Business processing method and device, nonvolatile storage medium and electronic equipment

Abstract

The invention discloses a business processing method and device, a nonvolatile storage medium and electronic equipment. The method comprises the following steps: performing feature extraction on a first service dialogue text to obtain a first feature set of the first service dialogue text; determining an alternative service scene from a plurality of preset service scenes according to the first feature set; respectively determining target distinguishing feature names of the plurality of alternative service scenes; determining a target feature quantity value, corresponding to the target distinguishing feature name, of the first business dialogue text, and determining a first target business scene matched with the first business dialogue text from a plurality of alternative business scenes according to the target feature quantity value; and generating a service reply text corresponding to the first service dialogue text according to the first target service scene. The technical problem that the reply text matched with the text input by the user cannot be accurately generated due to the fact that the business scene corresponding to the text input by the user cannot be accurately determined in the related technology is solved.

Description

Technical Field

[0001] The present application relates to the field of text data processing, and in particular, to a business processing method, device, non-volatile storage medium and electronic device. Background Art

[0002] In the related art, when natural language processing technology is used to reply to the text input by the user, it is necessary to first identify the business scenario corresponding to the user input text. However, in actual application, since some scenarios are similar, it is impossible to determine which specific scenario the user input text corresponds to, resulting in that the reply text generated in the related art usually cannot match the user input text.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present application provide a business processing method, device, non-volatile storage medium and electronic device to at least solve the technical problem in the related art that the business scenario corresponding to the user input text cannot be accurately determined, resulting in the inability to accurately generate a reply text that matches the user input text.

[0005] According to one aspect of an embodiment of the present application, a business processing method is provided, including: performing feature extraction on a first business conversation text to obtain a first feature set of the first business conversation text; determining an alternative business scenario from a plurality of preset business scenarios based on the first feature set; in the case where there are multiple alternative business scenarios, respectively determining target distinguishing feature names of the multiple alternative business scenarios, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; determining a target feature value corresponding to the first business conversation text and the target distinguishing feature name, and determining a first target business scenario matching the first business conversation text from a plurality of alternative business scenarios based on the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text; and generating a business reply text corresponding to the first business conversation text based on the first target business scenario.

[0006] Optionally, determining target distinguishing feature names for multiple alternative business scenarios includes: determining sample feature elements for each alternative business scenario, wherein the sample feature element for each alternative business scenario includes a feature name and a feature value corresponding to the feature name, wherein the feature value corresponding to the feature name is used to indicate the number of occurrences of the feature corresponding to the feature name in the alternative business scenario; for each feature name, determining a value dispersion coefficient of the feature value of each alternative business scenario, wherein the value dispersion coefficient is used to reflect the degree of dispersion of the feature value of each alternative business scenario, and the larger the value dispersion coefficient, the greater the degree of dispersion; determining the feature name with the largest value dispersion coefficient as the target distinguishing feature name.

[0007] Optionally, for each feature name, determining the value dispersion coefficient of the feature value of each candidate business scenario includes: for each feature name, determining the variance between the feature values ​​of each candidate business scenario, and determining the variance as the value dispersion coefficient.

[0008] Optionally, determining the first target business scenario that matches the first business dialogue text from multiple alternative business scenarios based on the target feature value includes: determining the sample feature value of each alternative business scenario relative to the target distinguishing feature name; in the case where there is a sample feature value equal to the target feature value, determining the alternative business scenario whose corresponding sample feature value is equal to the target feature value as the first target business scenario; in the case where there is no sample feature value equal to the target feature value, determining the absolute value of the difference between the sample feature value of each alternative business scenario and the target feature value, and determining the alternative business scenario whose corresponding difference absolute value is the smallest as the first target business scenario.

[0009] Optionally, determining alternative business scenarios from multiple preset business scenarios based on the first feature set includes: determining a business network topology model, wherein each node in the business network topology model corresponds to a business scenario, and different nodes correspond to different business scenarios, and the nodes are connected based on the relevance of business logic between the business scenarios corresponding to the nodes; determining a matching degree index between the business scenarios corresponding to each node and the first business dialogue text based on the connection relationship between the first feature set and each node in the business network topology model; and determining a business scenario whose matching degree index is greater than a preset threshold as an alternative business scenario.

[0010] Optionally, the business processing method also includes: after receiving the second business conversation text, performing feature extraction on the second business conversation text to obtain a second feature set of the second business conversation text; determining a matching degree index between the second business conversation text and the first target business scenario based on the second feature set; when the matching degree index is lower than a preset threshold, determining a second target business scenario that matches the second business conversation text, and generating a business reply text corresponding to the second business conversation text based on the second target business scenario.

[0011] Optionally, the feature type of the first feature set includes at least one of the following: lexical feature, grammatical feature and semantic feature.

[0012] According to another aspect of an embodiment of the present application, a business processing device is also provided, including: a first processing module, used to extract features from a first business conversation text to obtain a first feature set of the first business conversation text; a second processing module, used to determine alternative business scenarios from multiple preset business scenarios based on the first feature set; a third processing module, used to determine target distinguishing feature names of multiple alternative business scenarios respectively when there are multiple alternative business scenarios, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; a fourth processing module, used to determine the target feature values ​​corresponding to the target distinguishing feature names of the first business conversation text, and determine a first target business scenario matching the first business conversation text from multiple alternative business scenarios based on the target feature values, wherein the target feature values ​​are used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text; a fifth processing module, used to generate a business reply text corresponding to the first business conversation text based on the first target business scenario.

[0013] According to another aspect of an embodiment of the present application, a non-volatile storage medium is provided, in which a program is stored, wherein when the program is running, a device where the non-volatile storage medium is located is controlled to execute a business processing method.

[0014] According to another aspect of an embodiment of the present application, there is further provided an electronic device, including: a memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the business processing method is executed when the program is run.

[0015] According to another aspect of an embodiment of the present application, a computer program product is also provided, including a computer program, and the computer program implements the business processing method when executed by a processor.

[0016] In an embodiment of the present application, a feature extraction is performed on a first business conversation text to obtain a first feature set of the first business conversation text; an alternative business scenario is determined from a plurality of preset business scenarios based on the first feature set; in the case where there are multiple alternative business scenarios, target distinguishing feature names of the multiple alternative business scenarios are respectively determined, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; a target feature value corresponding to the first business conversation text and the target distinguishing feature name is determined, and a first target business scenario matching the first business conversation text is determined from the plurality of alternative business scenarios based on the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text; based on the first A method for generating a business reply text corresponding to a first business dialogue text from a target business scenario, by determining a target distinguishing feature name of a target distinguishing feature that can distinguish multiple alternative business scenarios, and determining a target feature value of the first business dialogue text based on the target distinguishing feature name, and then determining a first target business scenario that matches the first business dialogue text based on the target feature value, thereby achieving the purpose of accurately determining a business scenario that matches the first business dialogue text from multiple alternative business scenarios, thereby achieving the technical effect of accurately generating a business reply text that matches the first business dialogue text, and thereby solving the technical problem in the related art that the business scenario corresponding to the user input text cannot be accurately determined and the reply text that matches the user input text cannot be accurately generated. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 It is a structural schematic diagram of a computer terminal according to the related art;

[0019] Figure 2 It is a flowchart of a business processing method provided according to an embodiment of the present application;

[0020] Figure 3 It is a flow chart of a process for calculating a discrete coefficient of magnitude provided in an embodiment of the present application;

[0021] Figure 4 It is a flowchart of a reply text generation process provided according to an embodiment of the present application;

[0022] Figure 5 It is a structural diagram of a business processing device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] In order to better understand the embodiments of the present application, the technical terms involved in the embodiments of the present application are explained as follows:

[0026] NLP: NLP (Natural Language Processing) refers to natural language processing technology, which is a branch of artificial intelligence that aims to enable computers to understand, interpret and generate human language. This technology covers many different tasks and applications, including text classification, named entity recognition, sentiment analysis, machine translation, language generation, etc. These methods enable computers to learn the structure and rules of language through a large amount of language data, and can perform semantic understanding and context analysis of text.

[0027] Feature engineering refers to the conversion and processing of raw data in natural language machine learning and data mining to create new features that can better describe the characteristics of the data. These new features can help machine learning models better understand the data and improve the performance and generalization ability of the model. Feature elements can be the frequency of word occurrence, sentence length, part-of-speech tagging, named entity recognition results, etc. Through the creation of these feature elements, the original text data can be converted into a form that is easier for machine learning models to understand, thereby improving the performance of the model. These feature elements can help the model better capture the semantic and structural information of text data, thereby improving the performance of the model on various tasks.

[0028] With the rapid development of information technology, natural language processing (NLP) technology has become an important part of the human-computer interaction field. Especially in scenarios such as enterprise services, intelligent customer service, and online education, NLP technology can greatly improve user experience and achieve efficient and accurate automatic responses. At present, the existing technology directly uses natural language processing technology for language processing or scene recognition. For example, CN116992397A discloses a business scene recognition method, electronic device, and storage medium. When receiving a question consultation request sent by a target user, the user question and user background information of the target user are obtained, and the user question and user background are evaluated by natural language processing technology to obtain an effective background score. The target business scene corresponding to the user question is determined based on the effective background score set. Direct scene recognition is used. If the similarity between scenes is relatively strong, there will be problems such as low recognition accuracy and scene matching failure.

[0029] Therefore, in the face of complex and ever-changing business scenarios and the high similarity of scenarios, the single feature extraction and recognition method in the existing technology often fails to achieve ideal results. In particular, in the telecommunications dialogue scenario, there are many business branches, each branch has small branches, and the scenarios between each branch are very similar. Ordinary language feature extraction often matches many similar business scenarios of small branches, making it impossible to accurately identify and respond.

[0030] In order to solve the above problems, relevant solutions are provided in the embodiments of the present application, which are described in detail below.

[0031] According to an embodiment of the present application, a method embodiment of a business processing method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0032] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing a service processing method. Figure 1As shown, the computer terminal 10 (or mobile device 10) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0033] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0034] The memory 104 can be used to store software programs and modules of application software, such as program instructions / data storage devices corresponding to the business processing method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the above-mentioned business processing method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0035] The transmission device 106 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a Radio Frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0036] The display can be, for example, a touch-screen liquid crystal display (LCD), which enables users to interact with the user interface of the computer terminal 10 (or mobile device).

[0037] Under the above operating environment, an embodiment of the present application provides a service processing method, as Figure 2 shown, the method includes the following steps:

[0038] Step S202, extract features from the first service dialogue text to obtain a first feature set of the first service dialogue text;

[0039] In the technical solution provided in step S202, the feature types of the first feature set include at least one of the following: lexical features, syntactic features, and semantic features.

[0040] As an alternative implementation, the NLP natural language processing system can be connected from the service intelligent customer service system to extract the consultation information of the current user (that is, the first service dialogue text), and then perform multi-feature extraction on the consultation information of the current user to obtain the first feature set. The first feature set includes lexical features, syntactic features, and semantic features. These features can help extract information from different aspects of the text, so as to better understand the user's intention and emotion.

[0041] Among them, lexical features include the frequency of each word appearing in the text, the number of different words in the text and their coverage, extract the most representative or key words in the text, and remove common stop words, such as "de", "shi", "zai", etc., to reduce noise. Syntactic features include the average length of sentences in the text, the longest sentence length, etc., analyze the sentence structure, such as subject-predicate-object, parallel relationship, etc., identify the part of speech of each word in the text, such as noun, verb, adjective, etc., and analyze the emotional tendency of the sentence, such as positive, negative or neutral. Semantic features include identifying the theme or topic in the text to understand the main content of the text, analyzing the emotional expression in the text, such as joy, anger, sorrow, etc., replacing synonyms in the text with unified words to reduce the feature dimension, and calculating the semantic similarity between texts for relevance matching or recommendation.

[0042] Step S204, determining an alternative business scenario from a plurality of preset business scenarios according to the first feature set;

[0043] In some embodiments of the present application, at least one alternative business scenario can be obtained by processing the first feature set through a trained business scenario recognition model. Among them, the trained business scenario recognition model is obtained by establishing a telecommunications business scenario in advance, and identifying a corpus sample of each business scenario in each business scenario as a training set. Specifically, when performing multivariate feature extraction based on the corpus samples, the extracted features include lexical features, grammatical features, and semantic features, which helps the model recognize the current dialogue scenario. It should be noted that since there are many similarity branches between business scenarios, it is easy to cause confusing scenes, resulting in inaccurate positioning of dialogue scenes, so the model is prone to output multiple similar scenes.

[0044] When obtaining a training set, you can obtain training data through a pre-built corpus. Specifically, when building a corpus, you can collect conversations, texts, or corpora related to various business scenarios, such as from public data sets, forums on websites, social media, professional papers, or internal company data, and manually annotate the collected data to determine the useful information and features contained therein. When annotating, specific annotation tasks include entity recognition, sentiment analysis, topic classification, etc., so as to provide high-quality data support for specific application scenarios, thereby improving the effect and performance of the model.

[0045] As an optional implementation, the business scenario recognition model can also be used to recognize and understand different business scenarios. The vocabulary features, grammatical features and the semantic features are input into the business scenario recognition model. The business scenario recognition model will calculate the matching degree with different business scenarios based on these features. According to the matching degree output by the business scenario recognition, the business scenario with a matching degree greater than a preset matching degree is screened out as the identified business scenario. The preset matching degree is a threshold used to determine which business scenarios sufficiently match the input dialogue information.

[0046] In the technical solution provided in step S204, the step of determining alternative business scenarios from multiple preset business scenarios based on the first feature set includes: determining a business network topology model, wherein each node in the business network topology model corresponds to a business scenario, and different nodes correspond to different business scenarios, and the nodes are connected based on the relevance of the business logic between the business scenarios corresponding to the nodes; determining a matching degree index between the business scenarios corresponding to each node and the first business dialogue text based on the connection relationship between the first feature set and each node in the business network topology model; determining a business scenario whose matching degree index is greater than a preset threshold as an alternative business scenario.

[0047] In some embodiments of the present application, the business scenario recognition model can be constructed based on the business network topology model, wherein each node in the business network topology model represents each business scenario and its multivariate feature samples, including lexical feature samples, grammatical feature samples and semantic feature samples.

[0048] Specifically, when constructing a business scenario recognition model, each business scenario sample can be used as a node to construct a business network topology model. These nodes are connected based on the relevance of business logic to form a network structure. Identify the multivariate feature samples of each business scenario sample, including lexical feature samples, grammatical feature samples, and semantic feature samples. Specifically, for each business scenario sample, extract its lexical feature samples, including but not limited to keywords, word frequency, phrase patterns, etc.; analyze the grammatical structure of the business scenario sample, extract grammatical feature samples, such as sentence length, sentence type (declarative sentence, interrogative sentence, etc.), part-of-speech tagging, etc.; mine the deep semantic information of the business scenario sample, and extract semantic feature samples. Encode the extracted lexical feature samples, grammatical feature samples, and semantic feature samples so as to associate them with the nodes in the business network topology model. Match and associate the encoded feature samples with the nodes in the business network topology model, and add corresponding feature identifiers to each node. In this way, each node not only represents a business scenario sample, but also contains the multivariate feature information of the scenario. The business network topology model containing multiple feature identifiers is integrated into a complete business scenario recognition model. The business scenario recognition model can identify the corresponding business scenario based on the input text or data by matching and analyzing feature samples.

[0049] Step S206, when there are multiple candidate business scenarios, respectively determine target distinguishing feature names of the multiple candidate business scenarios, wherein the target distinguishing feature names are used to distinguish the candidate business scenarios;

[0050] In the technical solution provided in step S206, the step of determining the target distinguishing feature names of multiple alternative business scenarios includes: determining sample feature elements of each alternative business scenario, wherein the sample feature elements of each alternative business scenario include a feature name and a feature value corresponding to the feature name, wherein the feature value corresponding to the feature name is used to indicate the number of occurrences of the feature corresponding to the feature name in the alternative business scenario; for each feature name, determining the value dispersion coefficient of the feature value of each alternative business scenario, wherein the value dispersion coefficient is used to reflect the degree of dispersion of the feature value of each alternative business scenario, and the larger the value dispersion coefficient, the greater the degree of dispersion; determining the feature name with the largest value dispersion coefficient as the target distinguishing feature name.

[0051] As an optional implementation, for each feature name, determining the value dispersion coefficient of the feature value of each candidate business scenario includes: for each feature name, determining the variance between the feature values ​​of each candidate business scenario, and determining the variance as the value dispersion coefficient.

[0052] Specifically, when calculating the dispersion coefficient, you can use the following method: Figure 3 The calculation process shown includes the following steps:

[0053] Step S302, encoding each candidate service scenario;

[0054] Specifically, a unique code may be assigned to each candidate business scenario, and the unique code may be a number, a string, or other identifier.

[0055] Step S304, establishing a rectangular coordinate system with the encoding results of each candidate service scenario as the horizontal axis and the feature value as the vertical axis;

[0056] When constructing the coordinate system, the coordinate value orientations on the horizontal and vertical axes of the coordinate system may also be set according to the actual number of business scenarios and the value range of the characteristic quantity value.

[0057] Step S306, using a rectangular coordinate system to input data of the feature metadata of each candidate business scenario, and obtaining the feature value distribution corresponding to each feature metadata;

[0058] Specifically, by entering the feature metadata into a rectangular coordinate system, the feature metadata can be represented as a point on the rectangular coordinate system, and then the feature value distribution of each feature element in all identified business scenarios can be obtained. For each feature element, the variance of its feature value in all business scenarios is calculated. The calculated variance is output as the value dispersion coefficient of each feature element. The larger the value dispersion coefficient, the greater the difference between the feature element in different business scenarios, and the stronger the ability to distinguish business scenarios.

[0059] Step S308: performing variance calculation on the distribution of the feature values ​​corresponding to each feature name, and using the variance calculation result as the value dispersion coefficient corresponding to each feature name.

[0060] Step S208, determining the target feature value corresponding to the first business conversation text and the target distinguishing feature name, and determining a first target business scenario matching the first business conversation text from a plurality of candidate business scenarios according to the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text;

[0061] In the technical solution provided in step S208, determining the first target business scenario that matches the first business dialogue text from multiple alternative business scenarios based on the target feature value includes: determining the sample feature value of each alternative business scenario relative to the target distinguishing feature name; in the case where there is a sample feature value equal to the target feature value, determining the alternative business scenario whose corresponding sample feature value is equal to the target feature value as the first target business scenario; in the case where there is no sample feature value equal to the target feature value, determining the absolute value of the difference between the sample feature value of each alternative business scenario and the target feature value, and determining the alternative business scenario whose corresponding difference absolute value is the smallest as the first target business scenario.

[0062] Specifically, when there are multiple alternative business scenarios, their lexical features, grammatical features and semantic features can be extracted again, and the features of different business scenarios can be compared to find those that show significant differences in all scenarios. From the features with significant differences, select one or more features with the highest representativeness and highest discrimination as the first distinguishing feature element, and determine the target distinguishing feature name corresponding to the first distinguishing feature element.

[0063] Then, the corresponding features can be extracted from the first business dialogue text according to the target distinguishing feature name, and the extracted features can be converted into specific values, namely, the target feature values, by calculating the word frequency and counting the number of occurrences of the specific grammatical structure. The feature values ​​are matched in all the identified business scenarios according to the first feature value, and the matching business scenarios are output, that is, the matching business scenarios whose corresponding sample feature values ​​are equal to the target feature values ​​are found.

[0064] If there is no business scenario whose characteristic value is equal to the target characteristic value, a difference operation can be performed between the sample characteristic values ​​and the target characteristic value of all candidate business scenarios, and the candidate business scenario with the smallest absolute value of the difference is determined as the first matching target business scenario.

[0065] Step S210: Generate a business response text corresponding to the first business dialogue text according to the first target business scenario.

[0066] In the technical solution provided in step S210, a business reply text can be generated through a corpus corresponding to the first target business scenario. Specifically, the corpus contains common dialogue texts, reply templates, keywords, etc. in the business scenario. By searching the corpus corresponding to the matching business scenario, a suitable reply template or predefined reply text can be selected according to the keywords or intentions of the first business dialogue information to obtain the business reply text.

[0067] As an optional implementation, the business processing method also includes: after receiving the second business conversation text, performing feature extraction on the second business conversation text to obtain a second feature set of the second business conversation text; determining a matching degree index between the second business conversation text and the first target business scenario based on the second feature set; when the matching degree index is lower than a preset threshold, determining a second target business scenario that matches the second business conversation text, and generating a business reply text corresponding to the second business conversation text based on the second target business scenario.

[0068] Specifically, when processing continuous business dialogues, especially when the business scenarios involved in the second round of dialogue (i.e., the second business dialogue text) and the first round of dialogue (i.e., the first business dialogue text) may change, it is necessary to re-evaluate the business scenarios based on the second round of dialogue information. Specifically, a second business dialogue text related to the first business dialogue text is obtained, and features related to the business scenario are extracted from the second business dialogue text. These features may be different from the features of the first business dialogue text because they may reflect the progress of the dialogue or changes in user intentions. Using these newly extracted features, the matching degree between the second business dialogue text and the preferred business scenario is calculated. The calculated matching degree is compared with a preset matching degree threshold. If the matching degree is less than the preset threshold, it means that the current preferred business scenario does not match the second business dialogue text very well. If the scenario matching degree is less than the preset matching degree, a return instruction is obtained, and the return instruction indicates that the system needs to re-evaluate all identified business scenarios to find a more suitable business scenario. According to the return instruction, all identified business scenarios are re-evaluated with the preset matching degree as the target, and the matching degree of each business scenario with the second business dialogue text is calculated, and the business scenario with the highest matching degree is selected as the new preferred business scenario.

[0069] In some embodiments of the present application, there is also provided a Figure 4 The reply text generation process shown includes the following steps:

[0070] Step S402, connecting to the NLP natural language processing system to obtain the first business dialogue text;

[0071] Step S404, extracting features from the first business dialogue text to obtain multi-dimensional features of the first business dialogue text, wherein the multi-dimensional features include lexical features, grammatical features, and semantic features;

[0072] Step S406, processing the multivariate features through the business scenario recognition model to obtain an alternative business scenario whose matching degree with the first business dialogue text is greater than a preset matching degree;

[0073] Step S408, when there are multiple candidate business scenarios, determining first distinguishing feature elements corresponding to the multiple candidate business scenarios, wherein the first distinguishing feature element is used to indicate feature names that can distinguish the multiple candidate business scenarios;

[0074] Step S410, determining a matching business scenario that matches the first business dialogue text from a plurality of candidate business scenarios according to the first distinguishing feature element;

[0075] Step S412: Generate a reply text corresponding to the first business dialogue text according to the corpus matching the business scenario.

[0076] By extracting features from a first business conversation text, a first feature set of the first business conversation text is obtained; an alternative business scenario is determined from a plurality of preset business scenarios according to the first feature set; when there are a plurality of alternative business scenarios, target distinguishing feature names of the plurality of alternative business scenarios are respectively determined, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; a target feature value corresponding to the first business conversation text and the target distinguishing feature name is determined, and a first target business scenario matching the first business conversation text is determined from the plurality of alternative business scenarios according to the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text; and a first target business scenario matching the first business conversation text is determined according to the first target business conversation text. The method of generating a business reply text corresponding to the first business dialogue text by scenario, by determining the target distinguishing feature name of the target distinguishing feature that can distinguish multiple candidate business scenarios, and determining the target feature value of the first business dialogue text based on the target distinguishing feature name, and then determining the first target business scenario matching the first business dialogue text according to the target feature value, achieves the purpose of accurately determining the business scenario matching the first business dialogue text from multiple candidate business scenarios, thereby achieving the technical effect of accurately generating a business reply text matching the first business dialogue text, and then solves the technical problem that the business scenario corresponding to the user input text cannot be accurately determined in the related art, resulting in the inability to accurately generate a reply text matching the user input text. It also solves the technical problem that there are many similar business scenarios in the communication dialogue scenario in the related art, resulting in difficulty in identifying the language scenario, and achieves the technical effect of accurately locating the dialogue scenario.

[0077] The present application provides a service processing device. Figure 5 is a schematic diagram of the structure of the device. Figure 5As can be seen from the figure, the device includes: a first processing module 50, which is used to extract features from the first business dialogue text to obtain a first feature set of the first business dialogue text; a second processing module 52, which is used to determine an alternative business scenario from a plurality of preset business scenarios based on the first feature set; a third processing module 54, which is used to determine target distinguishing feature names of a plurality of alternative business scenarios respectively when the number of alternative business scenarios is multiple, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; a fourth processing module 56, which is used to determine the target feature value corresponding to the target distinguishing feature name of the first business dialogue text, and determine the first target business scenario matching the first business dialogue text from a plurality of alternative business scenarios based on the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business dialogue text; a fifth processing module 58, which is used to generate a business reply text corresponding to the first business dialogue text based on the first target business scenario.

[0078] In some embodiments of the present application, the feature type of the first feature set includes at least one of the following: lexical feature, grammatical feature and semantic feature.

[0079] In some embodiments of the present application, the step of the second processing module 52 determining alternative business scenarios from multiple preset business scenarios based on the first feature set includes: determining a business network topology model, wherein each node in the business network topology model corresponds to a business scenario, and different nodes correspond to different business scenarios, and the nodes are connected based on the business logic association between the business scenarios corresponding to the nodes; determining a matching degree index between the business scenarios corresponding to each node and the first business dialogue text based on the connection relationship between the first feature set and each node in the business network topology model; and determining a business scenario whose matching degree index is greater than a preset threshold as an alternative business scenario.

[0080] In some embodiments of the present application, the step in which the third processing module 54 determines the target distinguishing feature names of multiple alternative business scenarios includes: determining sample feature elements of each alternative business scenario, wherein the sample feature elements of each alternative business scenario include a feature name and a feature value corresponding to the feature name, wherein the feature value corresponding to the feature name is used to indicate the number of occurrences of the feature corresponding to the feature name in the alternative business scenario; for each feature name, determining the value dispersion coefficient of the feature value of each alternative business scenario, wherein the value dispersion coefficient is used to reflect the degree of dispersion of the feature value of each alternative business scenario, and the larger the value dispersion coefficient, the greater the degree of dispersion; determining the feature name with the largest value dispersion coefficient as the target distinguishing feature name.

[0081] In some embodiments of the present application, the step in which the third processing module 54 determines the value dispersion coefficient of the feature value of each alternative business scenario for each feature name includes: for each feature name, determining the variance between the feature values ​​of each alternative business scenario, and determining the variance as the value dispersion coefficient.

[0082] In some embodiments of the present application, the fourth processing module 56 determines the first target business scenario that matches the first business dialogue text from multiple alternative business scenarios based on the target feature value, including: determining the sample feature value of each alternative business scenario relative to the target distinguishing feature name; in the case where there is a sample feature value equal to the target feature value, determining the alternative business scenario whose corresponding sample feature value is equal to the target feature value as the first target business scenario; in the case where there is no sample feature value equal to the target feature value, determining the absolute value of the difference between the sample feature value of each alternative business scenario and the target feature value, and determining the alternative business scenario whose corresponding difference absolute value is the smallest as the first target business scenario.

[0083] In some embodiments of the present application, the business processing device is also used to: after receiving the second business conversation text, perform feature extraction on the second business conversation text to obtain a second feature set of the second business conversation text; determine a matching degree index between the second business conversation text and the first target business scenario based on the second feature set; when the matching degree index is lower than a preset threshold, determine a second target business scenario that matches the second business conversation text, and generate a business reply text corresponding to the second business conversation text based on the second target business scenario.

[0084] It should be noted that the various modules in the above-mentioned business processing device can be program modules (for example, a set of program instructions that implement a certain specific function) or hardware modules. For the latter, it can be expressed in the following forms, but is not limited to this: the expression form of each of the above-mentioned modules is a processor, or the functions of each of the above-mentioned modules are implemented by a processor.

[0085] According to another aspect of an embodiment of the present application, a non-volatile storage medium is also provided, in which a program is stored, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the following business processing method: feature extraction is performed on a first business conversation text to obtain a first feature set of the first business conversation text; alternative business scenarios are determined from multiple preset business scenarios based on the first feature set; when there are multiple alternative business scenarios, target distinguishing feature names of the multiple alternative business scenarios are determined respectively, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; target feature values ​​corresponding to the first business conversation text and the target distinguishing feature names are determined, and a first target business scenario matching the first business conversation text is determined from multiple alternative business scenarios based on the target feature values, wherein the target feature values ​​are used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text; and a business reply text corresponding to the first business conversation text is generated based on the first target business scenario.

[0086] According to another aspect of an embodiment of the present application, an electronic device is also provided, including a memory and a processor, the processor being used to run a program stored in the memory, wherein the following business processing method is executed when the program is running: performing feature extraction on a first business conversation text to obtain a first feature set of the first business conversation text; determining an alternative business scenario from a plurality of preset business scenarios based on the first feature set; in the case where there are multiple alternative business scenarios, respectively determining target distinguishing feature names of the plurality of alternative business scenarios, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; determining a target feature value corresponding to the first business conversation text and the target distinguishing feature name, and determining a first target business scenario matching the first business conversation text from a plurality of alternative business scenarios based on the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text; and generating a business reply text corresponding to the first business conversation text based on the first target business scenario.

[0087] According to another aspect of an embodiment of the present application, a computer program product is also provided, including a computer program, which implements the following business processing method when executed by a processor: extracting features from a first business conversation text to obtain a first feature set of the first business conversation text; determining alternative business scenarios from multiple preset business scenarios based on the first feature set; in the case where there are multiple alternative business scenarios, respectively determining target distinguishing feature names of the multiple alternative business scenarios, wherein the target distinguishing feature names are used to distinguish the alternative business scenarios; determining target feature values ​​corresponding to the first business conversation text and the target distinguishing feature names, and determining a first target business scenario matching the first business conversation text from multiple alternative business scenarios based on the target feature values, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business conversation text; and generating a business reply text corresponding to the first business conversation text based on the first target business scenario.

[0088] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0089] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0090] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0091] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0092] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the relevant technology or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0093] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A business processing method, characterized in that: include: Extracting features from the first business conversation text to obtain a first feature set of the first business conversation text; Determining an alternative business scenario from a plurality of preset business scenarios according to the first feature set; In the case that there are multiple candidate business scenarios, respectively determine target distinguishing feature names of the multiple candidate business scenarios, wherein the target distinguishing feature names are used to distinguish the candidate business scenarios; Determine the target feature value corresponding to the first business dialogue text and the target distinguishing feature name, and determine a first target business scenario matching the first business dialogue text from the multiple candidate business scenarios according to the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business dialogue text; A business response text corresponding to the first business dialogue text is generated according to the first target business scenario.

2. The service processing method according to claim 1, characterized in that: Determining target distinguishing feature names of the plurality of candidate business scenarios includes: Determine a sample feature element of each candidate business scenario, wherein the sample feature element of each candidate business scenario includes a feature name and a feature value corresponding to the feature name, wherein the feature value corresponding to the feature name is used to indicate the number of occurrences of the feature corresponding to the feature name in the candidate business scenario; For each of the feature names, determining a value dispersion coefficient of the feature value of each of the candidate business scenarios, wherein the value dispersion coefficient is used to reflect the degree of dispersion of the feature value of each of the candidate business scenarios, and the greater the value dispersion coefficient, the greater the degree of dispersion; The feature name with the largest value dispersion coefficient is determined as the target distinguishing feature name.

3. The service processing method according to claim 2, characterized in that: For each of the feature names, determining the value dispersion coefficient of the feature value of each of the candidate business scenarios includes: For each of the feature names, the variance between the feature values ​​of the candidate business scenarios is determined, and the variance is determined as the value dispersion coefficient.

4. The service processing method according to claim 1, characterized in that: Determining a first target business scenario matching the first business dialogue text from a plurality of the candidate business scenarios according to the target feature value includes: Determining a sample feature value of each of the candidate business scenarios relative to the target distinguishing feature name; In the case where there is the sample feature value equal to the target feature value, determining the candidate business scenario whose corresponding sample feature value is equal to the target feature value as the first target business scenario; In the absence of the sample feature value equal to the target feature value, the absolute value of the difference between the sample feature value and the target feature value of each alternative business scenario is determined, and the alternative business scenario with the smallest corresponding absolute value of the difference is determined as the first target business scenario.

5. The service processing method according to claim 1, characterized in that: Determining the candidate business scenario from a plurality of preset business scenarios according to the first feature set includes: Determine a business network topology model, wherein each node in the business network topology model corresponds to a business scenario, and different nodes correspond to different business scenarios, and the nodes are connected based on the association of business logics between the business scenarios corresponding to the nodes; Determining a matching degree index between the business scenario corresponding to each of the nodes and the first business dialogue text according to the first feature set and the connection relationship between each of the nodes in the business network topology model; Determine the business scenario whose matching degree index is greater than a preset threshold as the candidate business scenario.

6. The service processing method according to claim 1, characterized in that: The business processing method further includes: After receiving the second business conversation text, performing feature extraction on the second business conversation text to obtain a second feature set of the second business conversation text; Determining a matching degree index between the second business dialogue text and the first target business scenario according to the second feature set; When the matching degree index is lower than a preset threshold, a second target business scenario matching the second business dialogue text is determined, and a business reply text corresponding to the second business dialogue text is generated according to the second target business scenario.

7. The service processing method according to claim 1, characterized in that: The feature type of the first feature set includes at least one of the following: lexical feature, grammatical feature and semantic feature.

8. A service processing device, characterized in that: include: A first processing module, configured to extract features from the first business conversation text to obtain a first feature set of the first business conversation text; A second processing module, configured to determine an alternative business scenario from a plurality of preset business scenarios according to the first feature set; A third processing module is used to determine target distinguishing feature names of the multiple candidate business scenarios respectively when there are multiple candidate business scenarios, wherein the target distinguishing feature names are used to distinguish the candidate business scenarios; a fourth processing module, configured to determine a target feature value corresponding to the target distinguishing feature name and the first business dialogue text, and determine a first target business scenario matching the first business dialogue text from the multiple candidate business scenarios according to the target feature value, wherein the target feature value is used to indicate the number of occurrences of the feature indicated by the target distinguishing feature name in the first business dialogue text; A fifth processing module is used to generate a business response text corresponding to the first business dialogue text according to the first target business scenario.

9. A non-volatile storage medium, characterized in that: The non-volatile storage medium stores a program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the service processing method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program executes the business processing method described in any one of claims 1 to 7 when running.

11. A computer program product, characterized in that The invention comprises a computer program, which implements the service processing method according to any one of claims 1 to 7 when being executed by a processor.

Images