CN115827833A

CN115827833A - Dialog structure processing method and device, storage medium and electronic equipment

Info

Publication number: CN115827833A
Application number: CN202211347795.4A
Authority: CN
Inventors: 李凤; 屈瑞麟
Original assignee: Chongqing Ant Consumer Finance Co ltd
Current assignee: Chongqing Ant Consumer Finance Co ltd
Priority date: 2022-10-31
Filing date: 2022-10-31
Publication date: 2023-03-21

Abstract

The specification discloses a dialog structure processing method, a dialog structure processing device, a storage medium and an electronic device, wherein the method comprises the following steps: the method comprises the steps of determining sentence intentions of conversation sentences in a plurality of conversation data, determining at least one conversation frequent item set consisting of a plurality of conversation reference items corresponding to the plurality of conversation data based on the sentence intentions of the conversation sentences, and then obtaining a conversation structure sequence corresponding to each reference conversation item in the conversation frequent item set to combine the conversation frequent item set so as to determine at least one conversation structure aiming at the plurality of conversation data.

Description

Dialog structure processing method and device, storage medium and electronic equipment

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a method and an apparatus for processing a dialog structure, a storage medium, and an electronic device.

Background

With the development of computer technology, electronic devices are rapidly popularized, and various application programs and web page end programs for providing life convenience services are also diversified, so as to provide services (such as travel services, take-out services, consumption financial services and the like) for the eating and wearing of users. In the process of using the services, users often involve multiple rounds of conversations, single rounds of conversations, customer service robots and the like in a customer service system to ask questions to be solved or communicate with customer services for service matters to be solved.

Disclosure of Invention

The specification provides a dialog structure processing method, a dialog structure processing device, a storage medium and an electronic device, and the technical scheme is as follows:

in a first aspect, the present specification provides a dialog structure processing method, including:

acquiring a plurality of dialogue data, wherein the dialogue data comprises at least one dialogue statement;

determining a sentence intention of the dialogue sentences in each dialogue data, and determining at least one dialogue frequent item set corresponding to the plurality of dialogue data based on the sentence intention of the dialogue sentences, wherein the dialogue frequent item set comprises a plurality of dialogue reference items;

and acquiring a dialog structure sequence corresponding to each reference dialog item in each dialog frequent item set, and determining at least one dialog structure aiming at the plurality of dialog data based on the dialog structure sequence and each reference dialog item in the dialog frequent item set.

In a second aspect, the present specification provides a dialog structure processing apparatus, the apparatus comprising:

the system comprises an intention determining module, a judging module and a judging module, wherein the intention determining module is used for acquiring a plurality of dialogue data, the dialogue data comprises at least one dialogue statement and determining the sentence intention of the dialogue statement in each dialogue data;

a term set determination module, configured to determine at least one conversation frequent item set corresponding to the plurality of conversation data based on the sentence intent of the conversation sentence, where the conversation frequent item set includes a plurality of conversation reference items;

and the structure determining module is used for acquiring a conversation structure sequence corresponding to each reference conversation item in each conversation frequent item set, and determining at least one conversation structure aiming at the plurality of conversation data based on the conversation structure sequence and each reference conversation item in the conversation frequent item set.

In a third aspect, the present specification provides a computer storage medium having stored thereon a plurality of instructions adapted to be loaded by a processor and to carry out the above-mentioned method steps.

In a fourth aspect, the present specification provides an electronic device, which may comprise: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the above-mentioned method steps.

The technical scheme provided by some embodiments of the present description brings beneficial effects at least including:

in one or more embodiments of the present specification, the electronic device determines at least one dialog structure for the plurality of dialog data by determining a sentence intention of a dialog sentence in the plurality of dialog data, determining at least one dialog frequent item set composed of a plurality of dialog reference items corresponding to the plurality of dialog data based on the sentence intention of the dialog sentence, and then obtaining a dialog structure order corresponding to each reference dialog item in the dialog frequent item set to combine with the dialog frequent item set. The method has the advantages that the orderly conversation structure can be obtained by mining the disordered conversation frequent item set based on the sentence intention of the conversation sentence and then determining the conversation structure sequence of the reference conversation item, the integral conversation characteristic under the integral conversation scene can be sensed through the orderly conversation structure, the conversation limitation caused by focusing local single-round conversation reply is avoided, the intelligence of conversation processing is improved, and the conversation quality and the accuracy of conversation expression can be greatly improved when the plurality of conversation structures based on the integral conversation characteristic are applied to the local single-round conversation scene.

Drawings

In order to more clearly illustrate the technical solutions in the present specification or prior art, the drawings used in the embodiments or prior art descriptions will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present specification, and other drawings can be obtained by those skilled in the art without creative efforts.

FIG. 1 is a schematic diagram of a dialog structure processing system in a scenario provided herein;

FIG. 2 is a flow chart diagram of a dialog structure processing method provided in the present specification;

FIG. 3 is a flow diagram illustrating another dialog structure processing method provided in the present specification;

FIG. 4 is a flow chart diagram of another dialog structure processing method provided in the present specification;

fig. 5 is a schematic structural diagram of a dialog structure processing apparatus provided in the present specification;

FIG. 6 is a block diagram of an item set determination module provided in the present specification;

fig. 7 is a schematic structural diagram of an item set determination unit provided in the present specification;

FIG. 8 is a schematic diagram of a structure determining module provided in the present specification;

fig. 9 is a schematic structural diagram of an electronic device provided in this specification;

FIG. 10 is a schematic diagram of the operating system and user space provided in this specification;

FIG. 11 is an architectural diagram of the android operating system of FIG. 10;

FIG. 12 is an architecture diagram of the IOS operating system of FIG. 10.

Detailed Description

The technical solutions in the present specification will be clearly and completely described below with reference to the drawings in the present specification, and it is obvious that the described embodiments are only a part of the embodiments in the present specification, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present specification without any creative effort belong to the protection scope of the present specification.

In the description of the present specification, it is to be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In the description of the present specification, it is to be noted that, unless explicitly stated or limited otherwise, "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus. The specific meanings of the above terms in the present specification can be understood in specific cases by those of ordinary skill in the art. Further, in the description of the present specification, "a plurality" means two or more unless otherwise specified. "and/or" describes the association relationship of the associated objects, meaning that there may be three relationships, e.g., a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship.

In the related art, in order to better provide a dialog interaction experience for a user or customer service, dialog structure processing in a common dialog corpus is often involved, for example, in a dialog interaction scenario, attention is often paid to how to quickly assist in determining a candidate dialog statement as a current dialog to reply, that is, to quickly reply to a current query statement of the user, while in a dialog scenario, multiple rounds of dialog are often involved with a high probability, and a great limitation must be imposed on a reply focused on a single round of local dialog in the dialog scenario, so that an overall dialog form in an actual application scenario cannot be perceived, and particularly, an abnormal dialog situation in some service transactions (such as consumption financial transactions) is difficult to perceive.

The present specification will be described in detail with reference to specific examples.

Please refer to fig. 1, which is a schematic view of a dialog structure processing system according to the present disclosure. As shown in FIG. 1, the dialog structure processing system may include at least a customer service cluster and a service platform 100.

The client cluster may include at least one client, an associated object of the client may be a client or a service end, and the service end is associated with the service platform 100 for improving the transaction service, as shown in fig. 1, specifically includes a client 1 corresponding to an associated object 1, a client 2 corresponding to an associated object 2, …, and a client n corresponding to an associated object n, where n is an integer greater than 0.

Each client in the client cluster may be a communication-enabled electronic device including, but not limited to: wearable devices, handheld devices, personal computers, tablet computers, in-vehicle devices, smart phones, computing devices or other processing devices connected to a wireless modem, and the like. Electronic devices in different networks may be called different names, such as: user equipment, access terminal, subscriber unit, subscriber station, mobile station, remote terminal, mobile device, user terminal, wireless communication device, user agent or user equipment, cellular telephone, cordless telephone, personal Digital Assistant (PDA), electronic device in a 5G network or future evolution network, and the like.

The service platform 100 may be a separate server device, such as: rack, blade, tower or cabinet type server equipment, or hardware equipment with stronger computing power such as a workstation and a large computer; the server cluster may also be a server cluster composed of a plurality of servers, each server in the service cluster may be composed in a symmetric manner, where each server is functionally equivalent and functionally equivalent in the transaction link, and each server may provide services to the outside independently, where the independent provision of services may be understood as no assistance from another server.

In one or more embodiments of the present description, the service platform 100 may establish a communication connection with at least one client in the client cluster, and complete data interaction in the dialog structure processing process based on the communication connection, for example, the service platform 100 may collect a plurality of dialog data from the clients based on the dialog structure processing method of the present description, where each dialog data is a plurality of rounds of dialog statements corresponding to the clients; for another example, the service platform 100 may perform the dialog structure processing method of the present specification to obtain several dialog structures, and then configure the dialog structures to the servers in the client cluster. For another example, a user side in the client cluster may initiate a dialog window to the service platform 100 for performing a dialog, the service platform 100 may allocate a customer service side to the user side, the customer service side may perform a dialog reply process based on a plurality of dialog structures of the service platform 100 after receiving a dialog statement of the user side, and so on.

It should be noted that the service platform 100 establishes a communication connection with at least one client in the client cluster to perform interactive communication through a network, where the network may be a wireless network including but not limited to a cellular network, a wireless local area network, an infrared network, or a bluetooth network, and the wired network includes but not limited to an ethernet network, a Universal Serial Bus (USB), or a controller area network. In one or more embodiments of the specification, data (e.g., target compressed packets) exchanged over a network is represented using techniques and/or formats including Hyper Text Markup Language (HTML), extensible Markup Language (XML), and the like. All or some of the links may also be encrypted using conventional encryption techniques such as Secure Socket Layer (SSL), transport Layer Security (TLS), virtual Private Network (VPN), internet Protocol Security (IPsec). In other embodiments, custom and/or dedicated data communication techniques may also be used in place of, or in addition to, the data communication techniques described above.

The dialog structure processing system embodiment provided in this specification and the dialog structure processing method in one or more embodiments belong to the same concept, and an execution subject corresponding to the dialog structure processing method in one or more embodiments in this specification is an electronic device, which may be the service platform 100 described above; the execution subject corresponding to the dialog structure processing method referred to in one or more embodiments of the specification may also be a client, specifically determined based on an actual application environment. The embodiment of the dialog structure processing system, which embodies the implementation process, can be referred to in the following method embodiments, and is not described herein again.

Based on the scene diagram shown in fig. 1, the following describes in detail a dialog structure processing method provided in one or more embodiments of the present specification.

Referring to fig. 2, a flow diagram of a processing method of a dialog structure, which can be implemented by a computer program and can run on a processing device of a dialog structure based on von neumann architecture, is provided for one or more embodiments of the present specification. The computer program may be integrated into the application or may run as a separate tool-like application. The dialog structure processing means may be an electronic device.

Specifically, the dialog structure processing method includes:

s102: acquiring a plurality of dialogue data, wherein the dialogue data comprises at least one dialogue statement;

the dialogue data may be dialogue data in a collaborative dialogue interaction scenario, for example, several rounds of dialogue statements in a business consultation scenario on a service platform, for example, multiple rounds of dialogue statements in a query dialogue scenario issued by a user to a customer service based on details of a commodity, where the dialogue data generally includes at least one round of dialogue statement.

It can be understood that multiple rounds of historical conversations in multiple conversation scenarios can be obtained as the conversation data, for example, multiple rounds of conversation sentences of the user side and the server side can be saved as the conversation data each time the user side and the server side perform at least one round of conversation in the multiple conversation scenarios (e.g., a trip service conversation scenario, a take-out service, online shopping, a cash-out service conversation scenario, etc.).

In one or more embodiments of the present specification, after a plurality of pieces of dialog data are obtained, dialog denoising processing may be performed on the dialog data first, so as to filter out non-key contents in the dialog and retain key dialog contents, for example, non-key contents such as call dialog information, automatic response dialog information, and toast information in the dialog, and the dialog data after denoising only retains statements in a role inquiry dialog form or a role reply dialog form.

S104: determining a sentence intention of the dialogue sentences in each dialogue data, and determining at least one dialogue frequent item set corresponding to the plurality of dialogue data based on the sentence intention of the dialogue sentences, wherein the dialogue frequent item set comprises a plurality of dialogue reference items;

in one or more embodiments of the present specification, the obtained multiple sets of dialogue data may be regarded as samples for extracting a dialogue structure in a corresponding dialogue scene, after the multiple sets of dialogue data are obtained, sentence intent marking needs to be performed on multiple rounds or multiple sets of dialogue sentences contained in the dialogue data, that is, sentence intentions of all or part of the dialogue sentences in the dialogue data are determined, and then, extraction of the dialogue structure in the dialogue scene needs to be realized based on the sentence intentions of the multiple rounds of dialogue sentences in the dialogue data.

In a possible implementation manner, sentence feature extraction is performed through each dialogue statement in the dialogue data, the sentence feature extraction process may be to perform dialogue slot completion, dialogue statement word segmentation, noise word filtering (e.g., stop word filtering), vector representation, and other processes in combination with a dialogue data context dialogue to obtain sentence features of the dialogue statement, where the word vector representation may employ a text vector extraction model such as DOC2VEC to extract sentence feature vectors, the dialogue statement may be mapped into high-dimensional feature space vectors in the form of feature vectors to be characterized, then, a dialogue sentence clustering process is performed on the sentence feature vectors to obtain a clustering sentence center, the clustering sentence center may feed back sentence intentions, a key intention is formed by extracting keywords with respect to a dialogue statement intention category corresponding to the clustering sentence center with TF-IDF to serve as an intention of a user, a set of sentence intentions of each dialogue statement in the plurality of data may form a user intention library, and then, and a labeling association between a plurality of dialogue statements in the dialogue data and their corresponding sentence intentions may be achieved based on regular intention labeling and/or manual labeling.

In a possible implementation, in the early stage of intent recognition, each intent is often only a few or dozen samples, considering that manually marking dialog sentences is time consuming and laborious and often does not have a large amount of marking data. In the face of such cold start problems, sentence intent can be identified using small sample Learning (Few-Shot Learning).

Illustratively, the intention semantic representation of the dialog sentence (the semantic representation may be in the form of a dialog semantic intention vector) may be performed on each dialog sentence in the dialog data by using an intention recognition model constructed based on a machine learning model in conjunction with the dialog data, for example, the intention semantic representation of each dialog sentence in the process may be recognized by using a machine learning model such as Bi-LSTM, BERT, etc. as input by using the dialog data;

illustratively, intent category features are then generalized from the sample semantic intent representation in the support set based on an intent recognition model;

illustratively, the intention semantic relationship between each dialog statement query and the intention category is measured based on the intention recognition model, and the intention classification result of the dialog statement is obtained.

Further, all the dialogue data can be used as a training set for small sample learning, and the training set may contain a large number of intention categories, and there are usually only a small number of dialogue samples in each intention category with explicit sentence intention labels. In the training stage, C intention categories are randomly extracted from a training set, K statement samples (C × K data in total) of each intention category construct a support set of an intention recognition model for recognizing the intention of a sentence, a batch of sentence samples are extracted from the C intention categories to serve as prediction objects of the intention recognition model, and in the small sample learning stage: to indicate how the intent recognition model learns from the C K data to distinguish the C intent classes.

In one possible implementation, semantic recognition of intent may be performed on several conversational sentences based on the conversational data, so as to obtain the sentence intent of the conversational sentence.

Illustratively, dialog data can be used as an intention semantic recognition object, intent semantic recognition for each dialog statement refers to a dialog context in addition to the dialog statement in the specification to extract slot information which is not related in the dialog statement but is expressed by a user, a semantic intent recognition model can be built based on a machine learning model, the dialog data is used as data input by adopting the semantic intent recognition model, the dialog data comprises a plurality of rounds of dialog statements, the dialog statements and the context information of the dialog statements in the dialog data are recognized based on the semantic intent recognition model to obtain accurate dialog semantic features of each dialog statement, the dialog semantic features fuse the semantics of the dialog statements and the context semantics of the dialog statements in the dialog data, and the semantic intent recognition model then performs intent recognition based on the dialog semantic features of each dialog statement to output a sentence intention corresponding to each dialog statement.

The dialogue semantic features are dialogue semantic attributes specific to unstructured data expressed in natural language, and comprise semantic elements such as dialogue intents, dialogue theme descriptions, underlying feature meanings and context semantics.

Alternatively, the semantic intention recognition model may be implemented based on fitting of one or more of a Convolutional Neural Network (CNN) model, a Deep Neural Network (DNN) model, a Recurrent Neural Network (RNN) model, a pre-training language model (BERT), an embedding (embedding) model, a Gradient Boosting Decision Tree (GBDT) model, a Logistic Regression (LR) model, and the like.

Schematically, a large amount of dialog text sample data can be obtained in advance, the semantic intent recognition model is trained by using the dialog text sample data, and the trained semantic intent recognition model can be obtained after the training is finished.

Further, after determining the sentence intent of the dialogue statement in each dialogue data, the electronic device may then determine at least one dialogue frequent item set corresponding to the plurality of dialogue data based on the sentence intent of the dialogue statement;

the conversation frequent item set feeds back a set which is characterized in a conversation reference item form, such as a conversation structure item, a conversation structure sequence or a sub-conversation structure and the like which frequently appear in a practical application transaction scene, the conversation frequent item set comprises a plurality of conversation reference items, and the plurality of conversation reference items in the conversation frequent item set do not usually contain conversation structure sequence/time sequence information.

The dialogue reference item can be understood as a reference dialogue statement in a dialogue scene, the dialogue reference item can be characterized by a sentence intention of the dialogue statement, and the understandable sentence intention of the same type can correspond to the dialogue statements of different forms; in some embodiments, the conversational reference term may also be a plurality of conversational sentences within the same class of sentence intent. Each of the dialog reference items may feed back a single-round dialog form of the data dialog mining, and the dialog reference item may be a character inquiry dialog form or a character reply dialog form among the scene dialog forms.

It is understood that after determining the dialog order of each dialog reference item based on the set of dialog frequentness items, the dialog structure under the dialog scenario in which these multiple pieces of dialog data are located can be obtained.

In one possible implementation, a dialogue frequent pattern tree may be constructed based on sentence intentions of each dialogue statement in a plurality of dialogue data, and at least one dialogue frequent item set may be mined from the constructed dialogue frequent pattern tree based on the sentence intentions;

illustratively, the conversation frequent item set comprises a plurality of reference conversation items, and disorder exists among the reference conversation items in the conversation frequent item set, and the plurality of reference conversation items in each conversation frequent item set can be understood as a multi-turn multi-conversation form which often appears in a conversation scene.

Illustratively, each tree node in the frequent dialog mode tree is not a dialog statement per se but a sentence intention of the dialog statement, and the sentence intention is used as a data mining object, so that the association rule relationship of the dialog in the dialog scene can be quickly and efficiently mined from the constructed frequent dialog mode tree, and at least one frequent dialog item set can be obtained.

Illustratively, the support degree of each sentence intention, which indicates the number of times all the dialog sentences of the sentence intention type appear in all the dialog data, may be first counted based on the sentence intention of each dialog sentence in the plurality of dialog data.

And then, sequentially inserting each sentence intention in the plurality of dialogue data into a tree with NULL as a root node according to a descending order of the support degree, and constructing a dialogue frequent pattern tree, wherein the dialogue frequent pattern tree comprises the NULL root node and branch nodes, the NULL root node is an invalid value, and the branch nodes correspond to an intention candidate frequent item and the support degree of the sentence intention thereof. And finally, mining at least one conversation frequent item set from the constructed conversation frequent pattern tree to be used as a conversation frequent item set.

Further, mining at least one dialog frequent item set from the constructed dialog frequent pattern tree as a dialog frequent item set can be realized by adopting a frequent item set mining algorithm in the related art.

Illustratively, a condition mode base of each intention frequent item in the constructed frequent pattern tree can be constructed, and the dialogue frequent pattern tree of the intention frequent item is constructed based on the constructed condition mode base, wherein the condition mode base is a path set of a plurality of prefix paths which take the intention frequent item as a suffix item and are connected with the suffix item.

And then updating the frequent pattern tree based on each constructed dialogue frequent pattern tree, continuously executing steps of aiming at each intention frequent item in the constructed frequent pattern tree based on the updated dialogue frequent pattern tree, constructing a condition pattern base of the intention frequent item, and constructing the dialogue frequent pattern tree of the intention frequent item based on the constructed condition pattern base, until the constructed dialogue frequent pattern tree is empty or only comprises one path, outputting the intention frequent item corresponding to the dialogue frequent pattern tree, wherein one intention frequent item path corresponds to one dialogue frequent item set, and a plurality of intention frequent item paths correspond to a plurality of dialogue frequent item sets.

Illustratively, when the constructed dialog frequent pattern tree is empty, the prefix path of the dialog frequent pattern tree is determined as the intention frequent item, and when the constructed dialog frequent pattern tree only contains one path, all the combined paths are connected with the prefix path of the dialog frequent pattern tree as the intention frequent item.

In one or more embodiments of the present specification, after at least one conversation frequent item set corresponding to a plurality of pieces of conversation data is determined based on the sentence intentions of the conversation sentences, each conversation frequent item set does not need to be deduplicated with the same or similar items, and mining the conversation frequent item sets based on the sentence intentions can avoid redundant computation of unnecessary identical or similar items in the association rule mining process, optimize the conversation structure mining process, and improve the computation processing efficiency.

S106: acquiring a dialogue structure sequence corresponding to each reference dialogue item in each dialogue frequent item set, and determining at least one dialogue structure aiming at the plurality of dialogue data based on the dialogue structure sequence and each reference dialogue item in the dialogue frequent item set;

in one or more embodiments of the present specification, a plurality of reference dialog items mined from a dialog frequent item set are unordered, and a dialog structure sequence of a reference dialog item with respect to the entire dialog frequent item set, that is, an interaction dialog sequence of a dialog statement corresponding to the reference dialog item in an actual dialog scene, may be determined with all reference dialog items included in one dialog frequent item set as common dialog structure forms that may occur in one dialog scene.

Schematically, setting D as a frequently conversational item set; d1 D2, D3.. Dk is a reference conversation item in a collection of conversation spuriously items, which usually mainly includes sentence intent. D1 D2, D3.. Dk represent different sentence intentions, respectively, D1, D2, D3.. Dk do not determine the dialog sequence between each other;

in some embodiments, the dialog structure order may be an interactive dialog intent order of the determined sentence intentions;

in some embodiments, the sentence intent may be associated with a dialog statement of several sentence intents in the original dialog data, and after determining the dialog structure order, the reference dialog item may feed back an interactive dialog order in which the sentence intent corresponds to the dialog statement, such as a dialog interaction order in which D1, D2, and D3.. Dk respectively correspond to dialog interactions that frequently occur in a dialog scenario.

Specifically, the dialogue structure order of the reference dialogue items can be determined by combining the sentence order of a plurality of dialogue sentences in the original plurality of dialogue data, in the specific implementation, each dialogue data is taken as an object, and the sentence intention of each dialogue sentence in the dialogue data is determined, so that the dialogue intention order of the dialogue sentences can be marked based on the sentence order of each dialogue sentence, and a plurality of groups of reference dialogue intention orders corresponding to the dialogue data can be obtained; then, combining the sentence intentions corresponding to each reference dialog item in the dialog frequent item set, the sequence of the sentence intentions corresponding to each reference dialog item can be determined according to the sequence of the reference dialog intentions, and the sequence is also the dialog structure sequence corresponding to the reference dialog item.

Optionally, an expert service may be invoked, and a dialog structure sequence corresponding to each reference dialog item in each dialog frequent item set is set based on an expert side of the expert service.

Furthermore, the electronic device then sorts each reference conversation item of the conversation frequent item set according to the conversation structure sequence based on the conversation structure sequence and each reference conversation item in the conversation frequent item set, a plurality of reference conversation items in the sorted conversation frequent item set have the conversation structure sequence, the conversation frequent item set with the conversation structure sequence can be used as a conversation structure of the plurality of conversation data, and a plurality of conversation structures can be obtained under the condition that the conversation frequent item set is multiple.

Referring to fig. 3, fig. 3 is a schematic flowchart of another embodiment of a dialog structure processing method according to one or more embodiments of the present disclosure. Specifically, the method comprises the following steps:

s202: acquiring a plurality of dialogue data, and determining sentence intentions of dialogue sentences in each dialogue data;

reference may be made specifically to method steps of other embodiments of the present disclosure, which are not described herein again.

S204: taking the sentence intention of the dialogue sentence as a candidate dialogue reference item to obtain at least one candidate dialogue frequent item set containing the candidate dialogue reference item;

it will be appreciated that the combination of candidate conversation reference items may be referred to as a set of items, which in embodiments of the present specification may also be referred to as a set of candidate conversation chores.

In one or more embodiments of the present specification, in a process of mining an association conversation rule between a plurality of pieces of conversation data in a conversation scene to obtain a conversation structure, data mining of a frequent set of conversations is performed by determining a sentence intention of a conversation sentence and referring to the sentence intention, so that large resource consumption caused by directly performing data mining on the conversation sentence is avoided, and a calculation processing process can be quickly converged and simplified.

Illustratively, the association dialog rule is an implication in the form of "dialog X → dialog Y", such as meaning that the "resulting" dialog Y can be derived by dialog X, where dialog X and dialog Y are referred to as the predecessor and successor of the association rule, respectively. The dialog elements involved in the associated dialog rule may be plural.

S206: determining candidate item support corresponding to each candidate dialogue frequent item set based on the sentence intention of each dialogue sentence in each dialogue data;

it is to be understood that the dialogue sentences intended for the same sentence may be different, that is, the dialogue sentences intended for a certain sentence in all the dialogue data may be a plurality of different dialogue sentences.

The candidate item support degree is used for measuring the frequency degree or frequency degree of a certain sentence intention in all the dialogue data; candidate support may be used to quantify associated conversation rules between conversations, which may reflect the usefulness and certainty of discovered conversation rules.

Illustratively, assuming that the candidate dialog frequent item set contains i candidate dialog reference items (i is an integer), the candidate support degree is further determined by counting the occurrence times of the i candidate dialog reference items in all the dialog data at the same time.

In an exemplary implementation, the candidate support degree of the candidate dialogue frequent item set is further determined by counting the occurrence times of the i candidate dialogue reference items corresponding to the sentence intentions in all the dialogue data at the same time.

Illustratively, the candidate support degree may be the number of occurrences of the i candidate dialogue reference items corresponding to the sentence intention in all the dialogue data at the same time, or may be the probability of occurrences of the i candidate dialogue reference items corresponding to the sentence intention in all the dialogue data at the same time.

S208: determining a candidate conversation set containing all candidate conversation frequent item sets, and performing associated conversation mining processing on the candidate conversation set based on candidate item support to obtain at least one conversation frequent item set corresponding to the plurality of conversation data, wherein the conversation frequent item set comprises a plurality of conversation reference items;

in one or more embodiments of the present disclosure, a priori knowledge of the set of conversational complexities is used to explore the set of i +1 terms using the current set of i terms, assuming the current set of candidate conversational complexities as the set of i terms, by an iterative method of layer-by-layer searching. Because all the dialogue data are traversed in advance, all the frequent 1-item dialogue sets are found out, the frequent 1-item dialogue sets also take sentence intentions as candidate dialogue reference items, one candidate dialogue reference item is a candidate dialogue frequent item set, and suppose that the sentence intentions are n, at the moment, the number of the frequent 1-item dialogue sets is also n candidate dialogue frequent item sets taking the sentence intentions as candidate dialogue reference items, and the candidate dialogue set containing all the candidate dialogue frequent item sets is marked as i1;

then, a candidate dialog set i2 of the frequent 2-item set is mined based on the candidate dialog set i1, and the candidate dialog set i2 is mined to a candidate dialog set i 3. And finally, finding out strong rules in all the candidate conversation sets ik, namely generating associated conversation rules of conversation interest characteristics, wherein all the conversation frequent item sets in the candidate conversation sets ik can be further subjected to conversation structure processing to obtain a plurality of conversation structures.

In a possible implementation manner, the step of determining a candidate dialog set including all candidate dialog frequent item sets, and performing associated dialog mining processing on the candidate dialog set based on candidate support degrees to obtain at least one dialog frequent item set corresponding to the plurality of dialog data is performed, and may specifically be in the following form:

a2, performing reference item set filtering processing on the candidate dialogue frequent item set in the candidate dialogue set based on the candidate support degree to obtain a processed first candidate dialogue set;

the reference item filtering process may be understood as filtering out a set of items that do not meet the support requirement.

In a possible implementation manner, the candidate support degree corresponding to each candidate dialogue frequent item set in the candidate dialogue set may be obtained, and then the candidate dialogue frequent item set with the candidate support degree smaller than the first support degree threshold value is filtered out from the candidate dialogue set, so as to obtain the processed first candidate dialogue set.

The first support threshold is a threshold or a threshold of candidate support for the candidate conversation frequent item set, and is used for filtering out the candidate conversation frequent item set smaller than the first support threshold in the data mining process.

Illustratively, assuming a first support threshold of 50%, the current set of candidate dialogs is A, B, C,

the conversation frequent item set A comprises a plurality of candidate reference items, such as A [ X1, X3 and X5], and the support degree of the conversation frequent item set A is 80%;

the conversation frequent item set B comprises a plurality of candidate reference items, such as B [ X1, X2 and X5], and the support degree of the conversation frequent item set B is 70%;

the conversation frequent item set C comprises a plurality of candidate reference items, such as C [ X2, X4 and X5], and the support degree of the conversation frequent item set C is 45%;

wherein, X1, X2, X3, X4 and X5 represent candidate reference items which are different sentence intentions;

the electronic equipment acquires the candidate support degree corresponding to each candidate conversation frequent item set in the candidate conversation set { A, B, C }, and then filters out the candidate conversation frequent item set C [ X2, X4 and X5] with the candidate support degree smaller than a first support degree threshold value from the candidate conversation set { A, B, C }, so as to obtain a processed first candidate conversation set { A, B }.

A4, performing reference item connection processing on the first candidate dialogue set based on the first candidate dialogue reference item of each first candidate dialogue frequent item set in the first candidate dialogue set to obtain a processed second candidate dialogue set;

specifically, the electronic device obtains the number of first sentence intention items corresponding to all first candidate dialogue reference items based on the first candidate dialogue reference items of each first candidate dialogue frequent item set in the first candidate dialogue set, and obtains the number of second sentence intention items based on the number of the first sentence intention items and the unit number of the incremental intention items;

illustratively, assuming that the number of terms in each first candidate dialog frequent term set in the current first candidate dialog set is i, that is, the number of first sentence intention terms is i, and the incremental intention term unit number is j (e.g., j = 1), the number of second sentence intention terms is obtained based on the number of first sentence intention terms and the incremental intention term unit number: "i + j";

specifically, the electronic device performs sentence intent combination on the first candidate conversation frequent item set in the first candidate conversation set based on the second sentence intent item number to obtain a second candidate conversation frequent item set indicated by the second sentence intent item number, and takes all the second candidate conversation frequent item sets as second candidate conversation sets.

Schematically, assuming that the number of terms in each first candidate conversation frequent item set in the current first candidate conversation set is i, and the unit number of incremental intention terms is j (e.g., j = 1), reference term connection processing performs reference term connection on the first candidate conversation frequent item set of i terms according to the number of "i + j" terms, where the number of terms in the connected first candidate conversation frequent item set is "i + j" as a new second candidate conversation frequent item set, and all the connected second candidate conversation frequent item sets are the second candidate conversation set;

it should be noted that, in the reference item connection processing process, a pruning strategy needs to be performed on the first candidate conversation frequent item set that does not satisfy the priori condition of the frequent item set.

The prior conditions of the frequent item set are as follows: any infrequent set of i items is not a subset of the frequent set of "i + j" items. Thus, if a subset of i items of a candidate "i + j" item set is not in the candidate conversation set Lk-j, the candidate is unlikely to be frequent, so that pruning that would effect a frequent item set can be eliminated from the candidate conversation set,

schematically, taking i =2,j =1 as an example, the generation process of the set C3 of frequent 3 items of the second candidate dialog is: the set L2 of the first set of candidate dialog frequent 2 items is: l2= { { I1, I2}, { I1, I3}, { I1, I5}, { I2, I4}, { I2, I3}, { I2, I4}, { I2, I5} }, starting from the connecting step, first C3 { { I1, I2, I3}, { I1, I2, I5}, { I1, I3, I5}, { I1, I2, I4}, { I2, I3, I5}, { I2, I4, I5}, { C3 is generated by L2 connecting to itself. Pruning is performed according to the priori condition of the frequent item set, all subsets of the frequent item set must also be frequent, it can be determined that there are 4 candidate sets { I1, I3, I5}, { I2, I3, I4}, { I2, I3, I5}, { I2, I4, I5} } that are unlikely to be frequent because there are subsets that do not belong to the frequent set and therefore they are deleted from C3, and the remaining two candidate sets "{ I1, I2, I3}, { I1, I2, I5}" to obtain the set L3 of the second frequent 3 item set of candidate dialogues.

A6, detecting whether a second candidate frequent item set in the second candidate dialogue set meets a frequent item set mining ending condition;

the frequent item set mining end condition may be that a second candidate support degree of a second candidate frequent item set in a second candidate conversation set does not satisfy a second support degree threshold;

the second support degree threshold is a threshold or a critical value for ending mining corresponding to the second candidate support degree, and the second support degree threshold may be the same as or different from the first support threshold. The difference between the second support threshold and the first support threshold may be used for system environment tolerance to better fit the actual dialog scenario.

Specifically, the electronic device may determine a second candidate support degree corresponding to each second candidate frequent item set in a second candidate conversation set;

illustratively, assuming that the second candidate dialog reference item set includes i second candidate dialog reference items (i is an integer), the candidate support degree is further determined by counting the occurrence times of the i second candidate dialog reference items in all the dialog data at the same time.

In an exemplary implementation, the candidate support degree of the candidate dialogue frequent item set is further determined by counting the occurrence times of the i second candidate dialogue reference items corresponding to the sentence intentions in all the dialogue data at the same time.

Illustratively, the candidate support degree may be the number of occurrences of the i second candidate dialogue reference items corresponding to the sentence intention in all the dialogue data at the same time, or may be the probability of the occurrence of the i second candidate dialogue reference items corresponding to the sentence intention in all the dialogue data at the same time.

Specifically, if all the second candidate support degrees are smaller than a second support degree threshold value, the electronic device determines that a second candidate frequent item set in the second candidate conversation set meets a frequent item set mining end condition;

specifically, if at least one second candidate support degree is greater than or equal to a second support degree threshold, the electronic device determines that a second candidate frequent item set in the second candidate dialog set does not satisfy the frequent item set mining end condition.

A8, if a second candidate frequent item set in the second candidate conversation set meets a frequent item set mining ending condition, acquiring an original first candidate frequent item set corresponding to the second candidate frequent item set in the second candidate conversation set, and determining the original first candidate conversation frequent item set as a conversation frequent item set aiming at the plurality of conversation data;

it can be understood that, in the case that the second candidate frequent item set in the second candidate conversation set satisfies the condition of ending the frequent item set mining, the original first candidate frequent item set corresponding to the second candidate frequent item set in the second candidate conversation set needs to be obtained, where the original first candidate frequent item set is also the first candidate frequent item set before the reference item connection processing is performed. Determining a plurality of original first candidate conversation frequent item sets as conversation frequent item sets for the plurality of conversation data;

and A10, if a second candidate frequent item set in the second candidate dialogue set does not meet the condition of ending the mining of the frequent item set, taking the second candidate dialogue set as the candidate dialogue set, and executing the step of filtering out a reference item set of the candidate dialogue frequent item set in the candidate dialogue set based on the candidate support degree to obtain a processed first candidate dialogue set.

It can be understood that, if the second candidate frequent item set in the second candidate dialog set does not satisfy the frequent item set mining end condition, the electronic device executes step A2 to continue data mining.

S210: acquiring a dialog structure sequence corresponding to each reference dialog item in each dialog frequent item set;

in one possible implementation, the electronic device may acquire a dialog sentence sequence of each dialog sentence in each dialog data, determine a dialog intention sequence of each sentence intention in the dialog data based on the dialog sentence sequence, and then determine a dialog structure sequence corresponding to each reference dialog item based on the dialog intention sequence and the reference sentence intention corresponding to the reference dialog item;

in some embodiments, the dialog structure order may be an interaction dialog intention order of the determined sentence intentions, and in a specific implementation, each dialog data is taken as an object, and as the sentence intention of each dialog sentence in the dialog data is determined, the dialog intention order of the dialog sentences may be marked based on the dialog sentence order of each dialog sentence, so that a plurality of groups of reference dialog intention orders corresponding to the dialog data may be obtained; then, in combination with the reference sentence intentions corresponding to each reference dialog item in the dialog frequent item set, the sequence of the reference sentence intentions corresponding to each reference dialog item, that is, the dialog structure sequence corresponding to the reference dialog item, may be determined according to the reference dialog intention sequence.

S212: determining at least one dialog structure for the plurality of dialog data based on the dialog structure order and each reference dialog item in the set of dialog chorus items.

In one or more embodiments of the present specification, at least one dialog structure of the dialog data is a dialog structure formed by reference dialog items corresponding to sentence intentions, and in an actual dialog scenario, after determining the dialog structure, the reference sentence intentions corresponding to each reference dialog item in the dialog structure may be associated with corresponding dialog statements in the original dialog data, so that the associated dialog structure may feed back common dialog statement forms in addition to the intended dialog structure.

Schematically, the electronic device may obtain a reference sentence intention corresponding to each reference dialogue item in each dialogue frequent item set, determine at least one reference dialogue sentence corresponding to the reference sentence intention from a plurality of dialogue data, and perform association processing on the reference dialogue sentence and the reference dialogue item in the dialogue frequent item set to obtain the dialogue structure after the association processing;

optionally, the associating process may be to associate the reference dialog item with the reference dialog statement in a data structure mapping manner.

In one or more embodiments of the specification, after a dialog structure is mined based on dialog data in an actual dialog transaction scene to obtain a plurality of frequently-used dialog structures, the dialog structures can be applied to a specific interactive dialog scene online and can be directly input into an interactive dialog system for applications such as question answering systems, customer service tools, self-service inquiry and the like, on one hand, the dialog structures sense the interactive characteristics of the whole dialog, and can better assist in generating subsequent dialogues, so that the dialog generation quality and the accuracy of dialog expression are improved; on the other hand, focusing on parts which are not related in the related technology, the overall intention of a conversation scene can be better fed back by sensing the overall structure of the conversation, conversation early warning and abnormal user group monitoring in the corresponding transaction scene can be expanded based on the conversation structure, and malicious behaviors of the abnormal user group on normal maintenance and management of corresponding transactions based on a common conversation form can be avoided.

Referring to fig. 4, fig. 4 is a schematic flowchart of another embodiment of a dialog structure processing method according to one or more embodiments of the present disclosure. Specifically, the method comprises the following steps:

s302: acquiring a plurality of session data corresponding to a user side and a customer service side in a target transaction scene;

the target transaction scene can be a travel service transaction scene, a take-out service transaction scene, an online shopping transaction scene, a fund-consuming service transaction scene and other collaborative conversation interaction scenes.

In one or more embodiments of the present specification, multiple rounds of historical conversations in one or more collaborative conversation interaction scenarios, such as a trip service transaction scenario, a take-out service transaction, an online shopping transaction, a cash transaction scenario, and the like, may be obtained as the conversation data, and for example, in a plurality of conversation scenarios (e.g., a trip service conversation scenario, a take-out service, an online shopping, a cash transaction scenario, and the like), multiple rounds of conversation statements between the user side and the service side may be saved as the conversation data each time the user side and the service side perform at least one round of conversation.

In one or more embodiments of the present disclosure, the information medium of the dialog data may be text, pictures, voice, etc.

Schematically, taking a money-consuming transaction scenario as an example, in a consumption financial transaction scenario, the user may negotiate repayment, service complaints, credit investigation, and the like, and the user terminal may initiate a dialog window to the platform service terminal to perform a corresponding interactive dialog. In the scene of a money-eliminating transaction, the recognition of a conversation structure aiming at the insurance customer service can be realized by executing the conversation structure processing method designed by the specification, and the conversation structure of a conversation user is excavated, so that on one hand, the conversation structure can be monitored and early warned, and a corresponding service party is assisted: 1. capturing conversation structures with suddenly increased quantity in time, 2, making a conversation scheme for dealing with common conversation structures, and 3, strictly auditing the related application flows of high-risk conversation groups, such as negotiation repayment, credit investigation and complaint complaints; on the other hand, the application of the extracted conversation structure to the consumption financial affair scene can improve the conversation processing efficiency of the service party, and the service party can quickly generate a conversation reply based on the conversation structure extracted in advance, thereby providing high-efficiency professional services.

S304: determining a sentence intention of the dialogue statement in each dialogue data, and determining at least one dialogue frequent item set corresponding to the plurality of dialogue data based on the sentence intention of the dialogue statement, wherein the dialogue frequent item set comprises a plurality of dialogue reference items;

S306: acquiring a dialogue structure sequence corresponding to each reference dialogue item in each dialogue frequent item set, and determining at least one dialogue structure aiming at the plurality of dialogue data based on the dialogue structure sequence and each reference dialogue item in the dialogue frequent item set;

reference may be made in detail to method steps in other embodiments of the present disclosure, which are not described in detail herein.

S308: and configuring the at least one dialogue structure for at least one customer service side in a target transaction scene to indicate that the customer service side inquires a reply statement based on the dialogue structure after receiving an inquiry statement of a user side and performs transaction processing based on the reply statement.

It can be understood that after the electronic device obtains the plurality of dialog structures in the target transaction scene, the plurality of dialog structures can feed back frequently-occurring and common multi-round dialog forms in the actual transaction scene, the electronic device can apply the dialog structures to an interactive dialog service in the target transaction scene, the interactive dialog service is associated with a plurality of customer service terminals, and the customer service terminals can quickly generate a dialog reply in the actual dialog application by means of the plurality of dialog structures.

Schematically, after the electronic device configures the dialogue structure to a plurality of customer service terminals in a target transaction scene, if a user terminal initiates a dialogue window in the target transaction scene based on actual needs and sends an inquiry statement in the dialogue window, the customer service terminal can receive the inquiry statement of the user terminal, then retrieve the inquiry statement by means of the plurality of dialogue structures, can quickly match the inquiry dialogue items related to the inquiry statement in the dialogue structure, quickly generate a dialogue reply statement based on a next reference dialogue item of the inquiry dialogue item, and send the dialogue reply statement to the user terminal for transaction processing.

S310: and under a target transaction scene, monitoring the structure use time index of each dialog structure, and performing transaction early warning processing based on the structure use time index.

The structure use time index can be understood as counting the use times of the dialog structure by taking a time period as a reference, and the structure use time index can feed back the structure use number of a certain time dimension (such as day, week and month).

In a target transaction scene, a situation that normal service management of a target transaction is damaged or maliciously influenced exists, taking the target transaction scene as a consumption financial transaction scene as an example, a large number of counter-solicited behaviors may be provided within a period of time to help a lending user to resist abnormal transaction behaviors such as legal repayment, malicious complaints, damage and credit, and an interactive conversation phenomenon between a user side and a customer service side is usually involved.

Furthermore, in order to identify the abnormal transaction behaviors, a plurality of session structures mined based on actual session conditions in a target transaction scene are monitored, structure use time indexes of each session structure are monitored, and when the structure use time indexes are larger than the set index threshold value, transaction early warning processing is performed.

Illustratively, the structure usage time index may be: monitoring the usage amount of each dialogue structure in a week, and monitoring and early warning based on the structure usage index. The customer service end can be assisted: capturing abnormal conversation structures with suddenly increased quantity in time, and monitoring abnormal conversation behaviors based on a normal conversation form; formulating a dialoging scheme and a transaction early warning strategy for dealing with the abnormal dialog structure; the monitoring level of the high-risk conversation users using the abnormal conversation structure is improved, and the user auditing strength is improved for related application flows such as negotiation repayment, credit complaint and the like under the consumption financial affairs.

In one or more embodiments of the present specification, the electronic device determines at least one dialog structure for the plurality of dialog data by determining a sentence intent of a dialog sentence in the plurality of dialog data, determining at least one dialog frequent item set composed of a plurality of dialog reference items corresponding to the plurality of dialog data based on the sentence intent of the dialog sentence, and then obtaining a dialog structure sequence corresponding to each reference dialog item in the dialog frequent item set to combine the dialog frequent item set. The method has the advantages that the orderly conversation structure can be obtained by mining the disordered conversation frequent item set based on the sentence intention of the conversation sentence and then determining the conversation structure sequence of the reference conversation item, the integral conversation characteristic under the integral conversation scene can be sensed through the orderly conversation structure, the conversation limitation caused by focusing local single-round conversation reply is avoided, the intelligence of conversation processing is improved, and the conversation quality and the accuracy of conversation expression can be greatly improved when the plurality of conversation structures based on the integral conversation characteristic are applied to the local single-round conversation scene.

The dialog structure processing device provided in this specification will be described in detail below with reference to fig. 5. It should be noted that the dialog structure processing apparatus shown in fig. 5 is used for executing the method of the embodiment shown in fig. 1 to 4 of this specification, and for convenience of description, only the part relevant to this specification is shown, and details of the technology are not disclosed, please refer to the embodiment shown in fig. 1 to 4 of this specification.

Please refer to fig. 5, which shows a schematic structural diagram of the dialog structure processing device in the present specification. The dialog structure processing device 1 may be implemented as all or part of a user terminal by software, hardware or a combination of both. According to some embodiments, the dialog structure processing apparatus 1 comprises an intent determination module 11, an item set determination module 12, and a structure determination module 13, specifically configured to:

an intention determining module 11, configured to obtain a plurality of dialogue data, where the dialogue data includes at least one dialogue statement, and determine a sentence intention of the dialogue statement in each dialogue data;

an item set determining module 12, configured to determine at least one dialog frequent item set corresponding to the plurality of dialog data based on the sentence intent of the dialog sentence, where the dialog frequent item set includes a plurality of dialog reference items;

a structure determining module 13, configured to obtain a dialog structure sequence corresponding to each reference dialog item in each dialog frequent item set, and determine at least one dialog structure for the plurality of dialog data based on the dialog structure sequence and each reference dialog item in the dialog frequent item set.

Optionally, as shown in fig. 6, the item set determining module 12 includes:

a candidate set determining unit 121, configured to use the sentence intent of the conversational sentence as a candidate conversational reference item, so as to obtain at least one candidate set of conversational frequentness items that includes the candidate conversational reference item;

a support degree determining unit 122, configured to determine a candidate support degree corresponding to each candidate conversational complexity item set based on a sentence intent of each conversational sentence in each piece of conversational data;

an item set determining unit 123, configured to determine a candidate dialog set including all the candidate dialog frequent item sets, and perform associated dialog mining processing on the candidate dialog set based on the candidate support degree to obtain at least one dialog frequent item set corresponding to the plurality of dialog data.

Optionally, as shown in fig. 7, the item set determining unit 123 is configured to:

a filtering subunit 1231, configured to perform reference item set filtering processing on the candidate conversation frequent item set in the candidate conversation set based on the candidate support degree, so as to obtain a processed first candidate conversation set;

a connection subunit 1232, configured to perform reference item connection processing on the first candidate dialog set based on a first candidate dialog reference item of each first candidate dialog frequent item set in the first candidate dialog set, so as to obtain a second candidate dialog set after processing;

a detecting subunit 1233, configured to detect whether a second candidate frequent item set in the second candidate conversation set meets a frequent item set mining end condition;

the detecting subunit 1233 is further configured to, if the second candidate frequent item set in the second candidate dialog set meets a frequent item set mining end condition, obtain an original first candidate frequent item set corresponding to the second candidate frequent item set in the second candidate dialog set, and determine the original first candidate dialog frequent item set as a dialog frequent item set for the plurality of dialog data;

the detecting subunit 1233 is further configured to, if the second candidate frequent item set in the second candidate conversation set does not satisfy the frequent item set mining end condition, take the second candidate conversation set as the candidate conversation set, and perform the step of performing reference item set filtering processing on the candidate conversation frequent item set in the candidate conversation set based on the candidate support degree to obtain the processed first candidate conversation set.

Optionally, the filtering subunit 1231 is configured to:

acquiring the candidate support degree corresponding to each candidate conversation frequent item set in the candidate conversation set;

and filtering the candidate dialogue frequent item set with the candidate support degree smaller than a first support degree threshold value from the candidate dialogue set to obtain a processed first candidate dialogue set.

Optionally, the connection subunit 1232 is configured to:

acquiring the number of first sentence intention items corresponding to all first candidate dialogue reference items based on the first candidate dialogue reference items of each first candidate dialogue frequent item set in the first candidate dialogue set, and acquiring the number of second sentence intention items based on the number of the first sentence intention items and the unit number of incremental intention items;

and sentence intention combination is carried out on the first candidate dialogue frequent item set in the first candidate dialogue set based on the second sentence intention item number so as to obtain a second candidate dialogue frequent item set indicated by the second sentence intention item number, and all the second candidate dialogue frequent item sets are used as a second candidate dialogue set.

Optionally, the detecting subunit 1233 is further configured to:

determining a second candidate support degree corresponding to each second candidate frequent item set in the second candidate dialogue set;

if all the second candidate support degrees are smaller than a second support degree threshold value, determining that a second candidate frequent item set in the second candidate conversation set meets a frequent item set mining end condition;

and if at least one second candidate support degree is larger than or equal to the second support degree threshold value, determining that a second candidate frequent item set in the second candidate conversation set does not meet the frequent item set mining end condition.

Optionally, the structure determining module 13 is configured to: and acquiring a reference sentence intention corresponding to each reference conversation item in the conversation frequent item sets, and determining a conversation structure sequence corresponding to each reference conversation item based on the reference sentence intentions.

Optionally, as shown in fig. 8, the structure determining module 13 includes:

an intention order determination unit 131 configured to acquire a dialogue sentence order of each dialogue sentence in each dialogue data, and determine a dialogue intention order of each sentence intention in the dialogue data based on the dialogue sentence order;

a structure order determining unit 132, configured to determine a dialog structure order corresponding to each reference dialog item based on each dialog intention order and the reference sentence intention corresponding to the reference dialog item.

Optionally, the intention determining module 11 is configured to:

and acquiring a plurality of session data corresponding to the client and the customer service in a target transaction scene.

Optionally, the apparatus 1 is further configured to:

acquiring a reference sentence intention corresponding to each reference dialogue item in each dialogue frequent item set, determining at least one reference dialogue sentence corresponding to the reference sentence intention from the plurality of dialogue data, and associating the reference dialogue sentence with the reference dialogue item in the dialogue frequent item set to obtain the dialogue structure after association processing; and/or the presence of a gas in the gas,

configuring at least one dialogue structure for at least one customer service end in a target transaction scene to indicate that the customer service end inquires a reply statement based on the dialogue structure after receiving an inquiry statement of a user end and performs transaction processing based on the reply statement; and/or the presence of a gas in the gas,

and under a target transaction scene, monitoring the structure use time index of each dialog structure, and performing transaction early warning processing based on the structure use time index.

It should be noted that, when the dialog structure processing apparatus provided in the foregoing embodiment executes the dialog structure processing method, only the division of the functional modules is illustrated, and in practical applications, the functions may be distributed to different functional modules according to needs, that is, the internal structure of the device may be divided into different functional modules to complete all or part of the functions described above. In addition, the dialog structure processing apparatus and the dialog structure processing method provided in the foregoing embodiments belong to the same concept, and details of implementation processes thereof are shown in the method embodiments, and are not described herein again.

The above-mentioned serial numbers are for description purposes only and do not represent the merits of the embodiments.

The present specification further provides a computer storage medium, where the computer storage medium may store a plurality of instructions, where the instructions are suitable for being loaded by a processor and executing the dialog structure processing method according to the embodiment shown in fig. 1 to 4, and a specific execution process may refer to specific descriptions of the embodiment shown in fig. 1 to 4, which is not described herein again.

The present specification further provides a computer program product, where at least one instruction is stored, and the at least one instruction is loaded by the processor and executes the dialog structure processing method according to the embodiment shown in fig. 1 to 4, where a specific execution process may refer to a specific description of the embodiment shown in fig. 1 to 4, and is not described herein again.

Referring to fig. 9, a block diagram of an electronic device according to an exemplary embodiment of the present disclosure is shown. The electronic device in this specification may include one or more of the following components: a processor 110, a memory 120, an input device 130, an output device 140, and a bus 150. The processor 110, memory 120, input device 130, and output device 140 may be connected by a bus 150.

Processor 110 may include one or more processing cores. The processor 110 connects various parts within the overall electronic device using various interfaces and lines, and performs various functions of the electronic device 100 and processes data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 120 and calling data stored in the memory 120. Alternatively, the processor 110 may be implemented in hardware using at least one of Digital Signal Processing (DSP), field-programmable gate Array (FPGA), and Programmable Logic Array (PLA). The processor 110 may integrate one or more of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a modem, and the like. The CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for rendering and drawing display content; the modem is used to handle wireless communications. It is understood that the modem may not be integrated into the processor 110, but may be implemented by a communication chip.

The Memory 120 may include a Random Access Memory (RAM) or a read-only Memory (ROM). Optionally, the memory 120 includes a non-transitory computer-readable medium. The memory 120 may be used to store instructions, programs, code sets, or instruction sets. The memory 120 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing various method embodiments described below, and the like, and the operating system may be an Android (Android) system, including a system based on Android system depth development, an IOS system developed by apple, including a system based on IOS system depth development, or other systems. The data storage area may also store data created by the electronic device during use, such as a phonebook, audio-visual data, chat log data, and the like.

Referring to fig. 10, the memory 120 may be divided into an operating system space, where an operating system is run, and a user space, where native and third-party applications are run. In order to ensure that different third-party application programs can achieve a better operation effect, the operating system allocates corresponding system resources for the different third-party application programs. However, the requirements of different application scenarios in the same third-party application program on system resources are different, for example, in a local resource loading scenario, the third-party application program has a higher requirement on the disk reading speed; in an animation rendering scene, the third-party application program has a high requirement on the performance of the GPU. The operating system and the third-party application program are independent from each other, and the operating system cannot sense the current application scene of the third-party application program in time, so that the operating system cannot perform targeted system resource adaptation according to the specific application scene of the third-party application program.

In order to enable the operating system to distinguish a specific application scenario of the third-party application program, data communication between the third-party application program and the operating system needs to be opened, so that the operating system can acquire current scenario information of the third-party application program at any time, and further perform targeted system resource adaptation based on the current scenario.

Taking an operating system as an Android system as an example, programs and data stored in the memory 120 are as shown in fig. 11, and a Linux kernel layer 320, a system runtime library layer 340, an application framework layer 360, and an application layer 380 may be stored in the memory 120, where the Linux kernel layer 320, the system runtime library layer 340, and the application framework layer 360 belong to an operating system space, and the application layer 380 belongs to a user space. The Linux kernel layer 320 provides underlying drivers for various hardware of the electronic device, such as a display driver, an audio driver, a camera driver, a bluetooth driver, a Wi-Fi driver, power management, and the like. The system runtime library layer 340 provides a main feature support for the Android system through some C/C + + libraries. For example, the SQLite library provides support for a database, the OpenGL/ES library provides support for 3D drawing, the Webkit library provides support for a browser kernel, and the like. Also provided in the system runtime library layer 340 is an Android runtime library (Android runtime), which mainly provides some core libraries capable of allowing developers to write Android applications using the Java language. The application framework layer 360 provides various APIs that may be used in building an application, and developers may build their own applications by using these APIs, such as activity management, window management, view management, notification management, content provider, package management, session management, resource management, and location management. At least one application program runs in the application layer 380, and the application programs may be native application programs carried by the operating system, such as a contact program, a short message program, a clock program, a camera application, and the like; or a third-party application developed by a third-party developer, such as a game application, an instant messaging program, a photo beautification program, and the like.

Taking an operating system as an IOS system as an example, programs and data stored in the memory 120 are shown in fig. 12, and the IOS system includes: a Core operating system Layer 420 (Core OS Layer), a Core Services Layer 440 (Core Services Layer), a Media Layer 460 (Media Layer), and a touchable Layer 480 (Cocoa Touch Layer). The kernel operating system layer 420 includes an operating system kernel, drivers, and underlying program frameworks that provide functionality closer to hardware for use by program frameworks located in the core services layer 440. The core services layer 440 provides system services and/or program frameworks, such as a Foundation framework, an account framework, an advertisement framework, a data storage framework, a network connection framework, a geographic location framework, a motion framework, and so forth, as required by the application. The media layer 460 provides audiovisual related interfaces for applications, such as graphics image related interfaces, audio technology related interfaces, video technology related interfaces, audio video transmission technology wireless playback (AirPlay) interfaces, and the like. Touchable layer 480 provides various common interface-related frameworks for application development, and touchable layer 480 is responsible for user touch interaction operations on the electronic device. Such as a local notification service, a remote push service, an advertising framework, a game tools framework, a message User Interface (UI) framework, a User Interface UIKit framework, a map framework, and so forth.

In the framework illustrated in FIG. 12, the framework associated with most applications includes, but is not limited to: a base framework in the core services layer 440 and a UIKit framework in the touchable layer 480. The base framework provides many basic object classes and data types, provides the most basic system services for all applications, and is UI independent. While the class provided by the UIKit framework is a basic library of UI classes for creating touch-based user interfaces, iOS applications can provide UIs based on the UIKit framework, so it provides an infrastructure for applications for building user interfaces, drawing, processing and user interaction events, responding to gestures, and the like.

The Android system may be referred to as a manner and a principle for implementing data communication between the third-party application program and the operating system in the IOS system, and details are not repeated herein.

The input device 130 is used for receiving input instructions or data, and the input device 130 includes, but is not limited to, a keyboard, a mouse, a camera, a microphone, or a touch device. The output device 140 is used for outputting instructions or data, and the output device 140 includes, but is not limited to, a display device, a speaker, and the like. In one example, the input device 130 and the output device 140 may be combined, and the input device 130 and the output device 140 are touch display screens for receiving touch operations of a user on or near the touch display screens by using any suitable object such as a finger, a touch pen, and the like, and displaying user interfaces of various applications. The touch display screen is generally provided on a front panel of the electronic device. The touch display screen may be designed as a full-face screen, a curved screen, or a profiled screen. The touch display screen can also be designed to be a combination of a full-face screen and a curved-face screen, and a combination of a special-shaped screen and a curved-face screen, which is not limited in the specification.

In addition, those skilled in the art will appreciate that the configurations of the electronic devices illustrated in the above-described figures do not constitute limitations on the electronic devices, which may include more or fewer components than illustrated, or some components may be combined, or a different arrangement of components. For example, the electronic device further includes a radio frequency circuit, an input unit, a sensor, an audio circuit, a wireless fidelity (WiFi) module, a power supply, a bluetooth module, and other components, which are not described herein again.

In this specification, the execution subject of each step may be the electronic device described above. Optionally, the execution subject of each step is an operating system of the electronic device. The operating system may be an android system, an IOS system, or another operating system, which is not limited in this specification.

The electronic device of this specification may further have a display device mounted thereon, and the display device may be various devices that can implement a display function, for example: a cathode ray tube display (CR), a light-emitting diode display (LED), an electronic ink panel, a Liquid Crystal Display (LCD), a Plasma Display Panel (PDP), and the like. A user may utilize a display device on the electronic device 101 to view information such as displayed text, images, video, and the like. The electronic device may be a smartphone, a tablet, a server, a gaming device, an AR (Augmented Reality) device, an automobile, a data storage device, an audio playback device, a video playback device, a notebook, a desktop computing device, a wearable device such as an electronic watch, an electronic glasses, an electronic helmet, an electronic bracelet, an electronic necklace, an electronic garment, or the like.

In the electronic device shown in fig. 9, where the electronic device may be a terminal, the processor 110 may be configured to call the 9 application programs stored in the memory 120, and specifically perform the following operations:

determining a sentence intention of the dialogue statement in each dialogue data, and determining at least one dialogue frequent item set corresponding to the plurality of dialogue data based on the sentence intention of the dialogue statement, wherein the dialogue frequent item set comprises a plurality of dialogue reference items;

In one embodiment, the processor 110, when executing the determining at least one dialog frequent item set corresponding to the plurality of dialog data based on the sentence intent of the dialog sentence, specifically performs the following steps:

taking the sentence intention of the dialogue sentence as a candidate dialogue reference item to obtain at least one candidate dialogue frequent item set containing the candidate dialogue reference item;

determining candidate item support corresponding to each candidate dialogue frequent item set based on the sentence intention of each dialogue sentence in each dialogue data;

and determining a candidate conversation set containing all the candidate conversation frequent item sets, and performing associated conversation mining processing on the candidate conversation set based on the candidate item support degree to obtain at least one conversation frequent item set corresponding to the plurality of conversation data.

In an embodiment, the processor 110 performs the associated dialog mining process on the candidate dialog set based on the candidate support degree to obtain at least one dialog frequent item set corresponding to the plurality of dialog data, and specifically performs the following steps:

performing reference item set filtering processing on the candidate dialogue frequent item set in the candidate dialogue set based on the candidate support degree to obtain a processed first candidate dialogue set;

performing reference item connection processing on the first candidate dialogue set based on a first candidate dialogue reference item of each first candidate dialogue frequent item set in the first candidate dialogue set to obtain a processed second candidate dialogue set;

detecting whether a second candidate frequent item set in the second candidate dialogue set meets a frequent item set mining end condition;

if a second candidate frequent item set in the second candidate conversation set meets a frequent item set mining ending condition, acquiring an original first candidate frequent item set corresponding to the second candidate frequent item set in the second candidate conversation set, and determining the original first candidate conversation frequent item set as a conversation frequent item set aiming at the plurality of conversation data;

and if the second candidate frequent item set in the second candidate dialogue set does not meet the end condition of frequent item set mining, taking the second candidate dialogue set as the candidate dialogue set, and executing the step of performing reference item set filtering processing on the candidate dialogue frequent item set in the candidate dialogue set based on the candidate support degree to obtain the processed first candidate dialogue set.

In an embodiment, the processor 110, after performing the reference item set filtering processing on the candidate dialog frequent item set in the candidate dialog set based on the candidate support degree to obtain a processed first candidate dialog set, specifically performs the following steps:

In an embodiment, the processor 110, after executing the first candidate dialog reference item based on each first candidate dialog frequent item set in the first candidate dialog set, performs reference item connection processing on the first candidate dialog set to obtain a second candidate dialog set after processing, and specifically executes the following steps:

In an embodiment, the processor 110 specifically performs the following steps when performing the detecting whether the second candidate frequent item set in the second candidate dialog set satisfies the frequent item set mining end condition:

determining a second candidate support degree corresponding to each second candidate frequent item set in the second candidate conversation set;

if all the second candidate support degrees are smaller than a second support degree threshold value, determining that a second candidate frequent item set in the second candidate dialogue set meets a frequent item set mining end condition;

In an embodiment, the processor 110 specifically performs the following steps in the step of obtaining the dialog structure sequence corresponding to each reference dialog item in each dialog frequent item set:

and acquiring a reference sentence intention corresponding to each reference conversation item in the conversation frequent item sets, and determining a conversation structure sequence corresponding to each reference conversation item based on the reference sentence intentions.

In one embodiment, the processor 110 specifically performs the following steps in determining the dialog structure order corresponding to each reference dialog item based on the respective reference sentence intentions:

acquiring a dialogue statement sequence of each dialogue statement in each dialogue data, and determining a dialogue intention sequence of each sentence intention in the dialogue data based on the dialogue statement sequence;

and determining the dialog structure sequence corresponding to each reference dialog item based on the dialog intention sequences and the reference sentence intention corresponding to the reference dialog item.

In an embodiment, the processor 110 performs the following steps in the step of acquiring the plurality of session data:

In one embodiment, the processor 110 further performs the following steps after performing the determining at least one dialog structure for the plurality of dialog data:

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a read-only memory or a random access memory.

The above disclosure is only for the purpose of illustrating the preferred embodiments of the present disclosure, and it is not intended to limit the scope of the present disclosure, so that the present disclosure will be covered by the claims and their equivalents.

Claims

1. A dialog structure processing method, the method comprising:

2. The method of claim 1, the determining at least one conversational popularity set corresponding to the plurality of conversational data based on the sentence intent of the conversational sentence, comprising:

and determining a candidate dialogue set containing all the candidate dialogue frequent item sets, and performing associated dialogue mining processing on the candidate dialogue set based on the candidate support degree to obtain at least one dialogue frequent item set corresponding to the plurality of dialogue data.

3. The method of claim 2, wherein said performing associated conversation mining on said candidate set of conversations based on said candidate support to obtain at least one set of conversation artifacts corresponding to said plurality of conversation data comprises:

4. The method of claim 3, wherein the performing reference item set filtering processing on the candidate dialogue frequent item set in the candidate dialogue set based on the candidate support degree to obtain a first candidate dialogue set after processing comprises:

5. The method of claim 3, wherein the performing reference item connection processing on the first candidate dialog set based on the first candidate dialog reference item of each first candidate dialog frequent item set in the first candidate dialog set to obtain a second candidate dialog set after processing comprises:

6. The method of claim 3, said detecting whether a second candidate frequent item set in the second candidate conversation set satisfies a frequent item set mining end condition, comprising:

7. The method of claim 1, wherein said obtaining a dialog structure order corresponding to each reference dialog item in each of said set of dialog bloom items comprises:

8. The method of claim 7, wherein determining a dialog structure order for each reference dialog item based on the respective reference sentence intent comprises:

9. The method of claim 1, the obtaining a plurality of session data, comprising:

10. The method of claim 1, after the determining at least one dialog structure for the plurality of dialog data, further comprising:

acquiring a reference sentence intention corresponding to each reference dialogue item in each dialogue frequent item set, determining at least one reference dialogue sentence corresponding to the reference sentence intention from the plurality of dialogue data, and performing association processing on the reference dialogue sentences and the reference dialogue items in the dialogue frequent item set to obtain the dialogue structure after association processing; and/or the presence of a gas in the atmosphere,

11. A dialog structure processing apparatus, the apparatus comprising:

12. A computer storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the method steps according to any of claims 1 to 10.

13. A computer program product having stored at least one instruction for being loaded by said processor and for performing the method steps according to any of claims 1 to 10.

14. An electronic device, comprising: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the method steps of any of claims 1 to 10.