the objection response data of the sales personnel, the user basic information is used for extracting user portraits, and the objection data is recommended in a speaking way and scored according to the user portraits;
3. A user portrait module: the part extracts user portraits, and extracts user portraits information such as age, gender, occupation, income situation, liability situation and the like of the user according to the input user basic information;
4. An objection response identification module: the part is a core module of the whole system, and is divided into objection identification and response identification, and the objection category and the category to which objection response data belong are identified through a BERT model;
Optionally, the method may further include:
the data marking module: firstly screening candidate objection data from sales session data through keywords, then predicting objection categories through a natural language processing model, and finally manually screening to obtain objection labeling data;
the feature processing module is used for carrying out online processing on the sales dialogue data and converting the sales dialogue data into feature vectors required by the objection response recognition model, such as input data required by the BERT model;
5. And a speaking recommendation module: the part carries out speaking recommendation, and after identifying the objection category, carries out speaking recommendation based on objection data, objection category, user portrait and standard speaking template library;
6. an objection response matching module: the part is responsible for matching the objection data of the user with the objection response data of the sales personnel, and mainly matching the objection category with the category to which the objection response data belongs and the conversation time sequence;
7. objection response scoring module: the part scores the objection response data, and scores the objection response data based on the user portrait and the standard speaking template library after identifying the objection response pair;
8. and an output display module: the part displays objection data, objection response data, categories, response scores, recommended utterances and the like;
9. Model iteration module: the part provides a foreign matter protocol and response correction function, sales personnel can manually correct the foreign matter data, the foreign matter category and recommended foreign matter response data, and the corrected data can be used as a training data iterative foreign matter response recognition model.
In summary, the method of the embodiment of the invention realizes the intelligent recommendation scheme of customer objection identification and sales operation, automatically identifies the objection data of customers and the objection response data of sales personnel, scores the objection response data, provides support for the duplication and training of sales managers, and can effectively improve the work efficiency of the sales managers; in addition, a real-time recommendation function of the objection operation is provided, references are provided for sales personnel when customer objections are solved based on user portraits, the customer objection resolution ratio and the listing rate are improved, and the company income is improved; furthermore, feedback information of sales personnel for identifying and recommending objections can be collected, and automatic iterative upgrade of the system is realized according to the feedback information.
The description of the objection recognition and objection response scoring device provided by the invention is provided below, and the objection recognition and objection response scoring device described below and the objection recognition and objection response scoring method described above can be correspondingly referred to each other.
Fig. 4 is a schematic structural diagram of an objection identification and objection response scoring device provided by the invention. As shown in fig. 4, the apparatus for identifying objections and scoring an objection response provided in this embodiment includes:
an acquisition module 410, configured to acquire sales session data;
a processing module 420, configured to determine an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs, using the objection response recognition model obtained through training;
the processing module 420 is further configured to match each of the objection data with each of the objection response data according to the objection category and time information of each of the objection data and the category and time information of each of the objection response data, so as to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data;
The processing module 420 is further configured to score the median objection response data according to the user portraits, the objection categories, and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
Optionally, the processing module 420 is specifically configured to:
performing feature conversion processing on the sales session data to obtain feature vectors;
Inputting the feature vector into the objection response identification model to obtain an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs; the objection response identification model is established based on the BERT model.
Optionally, the processing module 420 is specifically configured to:
matching, for any one of the objection data, objection response data having the same category as the objection data according to an objection category of the objection data;
Screening the objection response data with the same category as the objection data according to the time information of the objection data and the time information of the objection response data with the same category as the objection data, and forming at least one objection response pair by the screened at least one objection response data and the objection data.
Optionally, the processing module 420 is specifically configured to:
Aiming at any objection response pair, obtaining a characteristic vector of objection response data in the objection response pair;
Obtaining all the conversation templates corresponding to the objection response pairs from the conversation template library according to the user portraits and the objection categories;
According to the feature vector of the objection response to the objection response data and the vector representation of each of the speaking templates corresponding to the objection response pair, determining the similarity between the objection response data and each of the speaking templates;
and determining the scores of the objection response to the objection response data according to the similarity between the objection response to the objection response data and each session template.
Optionally, the processing module 420 is specifically configured to:
Taking the conversation template with the highest similarity as a conversation template matched with the objection response pair;
And determining the scores of the objection response pairs on the objection response data according to a conversation template matched with the objection response pairs.
Optionally, the processing module 420 is further configured to:
Acquiring an initial conversation corresponding to a sales service, and converting the initial conversation into a feature vector;
Determining objection response data matched with the initial conversation from a preset conversation database according to the feature vector corresponding to the initial conversation;
And generating a conversation template based on the initial conversation and the objection response data matched with the initial conversation, and storing the conversation template into the conversation template library.
Optionally, the objection response identification model is trained by:
screening a plurality of candidate objection data from the session data sample according to the keywords;
determining the objection category of each candidate objection data by using a natural language processing model obtained through training;
labeling each candidate objection data according to the objection category to obtain objection labeling data corresponding to each candidate objection data;
and training a pre-established objection response identification model by utilizing each objection labeling data.
Optionally, the sales session data only includes objection data, and the processing module 420 is further configured to:
identifying an objection class of objection data in the session data containing objections by using the objection response identification model;
And determining a recommended conversation corresponding to the objection data according to the objection data, the objection category of the objection data, the user portrait and the conversation template library.
Optionally, the processing module 420 is further configured to:
after determining a recommended conversation corresponding to the objection data according to the objection category of the objection data, the user portrait and the conversation template library, acquiring a correction result of the recommended conversation and a correction result of the objection category;
and iterating the objection response identification model according to the correction result of the objection class.
The device of the embodiment of the present invention is configured to perform the method of any of the foregoing method embodiments, and its implementation principle and technical effects are similar, and are not described in detail herein.
Fig. 5 illustrates a physical schematic diagram of an electronic device, as shown in fig. 5, which may include: processor 510, communication interface (Communications Interface) 520, memory 530, and communication bus 540, wherein processor 510, communication interface 520, memory 530 complete communication with each other through communication bus 540. Processor 510 may invoke logic instructions in memory 530 to perform the objection identification and objection answer scoring method comprising: acquiring sales session data;
Determining an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs by using an objection response identification model obtained through training;
Matching each piece of objection data with each piece of objection response data according to the objection type and time information of each piece of objection data and the type and time information of each piece of objection response data to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data;
Scoring the objection response data of each objection response pair according to the user portrait, the objection category and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
Further, the logic instructions in the memory 530 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. 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 a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of executing the method for identifying and scoring an objection response provided by the above methods, the method comprising: acquiring sales session data;
Determining an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs by using an objection response identification model obtained through training;
Matching each piece of objection data with each piece of objection response data according to the objection type and time information of each piece of objection data and the type and time information of each piece of objection response data to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data;
Scoring the objection response data of each objection response pair according to the user portrait, the objection category and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method of objection recognition and objection response scoring provided by the above methods, the method comprising: acquiring sales session data;
Determining an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs by using an objection response identification model obtained through training;
Matching each piece of objection data with each piece of objection response data according to the objection type and time information of each piece of objection data and the type and time information of each piece of objection response data to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data;
Scoring the objection response data of each objection response pair according to the user portrait, the objection category and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for objection identification and objection response scoring, comprising:
acquiring sales session data;
Determining an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs by using an objection response identification model obtained through training;
Matching each piece of objection data with each piece of objection response data according to the objection type and time information of each piece of objection data and the type and time information of each piece of objection response data to obtain at least one objection response pair; the objection response pair comprises objection data and objection response data;
Scoring the objection response data of each objection response pair according to the user portrait, the objection category and a preset conversation template library; the conversation template library comprises a plurality of conversation templates, and the conversation templates are used for responding to the objection data.
2. The method for identifying objections and scoring an objection response of claim 1, wherein determining an objection category of at least one item of objection data and a category to which at least one item of objection response data belongs in the sales session data using a trained objection response identification model, comprises:
performing feature conversion processing on the sales session data to obtain feature vectors;
Inputting the feature vector into the objection response identification model to obtain an objection category of at least one piece of objection data in the sales session data and a category to which the at least one piece of objection response data belongs; the objection response identification model is established based on the BERT model.
3. The method for identifying and scoring an objection response according to claim 1 or 2, wherein said matching each of said objection data and each of said objection response data according to an objection category and time information of each of said objection data and a category and time information of each of said objection response data to obtain at least one objection response pair, comprises:
matching, for any one of the objection data, objection response data having the same category as the objection data according to an objection category of the objection data;
Screening the objection response data with the same category as the objection data according to the time information of the objection data and the time information of the objection response data with the same category as the objection data, and forming at least one objection response pair by the screened at least one objection response data and the objection data.
4. The method for identifying objections and scoring an objection response of claim 1 or 2, wherein scoring the objection response data for each of the objection responses based on a user portrayal, an objection category, and a library of pre-established conversation templates, comprises:
Aiming at any objection response pair, obtaining a characteristic vector of objection response data in the objection response pair;
Obtaining all the conversation templates corresponding to the objection response pairs from the conversation template library according to the user portraits and the objection categories;
According to the feature vector of the objection response to the objection response data and the vector representation of each of the speaking templates corresponding to the objection response pair, determining the similarity between the objection response data and each of the speaking templates;
and determining the scores of the objection response to the objection response data according to the similarity between the objection response to the objection response data and each session template.
5. The method of claim 4, wherein determining the score of the objection response to the objection response data based on the similarity between the objection response to the objection response data and each of the conversation templates comprises:
Taking the conversation template with the highest similarity as a conversation template matched with the objection response pair;
And determining the scores of the objection response pairs on the objection response data according to a conversation template matched with the objection response pairs.
6. The method of objection recognition and objection response scoring of claim 1 or 2, further comprising:
Acquiring an initial conversation corresponding to a sales service, and converting the initial conversation into a feature vector;
Determining objection response data matched with the initial conversation from a preset conversation database according to the feature vector corresponding to the initial conversation;
And generating a conversation template based on the initial conversation and the objection response data matched with the initial conversation, and storing the conversation template into the conversation template library.
7. The method for identifying and scoring an objection response according to claim 1 or 2, wherein the model for identifying an objection response is trained by:
screening a plurality of candidate objection data from the session data sample according to the keywords;
determining the objection category of each candidate objection data by using a natural language processing model obtained through training;
labeling each candidate objection data according to the objection category to obtain objection labeling data corresponding to each candidate objection data;
and training a pre-established objection response identification model by utilizing each objection labeling data.
8. The objection identification and objection answer scoring method of claim 1 or 2, wherein the sales session data includes only objection data, the method further comprising:
identifying an objection class of objection data in the session data containing objections by using the objection response identification model;
And determining a recommended conversation corresponding to the objection data according to the objection data, the objection category of the objection data, the user portrait and the conversation template library.
9. The method for identifying objections and scoring an objection response of claim 8, wherein after determining a recommended session corresponding to the objection data based on an objection category of the objection data, a user portrayal, and the session template library, further comprising:
acquiring a correction result of the recommended conversation and a correction result of the objection class;
and iterating the objection response identification model according to the correction result of the objection class.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of identifying and scoring an objection response as claimed in any one of claims 1 to 9 when the program is executed by the processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410116300.XA CN117952468A (en) | 2024-01-26 | 2024-01-26 | Objection identification and objection response scoring method and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410116300.XA CN117952468A (en) | 2024-01-26 | 2024-01-26 | Objection identification and objection response scoring method and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117952468A true CN117952468A (en) | 2024-04-30 |
Family
ID=90802641
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410116300.XA Pending CN117952468A (en) | 2024-01-26 | 2024-01-26 | Objection identification and objection response scoring method and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117952468A (en) |
-
2024
- 2024-01-26 CN CN202410116300.XA patent/CN117952468A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109670035B (en) | Text abstract generating method | |
| CN105723362B (en) | Naturally processing method, processing and response method, equipment and system are expressed | |
| CN112000791A (en) | Motor fault knowledge extraction system and method | |
| CN111858877A (en) | Multi-type question intelligent question answering method, system, equipment and readable storage medium | |
| CN110717018A (en) | Industrial equipment fault maintenance question-answering system based on knowledge graph | |
| CN107886231B (en) | Service quality evaluation method and system for customer service | |
| CN110321564B (en) | Multi-round dialogue intention recognition method | |
| CN109857846B (en) | Method and device for matching user question and knowledge point | |
| CN111930914A (en) | Question generation method and device, electronic equipment and computer-readable storage medium | |
| CN114722839A (en) | Man-machine collaborative dialogue interaction system and method | |
| CN112287090A (en) | Financial question asking back method and system based on knowledge graph | |
| CN113807103B (en) | Recruitment method, device, equipment and storage medium based on artificial intelligence | |
| CN107291775A (en) | The reparation language material generation method and device of error sample | |
| CN111429157A (en) | Method, device and equipment for evaluating and processing complaint work order and storage medium | |
| CN110046232A (en) | Natural expression processing method, response method, equipment and the system of natural intelligence | |
| CN113468891A (en) | Text processing method and device | |
| CN109977213A (en) | A kind of optimal answer selection method towards intelligent Answer System | |
| CN115269836A (en) | Intention identification method and device | |
| CN113505606B (en) | Training information acquisition method and device, electronic equipment and storage medium | |
| CN117932009A (en) | ChatGLM model-based insurance customer service dialogue generation method, chatGLM model-based insurance customer service dialogue generation device, chatGLM model-based insurance customer service dialogue generation equipment and ChatGLM model-based insurance customer service dialogue generation medium | |
| CN110059167A (en) | Natural expression processing method, response method, equipment and the system of natural intelligence | |
| CN110008317A (en) | Natural expression processing method, response method, equipment and the system of natural intelligence | |
| CN112767940A (en) | Voice training recognition method, system, equipment and storage medium | |
| CN107886233B (en) | Service quality evaluation method and system for customer service | |
| CN110580899A (en) | Voice recognition method and device, storage medium and computing equipment |
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 |