CN117651093A - Information generation method, device, electronic equipment and readable storage medium - Google Patents
Information generation method, device, electronic equipment and readable storage medium Download PDFInfo
- Publication number
- CN117651093A CN117651093A CN202311498439.7A CN202311498439A CN117651093A CN 117651093 A CN117651093 A CN 117651093A CN 202311498439 A CN202311498439 A CN 202311498439A CN 117651093 A CN117651093 A CN 117651093A
- Authority
- CN
- China
- Prior art keywords
- node
- data
- user
- information
- generating
- 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
- 238000000034 method Methods 0.000 title claims abstract description 86
- 238000000605 extraction Methods 0.000 claims abstract description 30
- 239000011159 matrix material Substances 0.000 claims description 35
- 230000006854 communication Effects 0.000 claims description 17
- 238000004891 communication Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 15
- 230000007246 mechanism Effects 0.000 claims description 15
- 238000003062 neural network model Methods 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 9
- 230000014509 gene expression Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 3
- 238000007726 management method Methods 0.000 description 36
- 230000008569 process Effects 0.000 description 28
- 238000012550 audit Methods 0.000 description 9
- 238000003066 decision tree Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000009471 action Effects 0.000 description 5
- 238000012795 verification Methods 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 230000001960 triggered effect Effects 0.000 description 3
- 238000010200 validation analysis Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000012790 confirmation Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000008451 emotion Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000013524 data verification Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Abstract
The embodiment of the invention provides an information generation method, an information generation device, electronic equipment and a readable storage medium. The method comprises the following steps: acquiring a flow configuration file, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node; entering an initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; entering an answering node after the call is connected, acquiring voice data of a user according to execution information of the answering node, and converting the voice data into a text; entering an extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of a user; and entering a generating node after the keyword data are stored, and generating flow result information of the user based on the keyword data according to the execution information of the generating node. The invention can realize the automatic generation of the user information of the outbound flow and improve the information generation efficiency.
Description
Technical Field
The present invention relates to the field of computers, and in particular, to an information generating method, an information generating device, an electronic device, and a readable storage medium.
Background
The outbound system may be used to automatically initiate communication with the user, for example, to obtain opinion of the user, or to obtain a service rating of the user, etc.
At present, a call can be automatically or semi-automatically initiated to a user through an outbound system, the call is communicated with the user, some comments of the user are obtained, and then the comments are manually recorded.
However, this method requires manual operation, and has problems of high labor cost and low efficiency.
Disclosure of Invention
The invention provides an information generation method, an information generation device, electronic equipment and a readable storage medium, which are used for solving the problems of high labor cost and low efficiency in the related technology.
In a first aspect, the present invention provides an information generating method, including:
acquiring a flow configuration file, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node;
entering the initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls;
entering the answering node after the call is connected, acquiring voice data of the user according to the execution information of the answering node, and converting the voice data into text;
Entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user;
and entering the generating node after the keyword data are stored, and generating flow result information of the user based on the keyword data according to the execution information of the generating node.
In a second aspect, the present invention provides a service inspection device, including:
the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring a flow configuration file, and the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node;
the initial module is used for entering the initial node, and triggering the outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls;
the answering module is used for entering the answering node after the call is switched on, acquiring the voice data of the user according to the execution information of the answering node, and converting the voice data into a text;
the extraction module is used for entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user;
And the generation module is used for entering the generation node after the keyword data are stored, and generating the flow result information of the user based on the keyword data according to the execution information of the generation node.
In a third aspect, the present invention provides an electronic device comprising: a processor, a memory, and a computer program stored on the memory and executable on the processor, the processor implementing the information generating method of the first aspect when executing the program.
In a fourth aspect, the present invention provides a readable storage medium, which when executed by a processor of an electronic device, enables the electronic device to perform the information generating method of the first aspect.
In summary, a flow configuration file is obtained, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node; the method comprises the steps of carrying out a first treatment on the surface of the Entering the initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls; entering the answering node after the call is connected, acquiring voice data of the user according to the execution information of the answering node, and converting the voice data into text; entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user; and entering the generating node after the keyword data are stored, and generating flow result information of the user based on the keyword data according to the execution information of the generating node. The method and the device can combine the nodes based on the progress of the outbound process according to the pre-configured process configuration file, correspondingly execute related actions, finally generate the process result information of the user, automatically generate the user information of the outbound process, and improve the information generation efficiency.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the related art, the drawings that are required to be used in the embodiments or the related technical descriptions will be briefly described, and it is apparent that the drawings in the following descriptions are some embodiments of the present invention, and other drawings may be obtained according to the drawings without inventive effort for those skilled in the art.
Fig. 1 is a flowchart of steps of an information generating method according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating steps of another information generating method according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating steps of node conversion according to an embodiment of the present invention;
FIG. 4 is a block diagram of various portions of a blockchain provided by an embodiment of the present invention; the method comprises the steps of carrying out a first treatment on the surface of the
Fig. 5 is a data transmission flow chart of the outbound system, blockchain, and subscriber management system provided by the embodiment of the invention;
FIG. 6 is an interactive flow diagram of an outbound system, blockchain, and subscriber management system provided by an embodiment of the invention;
fig. 7 is a block diagram of an information generating apparatus according to an embodiment of the present invention;
fig. 8 is a block diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. 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 information generating method of the present application is further described below with reference to the accompanying drawings and examples:
fig. 1 is a flowchart of steps of an information generating method according to an embodiment of the present invention. Referring to fig. 1, the method may include:
step 101, acquiring a flow configuration file, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node;
in the disclosed embodiments, the flow profile may be implemented by a smart contract on the blockchain. Smart Contract (Smart contact) is a computer protocol that propagates, validates, or executes contracts in an informative manner. A smart contract is an automated execution contract whose contract terms are written directly in code into the blockchain. The intelligent contract can automatically execute specific business logic when the preset condition is met, and intervention or verification of any third party is not needed. A finite node automaton (FSM, finite State Machine), which is a computational model abstracted for studying finite memory computational processes and certain language classes, has a finite number of nodes, each of which can migrate to zero or more nodes, and under certain conditions or events, from one node to another.
In the embodiment of the disclosure, the intelligent contract can be built and generated based on a finite node automaton, and the finite node automaton can comprise a creation node, an answering node, an indicating node and a subscribing node, wherein the creation node represents that a new order is created in a user management system, and an outbound event is triggered; the answering node is a node for communication after the user answers after the outbound is triggered; the indication node is a node for generating an indication of the user about ordering according to the communicated content; the order node refers to a notice that the user management system generates an order of order according to the order instruction of the user. The user management system may be a customer relationship management system (CRM, customer Relationship Management), which is a system for realizing the interaction between enterprises and customers on sales, marketing and service by using corresponding information technology and internet technology.
It will be appreciated that the smart contract may also include more nodes, such as a creation node, a waiting node, an answer node, an indication node, and a subscription node, that is, one more node than the above process, and the waiting node may be a node that reaches a period of time before the user answers after the outbound trigger. The greater the number of nodes, the more precisely the overall process performed, and the particular number of nodes is not limited herein.
Step 102, triggering an outbound system to initiate a call to a user based on the creation node; the outbound system is a communication system for making telephone calls.
The initial node of the intelligent contract may be a creation node upon which an action is performed that triggers the outbound system to initiate a call to the user. The outbound system is a communication system and is mainly used for automatically or semi-automatically making a call. The system is generally used in sales, marketing, customer service and other scenes, and the outbound system can play pre-recorded voice to the user and then acquire voice fed back by the user.
Step 103, entering the answering node after the call is connected, acquiring the voice data of the user according to the execution information of the answering node, and converting the voice data into text.
In the disclosed embodiment, the nodes of the intelligent contract are entered into the answering node by the creating node after the call is completed. Based on the receiving node, the voice data of the user in the call is acquired and converted into text. The recognition of the voice to the words can be completed through four steps of feature extraction, an acoustic model, a language model and a decoder. The feature extraction refers to removing useless information for voice recognition in a voice signal, and retaining key information capable of reflecting essential features of the voice, and can be to sample the voice to obtain digitized features of the voice, for example, features such as phoneme features, loudness, pitch and the like can be extracted; the acoustic model may be modeling sound, converting a speech input into an output of an acoustic representation, for example, a mixture gaussian model + a hidden markov model, modeling time information by the mixture gaussian model, and modeling a probability distribution of a speech feature vector belonging to a node of the hidden markov model after the node is given; the language model is used for calculating the probability of occurrence of a sentence, and calculating the probability of occurrence of a sentence of a word, and can be a statistical-based language model or other language models based on deep learning.
And 104, entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating the keyword data of the user.
In the embodiment of the disclosure, the call end is the end of communication between the user and the outbound system, the user enters the indication node after the call end, and the keyword data is generated based on the indication node according to the text converted based on the voice data in the communication process. The keyword data may be keywords included only in text converted based on voice data, and for example, for ordering, the keywords may be some key information of user information, time information, item information to be ordered, intention representing information whether to be ordered, and the like, which may be used to determine whether the user is to order, when to order, what item to order.
And step 105, entering the generating node after the keyword data are stored, and generating the flow result information of the user based on the keyword data according to the execution information of the generating node.
In the embodiment of the present disclosure, the keyword data may be stored in a preset location, for example, a database. After the storage is completed, the process enters a generating node, and according to the execution information of the generating node, the execution can be, for example, the process result information of the user is generated based on the keyword data. For example, the keyword data may be "service", "good", "buyback", and the flow result information that may be generated according to the keyword data is "consider the service good, and buyback".
In summary, a flow configuration file is obtained, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node; the method comprises the steps of carrying out a first treatment on the surface of the Entering the initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls; entering the answering node after the call is connected, acquiring voice data of the user according to the execution information of the answering node, and converting the voice data into text; entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user; and entering the generating node after the keyword data are stored, and generating flow result information of the user based on the keyword data according to the execution information of the generating node. The method and the device can combine the nodes based on the progress of the outbound process according to the pre-configured process configuration file, correspondingly execute related actions, finally generate the process result information of the user, automatically generate the user information of the outbound process, and improve the information generation efficiency.
Fig. 2 is a further information generating method according to an embodiment of the present application, and referring to fig. 2, the method may include the following steps:
step 201, obtaining a flow configuration file, where the flow configuration file includes an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node
Step 202, entering the initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls;
step 203, entering the answering node after the call is connected, acquiring the voice data of the user according to the execution information of the answering node, and converting the voice data into text;
step 204, entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user;
step 205, entering the generating node after the keyword data is stored, and generating the flow result information of the user based on the keyword data according to the execution information of the generating node.
Step 206, when any node of the initial node, the answering node, the extracting node and the generating node is entered, generating an identifier corresponding to the any node through a preset function;
The identification is used for checking the data integrity of any node.
In the embodiment of the present disclosure, the conversion of the node may be performed by the following node conversion function δ:
δ:S n ×I n →S
wherein S is n Is a node I n Is triggered S n The node enters the event of the next node S. For example, when the smart contract is at node S 1 And receive I 1 Upon event, the node transitions to S 2 . As shown in fig. 3, is a schematic diagram of the transition between nodes. Starting from the initial node, switching to the answering node, switching to the extracting node from the answering node, and switching to the generating node from the extracting node. As previously mentioned, the actual number of nodes is not limited and may be more or less nodes.
And generating a corresponding identifier of the next node through a preset function each time the node is converted. The hash value of the current node can be generated through a hash function, the hash value is the mapping of the node corresponding to the hash value, one node corresponds to one hash value, the data of the node can be verified through the hash value, and the uniqueness and the non-tamper property of the node are ensured. The formula can be expressed as:
H:{0,1} * →{0,1} n
where H is the hash function and n is the number of bits of the output hash value.
Each node is associated with a particular hash value, which can be expressed by the formula:
H(S i )＝H i
wherein S is i Represents a node, H i Represent S i Corresponding specific hash values.
By having one hash value for each node, minor node changes can also result in significant changes in hash values, which are easy to detect and verify. Compared with the traditional finite node machine, the node machine of the scheme is added with a cryptographic hash function to verify the correctness and data integrity of node conversion. This is the trustworthiness of the data in the blockchain and smart contract scenarios. The relation between the nodes and the events is defined in a mathematical form, and the hash calculation process in the node transfer is adopted, so that the behavior of the node machine is transparent and verifiable, and the security analysis and audit of the intelligent contract are facilitated. The node transfer condition of the node machine is integrated with verification of the legality of the input event, so that node abnormality caused by illegal input is avoided. The robustness of the node machine in the public chain environment is enhanced.
Optionally, step 203 may include sub-steps 2031-2032:
sub-step 2031, adding a self-attention mechanism in the neural network model for speech recognition, by which a preset weight matrix of keywords associated with the order is set;
Sub-step 2032, identifying and converting the voice data into text by a neural network model including the preset weight matrix;
the preset keywords comprise an intention keyword and an item keyword, wherein the intention keyword represents that a user acquires an item or does not acquire the item, and the item keyword is a keyword for describing the characteristics of the item.
In the embodiment of the present disclosure, the keywords include an intention keyword representing that the user acquires an item or does not acquire an item, and an item keyword, which is a keyword for describing characteristics of the item. For example, intent keywords may include "agree", "disagree", "order", "no order", and so on, which may indicate whether the user wants to purchase a package. The item keywords may be "voice packages," "traffic packages," and the like, representing items or packages targeted by the user.
The neural network model for Speech recognition may be a Deep Speech model in which a self-attention mechanism is added that can help the model focus on Speech segments that are more relevant to the current recognition task while ignoring or reducing the effects of other irrelevant or noisy segments. In the weight matrix of the self-attention model, higher weights are preset for the intention keywords and the item keywords. The weight matrix of the self-attention mechanism for intent keywords and item keywords can be expressed by the following formula:
Wherein, attention represents self-Attention mechanism, Q is Query matrix, representing the speech segment that needs to be decoded (identified) at present; k (Key matrix) and V (Value matrix) represent all or a portion of the input speech sequence. d, d k : dimension of Key matrix for scaling product QK T . Softmax is a normalization function used to convert an arbitrary set of real numbers into real numbers representing a probability distribution. As indicated by the letter Hadamard product, i.e., the multiplication of elements by elements, W is the weight matrix. By calculation ofThe resulting weights are used to weight the average Value matrix, the weight matrix W is used to adjust QK T Is a value of (2).
For example, the weight matrix may be as shown in the following table:
| polarity of emotion | Order intent | Mobile phone | Flow bag | Voice packet | |
| Consent to | 1 | 0 | 0 | 0 | 0 |
| Disagree with | -1 | 0 | 0 | 0 | 0 |
| Ordering of | 0 | 1 | 0 | 0 | 0 |
| Not ordered | 0 | -1 | 0 | 0 | 0 |
| Mobile phone | 0 | 0 | 0.8 | 0 | 0 |
| Flow bag | 0 | 0 | 0 | 0.6 | 0 |
| Voice packet | 0 | 0 | 0 | 0 | 0.3 |
In the above table, the emotion polarity and the order intent may all belong to the intent keyword, and the mobile phone, the traffic package and the voice package all belong to the item keyword. For example, the weight of the cell phone is set to 0.8, and the weight of the traffic packet is set to 0.6.
Optionally, step 203 may include sub-step 2033:
sub-step 2033, adding a time series weight vector into the neural network model for voice recognition, and setting different weights for data of a first half time period and a second half time period in the voice data through the time series weight vector;
Wherein the weight of the data of the first half period is smaller than the weight of the data of the second half period.
In the embodiment of the disclosure, in voice communication with an outbound call, important information related to the subscription, such as the name of the item, the time of failure of the item, whether the user agrees, etc., will be broadcast. This information is often a key factor in determining whether a user will make a subscription. Thus, in calculating the self-attention weight, an additional weight is given to the voice data end (i.e. the part broadcasting the subscription related information), for example, the voice data may be equally divided into two parts, a first half part and a second half part; the last ten seconds or five seconds of the voice data may be set as the second half, the rest as the first half, and the weight of the voice data in the second half may be greater than that in the first half.
The timing weight vector may be dynamically adjusted by a decision tree model in which element values are set according to the time position of the speech data. The time sequence weight vector T is a one-dimensional vector having the same length as the number of time steps of the input sequence. Each element in T represents a weight coefficient for that time step. The value of which is usually in the range of 0 to 1. The weight of the current time step can be adjusted according to the attention allocation situation of the previous time step through dynamic prediction of the decision tree model. In a typical outbound voice, the last 10 seconds is critical information about the ordering of the item. In this case, T may be designed to have a higher weight value within the 10 seconds. For example, if the total length of speech is 60 seconds, the last 10 elements of T (corresponding to the last 10 seconds) are set to a higher value than the first 50 elements. Assuming that the last 5 seconds of speech is used to confirm whether the user agrees to the subscription, this is a very critical decision point, in which case T is designed to have a higher weight value in the last 5 seconds to ensure that the model can accurately capture the user's decision. Thus, it can be understood that the voice data may be divided into a plurality of parts in practice, but as with the division into two parts, the part farther toward the end of the voice data is set to be larger than the voice data weight value closer to the front end of the voice data. The setting of the weight matrix for the preset keyword and the setting of the time sequence weight vector can be represented by the following formula:
Wherein, attention represents self-Attention mechanism, Q is Query matrix, representing the speech segment that needs to be decoded (identified) at present; k (Key matrix) and V (Value matrix) represent all or a portion of the input speech sequence. d, d k Is the dimension of the Key matrix used to scale the product QK T . Softmax is a normalization function used to convert an arbitrary set of real numbers into real numbers representing a probability distribution. As indicated by the letter Hadamard product, i.e., the multiplication of elements by elements. W is a weight matrix. T is the timing weight vector.
The decision tree model may also predict the optimal weight for the current time step based on information of the previous time steps (previous attention weights, input features, etc.).
T t ＝DecisionTree(T t-1 ,T t-2 ,…,T t-n )
Here, T t Is the time sequence weight of the current time step, T t-1 ，T t-2 ，…,T t-n Is the timing weight of the previous n time steps and decision tree is the decision tree model.
Optionally, step 204 may include sub-steps 2041-2042:
a sub-step 2041 of extracting, based on the text, the intention keywords and entity information of the user in the text through a semantic understanding model; the entity information comprises user information and the item keywords;
sub-step 2042 generates keyword data comprising the intent keywords and the entity information.
In the disclosed embodiments, the semantic understanding model may be a natural language understanding model (NLU, natural Language Understanding), such as Rasa NLU model. After the speech is converted to text, the text is further normalized and structured using Rasa's Natural Language Understanding (NLU) module. This may generate data that can be used by the user management system. The Rasa NLU model may identify multiple intents in the text, for example, the text may be "how is the weather of city a tomorrow? "including two intents, the first is" how weather in city a "and the second is" how weather in tomorrow ".
It can be appreciated that in order to improve accuracy of the model, regular expressions and look-up tables may be used to add to training data by which to train the Rasa NLU model, so that the Rasa NLU model may have more accurate entity information. Regular expressions may be used to describe patterns of a set of string features, for example, using a regular expression ((\d {4} - \d {2} - }) \) to match a date format of the type "2023-09-07" may enable the Rasa NLU model to identify dates of that type. The lookup table comprises the intention key words, the entity information and the information with preset similarity with the intention key words or the entity information. For example, an item and an alias or alias for the item are present in a lookup table, such as a seat and seat, etc.
Keywords in the text are extracted through a natural language understanding model, keyword data comprising the intended keywords and the entity information are directly generated, the keyword data can be structured data, and the format can be JS key value data (JSON, javaScript Object notification) format. For example, a specific data structure may be as follows:
- "intent" means the intention of the user.
Identity information including the identified entity information.
```json
{ "intent": "acquire",
"user_intent": "agree to"
"entities":[
{"entity":"cust_name"，
"value" means "Zhang Sano"
"entity":"eff_date"，
"value":"2023-08-25"。
Optionally, step 205 may include sub-steps 2051-2052:
sub-step 2051, storing the keyword data in a blockchain, entering the generating node, and triggering to send the keyword data and an information construction request to a user management system according to the execution information of the generating node;
a substep 2052, generating, by the user management system, demand information of the user based on the information construction request and the keyword data;
wherein the user management system is a system for generating, deleting or modifying relevant information of the user.
In the disclosed embodiments, blockchain is a distributed ledger (Distributed Ledger) technology that maintains a single non-tamperable and highly transparent data record in common across multiple nodes on a network. Each data record is called a "Block" (Block) and a plurality of blocks are linked together by a cryptographic method to form a "Chain" (Chain) in which each Block contains the following main fields: a Block Header (Block Header) including a time stamp, a hash value of a previous Block, etc.; the Transaction List includes all Transaction data in the block. One block B can be expressed as:
B＝(BlockHeader,TransactionList)
Wherein, the BlockHeader is a block header, and the TransactionList is a transaction list.
Each time a new outbound campaign or user interaction occurs, a corresponding transaction T is created and added to the transaction pool. The transaction structure is as follows:
T＝(State,H(Data),Timestamp)
where State is the current node of the smart contract, H (Data) is the hash value of the relevant Data, and Timestamp is the Timestamp of transaction creation.
It will be appreciated that each new chunk on the blockchain contains the hash value of the previous chunk, forming a hash chain:
H(B i )＝H(B i |H(B i-1 ))
wherein H represents a hash value, B 1 Representing a block.
As shown in fig. 4, in the a scenario of fig. 4, the blockchain includes at least one block; in the B scenario of fig. 4, a block includes a block header and a block body; in the C scenario of FIG. 4, the chunk header includes a timestamp, the hash value of the last chunk, and a random value; in the D scenario of FIG. 4, the transaction is included in the tile body; in the E scenario of fig. 4, the information included in the transaction is, for example, the transaction from and to, as well as the value and data.
Keyword data is stored on the blockchain, i.e., the keyword data is uplinked, which refers to the process of recording data or transactions on the blockchain. And storing the keyword data in the blockchain, entering a generating node, and triggering to send the keyword data and the information construction request to a user management system according to the execution information of the generating node. And the user management system receives the information construction request and generates flow result information of the user according to the keyword data.
For example, the keyword data includes user a information, package name B, package duration C, etc., and the user management system may generate an order of package duration C for user a.
In the block header, identification number fields of the outbound system and the subscriber management system are placed for recording the data source of the block. The transaction data structure of the blockchain is added with a node field and a hash value field for recording the current outbound node and data hash. The transaction data is added with an outbound record field, and key data such as client identification, outbound time and the like are stored in the outbound record field. The transaction data incorporates an order field storing generated order key information such as order number, order content, etc. And the transaction data is stored in an extensible JSON format, so that customized data of outbound calls and orders can be conveniently inserted. The add timestamp field records the exact time of the order and outbound for order verification. And a user confirmation field is added, and confirmation information of the user in the outbound process is recorded, so that authenticity is ensured. The transaction memory structure is optimized so that key data related to outbound calls and orders can be retrieved quickly. The data structure is compatible with the existing blockchain system, and the customized storage is realized through the extension field. And the exception and security event records are enhanced, so that the post-hoc traceability or audit is facilitated. The above design improves the applicability and usability of the blockchain in outbound business scenarios.
May be encrypted by symmetry E sym (D, K) and asymmetric encryption E asym (D,K pub ) Is used to encrypt the user-related data. Wherein E is sym And E is asym The symmetric encryption and the asymmetric encryption functions, respectively, D is the data to be encrypted, K and K pub A symmetric key and a public key, respectively.
The correctness and the safety of the logic of the intelligent contract can be ensured by auditing the codes of the intelligent contract. The audit process can be expressed in mathematical logic as:
Audit(C)＝Verify(L(C))
where C is the code of the smart contract and Audit (C) is the Audit result of the code of the smart contract. L (C) is a logical representation of the code of the smart contract, verify (·) is a validation function.
In the blockchain, each transaction or event generates an associated hash value and digital signature for subsequent data verification.
Transaction Validation＝Verify(Sign(H(E)),H(E))
Where Transaction Validation is transaction verification, E is transaction or event, H (E) is a hash of E, sign (H (E)) is a digital signature, and Verify is a verification function.
The compliance of the intelligent contract can also be checked, specifically by the following formula:
ComplianceCheck(T)＝F(T,R)
where ComplipenceCheck represents a compliance check, T is the transaction or operation to be evaluated, R is the relevant regulations and prescriptions, and F is the compliance check function.
Security audit and risk assessment mechanisms may also be included, by which the system may be security checked using a security audit model M. The model can be expressed as:
SecurityAudit(S)＝M(S,P)
the security audio represents security check, S is the node or state where S is located, P is the security policy, and M is the security audit model.
A security assessment model may also be included to assess security levels:
Security Level＝Evaluate(Encryption,Compliance)
where Security Level represents Security Level, encryption is data Encryption Level, company is Compliance Level, and evaluation represents Security assessment model.
Benefit evaluation may also be performed based on a cost-benefit model:
wherein C is t And C s The total cost of the conventional system and the above embodiments, B t And B s The Cost-Benefit Ratio is the Cost-Benefit Ratio, if it is the total Benefit of the conventional system and the above-described embodiments, respectively>1, it is illustrated that the above embodiment is more cost effective.
Optionally, step 203 may further comprise sub-step 2034:
a substep 2034 of obtaining text for a plurality of users from a corresponding plurality of sets of voice data based on calls for the plurality of users;
and updating a preset weight matrix of the preset keywords based on the occurrence frequency of the preset keywords in the texts of the plurality of users.
In the embodiment of the disclosure, the outbound system can call a plurality of users, and in the calling process, multiple groups of voice data are acquired and texts of the multiple users are acquired. And acquiring the occurrence frequency of the preset keywords from the texts of a plurality of users, and updating the preset weight matrix of the preset keywords. The method may be that according to the fact that the occurrence frequency of a certain preset keyword is low in the texts of a plurality of users, a weight value corresponding to the preset keyword in a preset weight matrix is reduced; the method may be that according to the fact that the occurrence frequency of a certain preset keyword is higher in the texts of the multiple users, a weight value corresponding to the preset keyword in a preset weight matrix is increased.
Optionally, the method further comprises steps 207-208:
step 207, obtaining the data generated by the user management system and the outbound system, and sending the data to a message queue;
step 208, based on the message queue, acquiring the data of the outbound system through the user management system, and acquiring the data of the user management system through the outbound system.
In the embodiment of the disclosure, in order to realize the data synchronization of the outbound system and the user management system, a message queue may be used, where the message queue is a container for storing messages in the transmission process of the messages. The data generated by the user management system and the outbound system can be acquired respectively and sent to the message queue. Then, the user management system can acquire data generated by the outbound system through the message queue, and acquire the data of the user management system through the outbound system. The expression can be expressed by the following formula:
Wherein, MQ represents a message queue, outbound System represents an Outbound System, and CRM System represents a user management System.
Real-time data flow integration between two systems can be realized through the message queue, and business coordination is realized through the configured message exchange logic shielding the difference between the two systems.
As shown in FIG. 5, outbound system 501 and user management system 502 both interact with blockchain 503, outbound system 501 synchronizes data to user management system 502, outbound system 501 integrates data to blockchain 503, and user management system 502 integrates data to blockchain 503.
Optionally, sub-step 2031 may include steps A1-A3:
a1, acquiring a corpus comprising the preset keywords, and determining word frequency and inverse document frequency of the preset keywords in the corpus;
step A2, calculating a weight value of each preset keyword based on the word frequency and the inverse document frequency;
and A3, setting a preset weight matrix of the preset keywords through the self-attention mechanism based on the weight value of each preset keyword.
In the embodiment of the disclosure, the keyword may be assigned and set by using a word Frequency-inverse document Frequency (TF-IDF, term Frequency-Inverse Document Frequency), which is a statistical method for evaluating the importance of a word to one of a set of documents or a corpus. The importance of a word increases proportionally with the number of times it appears in the file, but at the same time decreases inversely with the frequency with which it appears in the corpus. For example, file a is in the file set, the frequency of occurrence of the word "good" in file a is high, and the frequency of occurrence in other files in the file set is low, so that the weight value of the word "good" is high; the word "good" appears very frequently in file a and also in other files in the set of files, and the weight value of this word "good" is lower than in the former case. By acquiring the weight value of each preset keyword, a preset weight matrix can be acquired, and the preset weight matrix is set in the self-attention mechanism.
In summary, a flow configuration file is obtained, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node; entering the initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls; entering the answering node after the call is connected, acquiring voice data of the user according to the execution information of the answering node, and converting the voice data into text; entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user; and entering the generating node after the keyword data are stored, and generating flow result information of the user based on the keyword data according to the execution information of the generating node. The method and the device can combine the nodes based on the progress of the outbound process according to the pre-configured process configuration file, correspondingly execute related actions, finally generate the process result information of the user, automatically generate the user information of the outbound process, and improve the information generation efficiency.
A flow chart of interactions between the outbound system, blockchain, and subscriber management system is provided herein, as shown in fig. 6. Step S1, an outbound system sends voice data to a block chain, and the block chain receives the voice data; step S2, after receiving voice data, the block chain generates a hash value and returns the hash value to the outbound system; step S3, the outbound system sends voice data to the client management system, and the client management system receives the voice data; and S4, the client management system receives the voice data and returns flow result information.
Fig. 7 is a block diagram of a service inspection device according to an embodiment of the present invention, where the service inspection device 70 may include:
an obtaining module 701, configured to obtain a flow configuration file, where the flow configuration file includes an initial node, an answer node, an extraction node, a generation node, and execution information corresponding to each node;
an initial module 702, configured to enter the initial node, trigger an outbound system to initiate a call to a user according to execution information of the initial node; the outbound system is a communication system for making telephone calls;
an answering module 703, configured to enter the answering node after the call is completed, obtain voice data of the user according to execution information of the answering node, and convert the voice data into text;
An extracting module 704, configured to enter the extracting node after the call is ended, extract keywords from the text according to the execution information of the extracting node, and generate keyword data of the user;
and the generating module 705 is configured to enter the generating node after the keyword data is stored, and generate, according to the execution information of the generating node, flow result information of the user based on the keyword data.
Optionally, the answering module includes:
the weight sub-module is used for adding a self-attention mechanism into the neural network model for voice recognition, and setting a preset weight matrix of preset keywords through the self-attention mechanism;
the first recognition sub-module is used for recognizing the voice data through a neural network model comprising the preset weight matrix and converting the voice data into a text;
the preset keywords comprise an intention keyword and an item keyword, wherein the intention keyword represents that a user acquires an item or does not acquire the item, and the item keyword is a keyword for describing the characteristics of the item.
Optionally, the answering module includes:
a time sequence weight sub-module, which is used for adding a time sequence weight vector into a neural network model for voice recognition, and setting different weights for data of a first half time period and a second half time period in the voice data through the time sequence weight vector;
The second recognition sub-module is used for recognizing the voice data through a neural network model comprising the time sequence weight vector and converting the voice data into text;
wherein the weight of the data of the first half period is smaller than the weight of the data of the second half period.
Optionally, the extraction module includes:
the keyword extraction sub-module is used for extracting the intention keywords and entity information of the user in the text through a semantic understanding model based on the text; the entity information comprises user information and the item keywords;
and the keyword data generation sub-module is used for generating keyword data comprising the intention keywords and the entity information.
Optionally, the apparatus further comprises:
the identifier generation module is used for generating identifiers corresponding to any node through a preset function when entering any node of the initial node, the answering node, the extracting node and the generating node;
the identification is used for checking the data integrity of any node.
Optionally, the generating module includes:
the request sub-module is used for storing the keyword data in a blockchain and entering the generating node, and triggering to send the keyword data and the information construction request to a user management system according to the execution information of the generating node;
The information generation sub-module is used for generating flow result information of the user through the user management system based on the information construction request and the keyword data;
wherein the user management system is a system for generating, deleting or modifying relevant information of the user.
Optionally, the apparatus further comprises:
the first updating module is used for acquiring texts of a plurality of users through corresponding multiple groups of voice data based on calls of the users;
and the second updating module is used for updating the preset weight matrix of the preset keywords based on the occurrence frequency of the preset keywords in the texts of the plurality of users.
Optionally, the apparatus further comprises:
the training module is used for training the semantic understanding model through a regular expression and a lookup table before the step of extracting the intention keywords and the entity information of the user in the text through the semantic understanding model;
the regular expression is used for describing a pattern of a group of character string characteristics, and the lookup table comprises the intention keyword, the entity information and the information with preset similarity with the intention keyword or the entity information.
Optionally, the apparatus further comprises:
the data sending module is used for acquiring the data generated by the user management system and the outbound system and sending the data to a message queue;
and the data acquisition module is used for acquiring the data of the outbound system through the user management system based on the message queue and acquiring the data of the user management system through the outbound system.
Optionally, the weighting submodule includes:
the parameter acquisition unit is used for acquiring a corpus comprising the preset keywords and determining word frequency and inverse document frequency of the preset keywords in the corpus;
the weight calculation unit is used for calculating the weight value of each preset keyword based on the word frequency and the inverse document frequency;
and the matrix setting unit is used for setting a preset weight matrix of the preset keywords through the self-attention mechanism based on the weight value of each preset keyword.
In summary, a flow configuration file is obtained, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node; the method comprises the steps of carrying out a first treatment on the surface of the Entering the initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls; entering the answering node after the call is connected, acquiring voice data of the user according to the execution information of the answering node, and converting the voice data into text; entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user; and entering the generating node after the keyword data are stored, and generating flow result information of the user based on the keyword data according to the execution information of the generating node. The method and the device can combine the nodes based on the progress of the outbound process according to the pre-configured process configuration file, correspondingly execute related actions, finally generate the process result information of the user, automatically generate the user information of the outbound process, and improve the information generation efficiency.
The present invention also provides an electronic device, see fig. 8, comprising: a processor 801, a memory 802, and a computer program 8021 stored on the memory and executable on the processor, the processor implementing the information generating method of the foregoing embodiment when executing the program.
The present invention also provides a readable storage medium which, when executed by a processor of an electronic device, enables the electronic device to perform the information generating method of the foregoing embodiment.
For the device embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and reference is made to the description of the method embodiments for relevant points.
It should be noted that, various information and data acquired in the embodiment of the present invention are acquired under the condition that the information/data holder is authorized.
The algorithms and displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general-purpose systems may also be used with the teachings herein. The required structure for a construction of such a system is apparent from the description above. In addition, the present invention is not directed to any particular programming language. It will be appreciated that the teachings of the present invention described herein may be implemented in a variety of programming languages, and the above description of specific languages is provided for disclosure of enablement and best mode of the present invention.
In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the above description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be construed as reflecting the intention that: i.e., the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the apparatus of the embodiments may be adaptively changed and disposed in one or more apparatuses different from the embodiments. The modules or units or components of the embodiments may be combined into one module or unit or component and, furthermore, they may be divided into a plurality of sub-modules or sub-units or sub-components. Any combination of all features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or units of any method or apparatus so disclosed, may be used in combination, except insofar as at least some of such features and/or processes or units are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings), may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functions of some or all of the components in a sorting device according to the present invention may be implemented in practice using a microprocessor or Digital Signal Processor (DSP). The present invention may also be implemented as an apparatus or device program for performing part or all of the methods described herein. Such a program embodying the present invention may be stored on a computer readable medium, or may have the form of one or more signals. Such signals may be downloaded from an internet website, provided on a carrier signal, or provided in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third, etc. do not denote any order. These words may be interpreted as names.
The user information (including but not limited to user equipment information, user personal information, etc.), related data, etc. related to the present invention are all information authorized by the user or authorized by each party.
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, and are not repeated herein.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
The foregoing is merely illustrative embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think about variations or substitutions within the technical scope of the present invention, and the invention should be covered. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (13)
1. An information generation method, characterized in that the method comprises:
acquiring a flow configuration file, wherein the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node;
Entering the initial node, and triggering an outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls;
entering the answering node after the call is connected, acquiring voice data of the user according to the execution information of the answering node, and converting the voice data into text; the method comprises the steps of carrying out a first treatment on the surface of the
Entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user;
and entering the generating node after the keyword data are stored, and generating flow result information of the user based on the keyword data according to the execution information of the generating node.
2. The method of claim 1, wherein the step of converting the speech data to text comprises:
adding a self-attention mechanism into a neural network model for voice recognition, and setting a preset weight matrix of preset keywords through the self-attention mechanism;
identifying the voice data through a neural network model comprising the preset weight matrix and converting the voice data into a text;
The preset keywords comprise an intention keyword and an item keyword, wherein the intention keyword represents that a user acquires an item or does not acquire the item, and the item keyword is a keyword for describing the characteristics of the item.
3. The method of claim 1, wherein the step of converting the speech data to text comprises:
adding a time sequence weight vector into a neural network model for voice recognition, and setting different weights for data of a first half time period and a second half time period in the voice data through the time sequence weight vector;
identifying and converting the voice data into text through a neural network model comprising the time sequence weight vector;
wherein the weight of the data of the first half period is smaller than the weight of the data of the second half period.
4. The method of claim 2, wherein the step of keyword extraction of the text and generating the user's keyword data comprises:
extracting intention keywords and entity information of the user in the text through a semantic understanding model based on the text; the entity information comprises user information and the item keywords;
And generating keyword data comprising the intention keywords and the entity information.
5. The method according to claim 1, wherein the method further comprises:
when any node of the initial node, the answering node, the extracting node and the generating node is entered, generating a corresponding identifier of the any node through a preset function;
the identification is used for checking the data integrity of any node.
6. The method according to claim 1, wherein the step of entering the generating node after the keyword data is stored, generating the flow result information of the user based on the keyword data according to the execution information of the generating node, comprises:
storing the keyword data in a blockchain, entering the generating node, and triggering to send the keyword data and an information construction request to a user management system according to the execution information of the generating node;
generating flow result information of the user through the user management system based on the information construction request and the keyword data;
wherein the user management system is a system for generating, deleting or modifying relevant information of the user.
7. The method according to claim 2, wherein the method further comprises:
acquiring texts of a plurality of users through corresponding multiple groups of voice data based on calls of the users;
and updating a preset weight matrix of the preset keywords based on the occurrence frequency of the preset keywords in the texts of the plurality of users.
8. The method of claim 4, further comprising, prior to the step of extracting the user's intent keywords and entity information in the text by a semantic understanding model:
training the semantic understanding model through a regular expression and a lookup table;
the regular expression is used for describing a pattern of a group of character string characteristics, and the lookup table comprises the intention keyword, the entity information and the information with preset similarity with the intention keyword or the entity information.
9. The method of claim 6, wherein the method further comprises:
acquiring data generated by the user management system and the outbound system and sending the data to a message queue;
based on the message queue, the data of the outbound system is acquired through the user management system, and the data of the user management system is acquired through the outbound system.
10. The method according to claim 2, wherein the step of setting a preset weight matrix of preset keywords by the self-attention mechanism comprises:
acquiring a corpus comprising the preset keywords, and determining word frequency and inverse document frequency of the preset keywords in the corpus;
calculating a weight value of each preset keyword based on the word frequency and the inverse document frequency;
and setting a preset weight matrix of the preset keywords through the self-attention mechanism based on the weight value of each preset keyword.
11. An information generating apparatus, comprising:
the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring a flow configuration file, and the flow configuration file comprises an initial node, a receiving node, an extracting node, a generating node and execution information corresponding to each node;
the initial module is used for entering the initial node, and triggering the outbound system to initiate a call to a user according to the execution information of the initial node; the outbound system is a communication system for making telephone calls;
the answering module is used for entering the answering node after the call is switched on, acquiring the voice data of the user according to the execution information of the answering node, and converting the voice data into a text;
The extraction module is used for entering the extraction node after the call is ended, extracting keywords from the text according to the execution information of the extraction node, and generating keyword data of the user;
and the generation module is used for entering the generation node after the keyword data are stored, and generating the flow result information of the user based on the keyword data according to the execution information of the generation node.
12. An electronic device, comprising:
a processor, a memory and a computer program stored on the memory and executable on the processor, the processor implementing the method according to any one of claims 1-10 when executing the program.
13. A readable storage medium, characterized in that instructions in the readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the method of any one of claims 1-10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311498439.7A CN117651093A (en) | 2023-11-10 | 2023-11-10 | Information generation method, device, electronic equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311498439.7A CN117651093A (en) | 2023-11-10 | 2023-11-10 | Information generation method, device, electronic equipment and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117651093A true CN117651093A (en) | 2024-03-05 |
Family
ID=90046921
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311498439.7A Pending CN117651093A (en) | 2023-11-10 | 2023-11-10 | Information generation method, device, electronic equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117651093A (en) |
-
2023
- 2023-11-10 CN CN202311498439.7A patent/CN117651093A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11714793B2 (en) | Systems and methods for providing searchable customer call indexes | |
| CN110352425B (en) | Cognitive regulatory compliance automation for blockchain transactions | |
| US10847136B2 (en) | System and method for mapping a customer journey to a category | |
| US20200118009A1 (en) | Improved onboarding of entity data | |
| US11955113B1 (en) | Electronic signatures via voice for virtual assistants' interactions | |
| US11687946B2 (en) | Systems and methods for detecting complaint interactions | |
| CN111222308A (en) | Case decision book generation method and device and electronic equipment | |
| CN107977678A (en) | Method and apparatus for output information | |
| CN111783144A (en) | Data processing method and device based on block chain | |
| CN113159901A (en) | Method and device for realizing financing lease service session | |
| CN112016850A (en) | Service evaluation method and device | |
| CN112837149A (en) | Method and device for identifying enterprise credit risk | |
| CN113609271B (en) | Knowledge graph-based service processing method, device, equipment and storage medium | |
| CN111666059A (en) | Reminding information broadcasting method and device and electronic equipment | |
| CN117651093A (en) | Information generation method, device, electronic equipment and readable storage medium | |
| US11681966B2 (en) | Systems and methods for enhanced risk identification based on textual analysis | |
| CN108062379B (en) | Data processing method, platform, device and computer readable storage medium | |
| CN110879835A (en) | Data processing method, device and equipment based on block chain and readable storage medium | |
| CN114662007B (en) | Data recommendation method and device, computer equipment and storage medium | |
| WO2023119520A1 (en) | Estimation device, estimation method, and program | |
| CN112653795A (en) | Identity verification method and related device | |
| Sergiu | USING MACHINE LEARNING ALGORITHMS TO DETECT FRAUDS IN TELEPHONE NETWORKS. | |
| US20150340026A1 (en) | Extracting candidate answers for a knowledge base from conversational sources | |
| CN115618120A (en) | Public number information pushing method, system, terminal equipment and storage medium | |
| CN115914463A (en) | Risk detection method and device and electronic 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 |