CN113836275B

CN113836275B - Dialogue model establishment method and device, nonvolatile storage medium and electronic device

Info

Publication number: CN113836275B
Application number: CN202010514454.6A
Authority: CN
Inventors: 曹元斌; 吴胜兰; 殷浩; 邵明光; 唐攀攀; 白明智
Original assignee: Cainiao Smart Logistics Holding Ltd
Current assignee: Cainiao Smart Logistics Holding Ltd
Priority date: 2020-06-08
Filing date: 2020-06-08
Publication date: 2023-09-05
Anticipated expiration: 2040-06-08
Also published as: CN113836275A

Abstract

The invention discloses a dialogue model building method and device, a nonvolatile storage medium and an electronic device. Wherein the method comprises the following steps: generating a segmentation log by using a question-answer sequence, wherein the segmentation log comprises a plurality of question-answer pairs converted by the question-answer sequence; performing de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text; generating a merging log according to the vector text and the segmentation log; generating a session flow chart according to the combined log; and executing preset operation through the session flow chart. The dialogue model establishing method, the dialogue model establishing device, the non-volatile storage medium and the electronic device solve the technical problems that in the prior art, the efficiency of purely manually establishing a dialogue system is low and objectivity is not enough.

Description

Dialogue model establishment method and device, nonvolatile storage medium and electronic device

Technical Field

The invention relates to the field of intelligent learning models, in particular to a dialogue model building method and device, a nonvolatile storage medium and an electronic device.

Background

With the continuous development of intelligent learning models, learning models are applied in various fields such as business, military, production, life, etc. At present, in a customer service dialogue robot scene, a service scene needs to be accurately modeled in advance, and a model is modeled on how the robot processes service problems through dialogue so as to simulate a coping strategy of real customer service and solve the customer service problems.

The general customized conversation robot system for company business is the first step of work, the traditional method divides algorithm automatic extraction and manual construction into two major categories, the automatic extraction often needs a large amount of data, and because of lack of supervision and labeling, the conversation modeled in this way often has insufficient business subdivision or is too careful, important business nodes are ignored, or the user energy is wasted in no conversation, so the method cannot be well applied to real business scenes.

In view of the fact that purely automated solutions do not serve business well, the industry is more often using artificially constructed strategies, traditional manual construction, i.e. the way domain experts model, needs to go through the following process: firstly, on an existing customer service, all kinds of queries are processed by manual customer service all the time; at the beginning of the project, a service expert is used for combing out tasks which need to be processed by the robot, and combing the dialogue flow to arrange the service requirements; and the engineer constructs a dialogue system and an algorithm model according to the carded dialogue flow and business requirements and the historical accumulated manual customer service data, and then deploys online.

In the business expert's business modeling work, the business expert often needs to comb the dialogue log empirically, or manually, to complete the business modeling. There are two problems with this: firstly, when the business is huge, the work efficiency of business experts directly influences the progress of the whole project, and the efficiency is not likely to be improved by randomly adding one person due to the purely manual work and the need of expert knowledge; secondly, when a business expert works in a fixed certain field, the working center of gravity of the business expert exists, and the dialogue log quantity which can be combed in a limited time is limited, so that deviation between the business modeling of the business expert and a scene which needs to be covered by a real business scene is unavoidable, or a point which is relatively deviated is established in a heavy way, or an important point which needs to be covered is ignored, so that the following algorithm development deviates from the direction; the two points are the unavoidable problems of manual work.

In view of the above problems, no effective solution has been proposed at present.

Disclosure of Invention

The embodiment of the invention provides a dialogue model building method and device, a nonvolatile storage medium and an electronic device, which are used for at least solving the technical problems of low efficiency and objectivity of purely manual dialogue system building in the prior art.

According to an aspect of the embodiment of the present invention, there is provided a session model building method, including: generating a segmentation log by using a question-answer sequence, wherein the segmentation log comprises a plurality of question-answer pairs converted by the question-answer sequence; performing de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text; generating a merging log according to the vector text and the segmentation log; generating a session flow chart according to the combined log; and executing preset operation through the session flow chart.

Optionally, before the generating a segmentation log by using the question-answer sequence, where the segmentation log includes a plurality of question-answer pairs converted by the question-answer sequence, the method further includes: and acquiring the question-answer sequence.

Optionally, the performing the operations of de-duplication and vectorization on the text in the segmentation log to obtain a vector text includes: performing a de-duplication operation on the repeated dialogue text in the segmentation log; and carrying out text vectorization operation on the dialogue text after the de-duplication operation is carried out, and generating the vector text.

Optionally, the generating a merging log according to the vector text and the segmentation log includes: combining the vector text and the segmentation log to obtain a first combination result; clustering the first combined result to generate a clustering result; and generating the merging log according to the first merging result and the clustering result.

Optionally, the generating the merging log according to the merging result and the clustering result includes: combining the first combination result with the clustering result to obtain a second combination result; and generating the merging log according to the second merging result.

Optionally, the generating a session flow chart according to the combined log includes: acquiring the number of the same session in the combined log; sequencing the same session number according to a preset rule to obtain a sequencing result; and generating the session flow chart according to the sequencing result.

Optionally, the preset operation includes at least one of: sorting, merging and inserting.

Optionally, after the performing the preset operation through the session flow chart, the method further includes: and constructing a dialogue model according to the conversation flow chart.

According to another aspect of the embodiment of the present invention, there is also provided a method for establishing a physical distribution customer service session, applied to a physical distribution customer service terminal, including: generating a return goods segmentation log by using a return goods question-answering sequence, wherein the return goods segmentation log comprises a plurality of return goods question-answering pairs converted by the return goods question-answering sequence; performing de-duplication and vectorization operations on the text in the refund segmentation log to obtain a vector text; generating a combined log according to the vector text and the returned goods segmentation log; generating a return goods session flow chart according to the combined log; and executing preset operation through the return goods conversation flow chart.

According to another aspect of the embodiment of the present invention, there is also provided a method for applying a dialog model, including: obtaining dialogue model data; extracting training features according to the dialogue model data; and training a dialogue management model and an intention recognition model according to the training characteristics.

Optionally, the obtaining dialogue model data includes: obtaining a batch of dialogue models; and acquiring batch dialogue model data according to the batch dialogue model.

Optionally, after the training of the dialog management model and the intention recognition model according to the training features, the method further comprises: transmitting the trained dialogue management model and the trained intention recognition model to a dialogue system; and completing the construction of the dialogue system according to the dialogue management model and the intention recognition model.

According to another aspect of the embodiment of the present invention, there is also provided a satisfaction dialogue model application method, applied to customer service satisfaction feedback, including: acquiring satisfaction dialogue model data; extracting training features according to the satisfaction dialogue model data; training a dialogue management model and an intention recognition model according to the training characteristics; the satisfaction degree dialogue model data is formed by establishing the dialogue model establishing method.

According to another aspect of the embodiment of the present invention, there is also provided a session model building apparatus, including: the system comprises a segmentation log module, a segmentation log module and a query and answer processing module, wherein the segmentation log module is used for generating a segmentation log by utilizing a query and answer sequence, and the segmentation log comprises a plurality of query and answer pairs converted by the query and answer sequence; the vector text module is used for carrying out de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text; the merging log module is used for generating a merging log according to the vector text and the segmentation log; the generation module is used for generating a session flow chart according to the combined log; and the execution module is used for executing preset operation through the session flow chart.

Optionally, the apparatus further includes: and the acquisition module is used for acquiring the question-answer sequence.

Optionally, the vector text module includes: the duplicate removal unit is used for performing duplicate removal operation on the repeated dialogue text in the segmentation log; and the vector unit is used for carrying out text vectorization operation on the dialogue text after the de-duplication operation is carried out, and generating the vector text.

Optionally, the merge log module includes: the first merging unit is used for merging the vector text and the segmentation log to obtain a first merging result; the clustering unit is used for carrying out clustering operation on the first combined result to generate a clustering result; and the generation unit is used for generating the merging log according to the first merging result and the clustering result.

Optionally, the generating unit includes: the second merging unit is used for merging the first merging result and the clustering result to obtain a second merging result; the generating unit is further configured to generate the merge log according to the second merge result.

Optionally, the generating module includes: an obtaining unit, configured to obtain the number of identical sessions in the merge log; the ordering unit is used for ordering the same session number according to a preset rule to obtain an ordering result; and the generating unit is used for generating the session flow chart according to the sequencing result.

Optionally, the apparatus further includes: and the modeling module is used for constructing a dialogue model according to the conversation flow chart.

According to another aspect of the embodiment of the present invention, there is also provided a dialog model application apparatus, including: the acquisition module is used for acquiring dialogue model data; the extraction module is used for extracting training features according to the dialogue model data; and the training module is used for training a dialogue management model and an intention recognition model according to the training characteristics.

Optionally, the acquiring module includes: the acquisition unit is used for acquiring a batch of dialogue models; the acquisition unit is further used for acquiring batch conversation model data according to the batch conversation model.

Optionally, the apparatus further includes: the sending module is used for sending the trained dialogue management model and the trained intention recognition model to a dialogue system; and the construction module is used for completing the construction of the dialogue system according to the dialogue management model and the intention recognition model.

According to another aspect of the embodiment of the present invention, there is also provided a physical distribution customer service session establishment device applied to a physical distribution customer service terminal, including:

the system comprises a segmentation log module, a processing module and a processing module, wherein the segmentation log module is used for generating a returned goods segmentation log by utilizing a returned goods question-answer sequence, and the returned goods segmentation log comprises a plurality of returned goods question-answer pairs converted by the returned goods question-answer sequence;

the vector text module is used for carrying out duplication removal and vectorization operations on the text in the refund segmentation log to obtain a vector text;

the merging log module is used for generating a merging log according to the vector text and the refund segmentation log;

The generation module is used for generating a return goods session flow chart according to the combined log;

and the execution module is used for executing preset operation through the return goods conversation flow chart.

According to another aspect of the embodiment of the present invention, there is also provided a satisfaction session model application device, applied to customer service satisfaction feedback, including:

the acquisition module is used for acquiring satisfaction degree dialogue model data;

the extraction module is used for extracting training features according to the satisfaction dialogue model data;

the training module is used for training a dialogue management model and an intention recognition model according to the training characteristics;

wherein the satisfaction session model data is formed by the session model building device.

According to another aspect of embodiments of the present invention, there is also provided a computer program product comprising instructions which, when run on a computer, cause the computer to perform a dialog model building method.

According to another aspect of the embodiment of the present invention, there is further provided a nonvolatile storage medium, where the nonvolatile storage medium includes a stored program, and when the program runs, the device where the nonvolatile storage medium is controlled to execute a session model building method.

According to another aspect of an embodiment of the present application, there is provided an electronic device, including a processor and a memory; the memory stores computer readable instructions, and the processor is configured to execute the computer readable instructions, where the computer readable instructions execute a method for creating a dialog model.

In the embodiment of the application, a segmentation log is generated by using a question-answer sequence, wherein the segmentation log comprises a plurality of question-answer pairs converted by the question-answer sequence; performing de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text; generating a merging log according to the vector text and the segmentation log; generating a session flow chart according to the combined log; by means of the conversation flow chart executing the preset operation, the intelligent automatic conversation model building and manual conversation model building are achieved by means of text processing and vectorizing and model building, and the technical problems that in the prior art, the efficiency of purely manual conversation system building is low and objectivity is not achieved are solved.

Drawings

The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute a limitation on the application. In the drawings:

FIG. 1 is a flow chart of a dialog model building method according to an embodiment of the present invention;

FIG. 2 is a flow chart of a dialog model application method, according to an embodiment of the present invention;

FIG. 3 is a block diagram of a dialog model application apparatus in accordance with an embodiment of the present invention;

FIG. 4 is a block diagram of a dialog model application apparatus in accordance with an embodiment of the present invention;

FIG. 5 is a schematic diagram of a dialog model building method according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a dialog model application method, according to an embodiment of the present invention;

FIG. 7 is a flow chart of a method of establishing a physical distribution customer service session according to an embodiment of the present invention;

FIG. 8 is a flow chart of a method of satisfaction session model application according to an embodiment of the invention;

fig. 9 schematically shows a block diagram of a terminal device for performing the method according to the invention; and

fig. 10 schematically shows a memory unit for holding or carrying program code for implementing the method according to the invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

According to an embodiment of the present invention, there is provided a method embodiment of a dialog model building method, it being noted that the steps shown in the flowchart of the figures may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is shown in the flowchart, in some cases the steps shown or described may be performed in an order different from that shown or described herein.

Example 1

Fig. 1 is a flowchart of a dialog model building method according to an embodiment of the invention, as shown in fig. 1, the method comprising the steps of:

step S101, a segmentation log is generated by using a question-answer sequence, wherein the segmentation log comprises a plurality of question-answer pairs converted by the question-answer sequence.

In particular, the question-answer sequence may be a dialog log. The dialogue can be segmented and assembled, and the question-answer sequences of customer service and users under the same communication session are converted into question-answer pairs, namely, one record comprises one question-answer interaction of two continuous persons in one session and the number of dialogue rounds to be located, so that a segmentation log is obtained, wherein the segmentation log can be a segmentation log (2); for example, in one embodiment, the customer service and user question and answer sequences include the following:

q1: is the package mailer?

A1: no mail is wrapped.

Q2: is the Tibet area?

A2: can be used in Tibet areas.

Q3: is shipping a few days?

A3: and shipping within 24 hours.

The first round of sessions includes (Q1, A1), the second round of sessions includes (Q2, A2), and the third round of sessions includes (Q3, A3). The split log is used for recording the number of dialogue rounds in which all question-answer interactions are already located, such as the first round, the second round and the third round. The same communication session may refer to a conversation between one customer service ID and one user ID, or a conversation between one user ID and a different customer service ID for the same problem, or the like.

For example, when the question and answer sequence of the embodiment of the invention occurs on the Taobao application software, the dialogue between the Taobao customer service personnel and the customers using Taobao for consumption forms question and answer sequence data, and the dialogue model is built according to the question and answer sequence data generated on the Taobao application

The question-answer sequence may be obtained from the user's history data stored in the server. Or from an instant dialogue where the user interacts with the customer service.

The question-answer sequence is converted into question-answer pairs, and the question-answer sequence is split into multiple rounds of conversations. The segmentation log generated based on the question-answer sequence comprises question-answer pairs formed by the transformation of the question-answer sequence. Such as the first through third round of sessions previously described.

Specifically, since the embodiment of the present invention needs to process the question-answer sequence, before the segmentation log is generated, a server side or a client side of an instant dialogue needs to obtain a historical or real-time question-answer sequence. And sending a question-answer sequence acquisition request to a server storage area or a dialogue client through a processor of the dialogue system, calling a locally-existing question-answer sequence by the corresponding server storage area or dialogue client according to the acquisition request, and sending question-answer sequence data to the processor to wait for the next question-answer sequence processing.

Step S102, performing de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text.

Specifically, in order to construct the dialogue model in the embodiment of the invention, the segmentation log needs to be vectorized, the vectorization may be that the data unit in the segmentation log is vectorized, and the construction and analysis of the dialogue statement are performed according to the segmentation log after the whole vectorization, for example, after the segmentation log is vectorized, the processor may generate a set of required dialogue statements according to the vectorization segmentation log result, and process the dialogue statement according to the set.

Specifically, when the segmentation log vectorization is processed, the embodiment of the invention can extract all dialogue texts in the dialogue log, and perform duplicate removal and summarization to form summarized dialogue texts. Vectorizing the dialogue text, wherein technologies such as bert or word2vec can be tried, and a whole sentence of text is converted into a fixed-length vector to obtain a batch of text and a vector representation corresponding to the text; wherein, bert or word2vec are both deep learning model algorithms.

It should be noted that, for the dialogue log, all dialogue texts can be extracted, if necessary, a duplication removing operation can be performed, and if the log text itself does not need duplication removing, a vectorization operation can be directly performed to obtain a vector text corresponding to the segmentation log.

For example, the deduplication operation for a conversation log may be the following: the customer service representative's dialogue is "please confirm for package information. As the customer does not correctly input the confirmation information, the customer service representative continues to send out a dialogue please confirm the package information again. In the conversation of the customer service representative, please confirm the package information and please confirm the package information again, the repeated conversation records can be judged, and the meaning of the expression is that the customer is hoped to confirm the package information, and in the duplication removing operation, one of the two same conversations can be removed, so as to reduce the conversation log and improve the accuracy of conversation modeling.

And step S103, generating a merging log according to the vector text and the segmentation log.

Specifically, the vector text and the segmentation log are combined, and the first combination result may be that the segmentation log formed by the prepared question-answer pairs and the text vector generated just before are combined, and the vector representation of the question-answer pair and the vector representation of the answer are supplemented for each question-answer pair, so as to obtain the combination log. In the combined logs, each log record comprises a group of user questions and answers, the number of rounds of the user questions and answers in the questions and answers session of the user and customer service, and text vectors corresponding to two sentences of the group of questions and answers respectively; for example, in the log after merging, each log record may include the following information: dialogue segmentation log data between customer service representatives and clients and dialogue vector data in all segmentation logs.

It should be noted that, the first merging result is a result of merging vector data and segmentation log data, and the clustering operation on the result includes that text clustering can be performed on the question-answer text in each round according to different rounds, for example, a k-means algorithm is used to cluster the content similar to the semantic into one category, so as to obtain a clustering result of each round of dialogue. The k-means algorithm needs to specify the number k of clusters in advance, starts to randomly select k record points as center points, then traverses each record of the whole data set, classifies each record into a cluster where the center point closest to the record is located, replaces the previous center point with the average center point of the records of each cluster, and then iterates until convergence. K-Means is a common clustering algorithm, and compared with other clustering algorithms, the K-Means has low time complexity and good clustering effect.

And merging the logs which are just merged with the generated clustering results again, wherein each record after merging contains question-answer pairs, the session to which the question-answer pairs belong, the number of rounds in which the question-answer pairs are located, text vectors corresponding to each item of the question-answer pairs and clustering centers corresponding to the vectors.

Specifically, each of the records after merging may include, for example, the following: question-answer pair: for example, a question information and an answer information, i.e., question-answer pairs themselves; the session to which the question-answer pair belongs: for example in the form of a session number; number of rounds the question and answer pair is in: for example, what turn the question-answer pair is in the dialog. Wherein, the text vector corresponding to each item comprises: a text vector generated with each text correspondence, and a cluster center obtained from the text vector.

Step S104, generating a conversation flow chart according to the combined log.

Specifically, after the merged log is generated according to the clustering operation, the embodiment of the present invention needs to generate a session flow chart according to the merged log, where the session flow chart is used for modeling the dialog system in combination with manual work. The conversation flow chart can be a conversation list calculated and sequenced according to conversation frequency and used for representing which conversations are frequent and which conversations are not frequent, and the conversations are matched with corresponding answers, namely, answer pairs of clients and customer service representatives.

Specifically, in order to count and analyze dialogue data in the merged log, the same number of sessions in the merged log is acquired first, and the log may be counted according to the following criteria: the number of sessions clustered into the same question and the same answer under the same dialog turn is counted. After such statistics, all similar question-answer pairs are counted together at the same position of the conversation, and then the statistics result is inverted according to the conversation number to represent the current conversation position and is ranked according to a preset rule, wherein the preset rule can be a ranking method from high to low, and then the conversation system processor calculates which types of conversations can appear and the frequency of the conversations according to the statistics data and the ranking algorithm. With such statistics, a flow chart of the session can be constructed, the drawing flow chart being a directed graph, flowing from the previous round of questions to the next round. Finally, in an alternative embodiment, the long-tail low-frequency question-answer pair can be truncated to finally form a session stream, and the session stream is led into a business modeling system for a business expert modeling dialogue.

Step S105, executing a preset operation through the session flow chart.

Specifically, after the session flow chart is established in the session establishment method of the embodiment of the invention, the manual operation can perform the operations of adding, deleting and checking according to the session flow chart, and the better session model establishment result can be achieved by changing the session flow chart.

In particular, when too many service nodes are found to be clustered together, the clustering can be performed again. Different cluster nodes correspond to the same intent and may then merge. Alternatively, functional nodes may be inserted in the dialog flow. When the clustering based on the text is found, too many different service nodes are clustered together, a service expert can select and edit the nodes, designate the clustering number and enable the nodes to be clustered again so as to achieve the purpose of service node subdivision, for example, under the default clustering, "good", "can", "etc", "I want to wait" are clustered together, can select and be clustered again into two categories, "good", "can" be clustered into one category "," I want to wait "are clustered into one category, and for the condition that the service nodes cannot be distinguished, additional characteristics and rules can be added to distinguish;

In another case, if different cluster nodes are found to actually correspond to the same intention in the service, the two types can be selected to be combined, so that all the following child nodes in the tool are combined under the same parent node, for example, "how can you help you? The two nodes can be hooked and combined from the figure, so that the purpose of simplifying the flow is achieved.

Finally, functional nodes, such as "query order information", "query express mail location", etc., are inserted into the conversation flow to complete the conversation configuration. The service expert models the service scene dialogue flow by adding and deleting the counted dialogue flow, so that the method is efficient, important service nodes are not ignored because of blind spots of people, the importance of the service nodes can be judged according to the statistical indexes, and excessive energy is not wasted on insignificant nodes.

Specifically, according to the conversation flow chart generated in the process, the operation of manual addition, deletion and examination is carried out, a complete conversation model can be constructed, the model can play a role of automatic reply according to conversation topics or contents aiming at different application scenes, and the working efficiency in business application is improved.

Through the steps, the technical effect of objectively and efficiently establishing the dialogue system can be realized

Example two

Fig. 2 is a flowchart of a dialog model application method according to an embodiment of the present invention, as shown in fig. 2, the method including:

step S201, session model data is acquired.

Specifically, the dialogue model data may be obtained according to the foregoing embodiment, and the dialogue data and other contents in the dialogue model are extracted for the application of the dialogue model in this embodiment.

Specifically, since the number of dialogue models is more than one in any one application scenario, it is necessary to obtain dialogue models in batch through the dialogue system and analyze and process the dialogue models according to the data of the dialogue models. For example, when a user uses a naughty device to make online shopping, the user needs to ask and answer for a commodity circulation situation and a customer service representative, and then the user asks and answers for a commodity quality problem and the customer service representative, and the two types of questions and answers need to be implemented by two dialogue models, so that batch acquisition of the commodity circulation and quality dialogue models is needed when the dialogue models are acquired.

Step S202, extracting training features according to the dialogue model data;

specifically, in the foregoing embodiment, the business dialogue model is constructed through dialogue modeling, and the rules and cluster models obtained after construction may be referred to as dialogue flow models, for example, to distinguish different dialogue flows from logs. After the dialog flow model is obtained in this embodiment, the training features required for each model of dialog may be separated from the log in bulk. For example, in dialog management, it is necessary to train a dialog management model with jumps through dialog; in the intention recognition, it is necessary to determine different intentions of the user and data of the failed child node corresponding to each parent node in a round of dialogue.

Step S203, training a dialogue management model and an intention recognition model according to the training features.

Specifically, the extracted features are respectively used for training a dialogue management model and an intention recognition model, and the trained models are sent to a final dialogue system to complete the construction of the dialogue system. And then, respectively training the dialogue management model and the intention recognition model by the extracted features, and sending the trained models to a final dialogue system to complete the construction of the dialogue system. For example, two models have been trained, wherein a dialog management model is used to manage a dialog flow for a user, an intent recognition model is used to recognize the intent of the user, and a complete dialog flow may be generated by the dialog management model and the intent recognition model.

By the device, the technical effect of objectively and efficiently establishing the dialogue system can be realized.

Example III

Fig. 3 is a block diagram of a dialog model application apparatus according to an embodiment of the present invention, as shown in fig. 3, including:

the segmentation log module 301 is configured to generate a segmentation log by using a question-answer sequence, where the segmentation log includes a plurality of question-answer pairs converted from the question-answer sequence.

Q1: is the package mailer?

A1: no mail is wrapped.

Q2: is the Tibet area?

A2: can be used in Tibet areas.

Q3: is shipping a few days?

A3: and shipping within 24 hours.

And the vector text module 302 is configured to perform de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text.

And the merging log module 303 is configured to generate a merging log according to the vector text and the segmentation log.

Specifically, in order to obtain a combined log for modeling, the log just combined and the generated clustering result may be combined again, and each record after combination includes question-answer pairs, a session to which the question-answer pairs belong, the number of rounds in which the question-answer pairs are located, a text vector corresponding to each item of the question-answer pairs, and a clustering center corresponding to the text vector.

And the generating module 304 is configured to generate a session flow chart according to the merge log.

And the execution module 305 is configured to execute a preset operation through the session flow chart.

Optionally, a device for establishing a physical distribution customer service session is also provided, which is applied to a physical distribution customer service terminal for returning and changing goods, and includes:

Example IV

Fig. 4 is a block diagram of a dialog model application apparatus according to an embodiment of the present invention, as shown in fig. 4, including:

an acquisition module 401 is configured to acquire dialogue model data.

An extraction module 402, configured to extract training features according to the dialogue model data;

A training module 404 for training a dialogue management model and an intention recognition model according to the training features.

Optionally, the embodiment further proposes a satisfaction dialogue model application device, applied to customer service satisfaction feedback, including:

Example five

Fig. 5 is a schematic diagram of a method for creating a dialogue model according to an embodiment of the invention, as shown in fig. 5, the method includes:

the dialogue log contains a plurality of dialogue text data, the dialogue log is segmented and assembled to generate a segmented log, the dialogue text in the log is vectorized, then the data is combined according to the vectorized dialogue text and the segmented log generated by segmentation and assembly, the combined data is clustered, the clustered data and the combined data are combined again, finally the statistics of the number of identical terms of questions and answers is carried out according to the combined data after the combination of the two times, a directed graph is constructed, the directed graph is a conversation flow chart, a person who builds a dialogue model can carry out addition, deletion and correction according to the conversation flow chart, a final dialogue model is completed, and the method is applied to an operation tool, and the operation tool can be an online shopping tool such as a panning.

FIG. 6 is a schematic diagram of a method for applying a dialog model, as shown in FIG. 5, according to an embodiment of the invention, the method comprising: .

The dialogue log contains a plurality of characteristic values for training a dialogue management model and an intention recognition model, wherein the dialogue management model is used for controlling the answer accuracy and the answer efficiency of the robot customer service when answering user questions, the intention recognition model is used for recognizing what answer the user hopes to obtain, and the whole user communication service process, namely the construction of the whole dialogue system, can be completed according to the intention of the user and the dialogue management model. The feature value extraction needs to be extracted and called from a dialogue flow model established by a dialogue log, which parameters are specifically defined as feature values, the parameters need to be determined by a user according to an actual application scene, for example, when the user performs online shopping in a Taobao, the user can extract and define dialogue question-answer data in package aspect, dialogue question-answer data in cargo parameter aspect and the like in the dialogue flow model as feature values to train a dialogue management model and an intention recognition model, and a dialogue system is established.

Example six

Fig. 7 is a flowchart of a method for establishing a physical distribution customer service session according to an embodiment of the present invention, as shown in fig. 7, the method includes:

Step S701, generating a refund cut log by using a refund question-answer sequence, wherein the refund cut log comprises a plurality of refund question-answer pairs converted by the refund question-answer sequence.

Step S702, performing de-duplication and vectorization operations on the text in the refund segmentation log to obtain a vector text.

Step S703, generating a merge log according to the vector text and the refund segmentation log.

Step S704, generating a refund session flow chart according to the combined log.

Step S705, executing a preset operation through the refund session flowchart.

Specifically, the embodiment is used for performing a dialogue with a user by a customer service and performing a dialogue model establishment for a return shipment according to the method of the embodiment when the return shipment occurs, for example: the return exchange dialogue log contains several dialogue text data, such as: "ask: is the goods returned? The method comprises the steps of carrying out a first treatment on the surface of the Answering: can-! "ask: when is the return to stock? The method comprises the steps of carrying out a first treatment on the surface of the Answering: the present invention is for this evening-! "ask: is the return package mailer? The method comprises the steps of carrying out a first treatment on the surface of the Answering: of course package post-! And finally, counting the number of identical terms by questioning and answering according to the combined data after the twice combination, and constructing a directed graph, wherein the directed graph is a return goods conversation flow chart, namely, a person establishing a return goods conversation model can perform addition and subtraction check according to the conversation flow chart to complete a final return goods conversation model and is applied to an operation tool, such as an online shopping tool like a pannage.

Example seven

FIG. 8 is a flow chart of a satisfaction session model application method according to an embodiment of the invention, as shown in FIG. 8, comprising:

in step S801, satisfaction session model data is acquired.

Step S802, extracting training features according to the satisfaction dialogue model data.

Step S803, training a dialogue management model and an intention recognition model according to the training features.

Specifically, the embodiment is applied to training the existing dialogue model by taking the answer result of the customer service satisfaction degree of the user as the training value, so that the technical effect of improving the existing dialogue model through the feedback of the user is achieved. For example: the satisfaction dialogue log contains a plurality of characteristic values, which are used for training a dialogue management model and an intention recognition model, such as' query: is you satisfied? The method comprises the steps of carrying out a first treatment on the surface of the Answering: dissatisfaction-! "ask: where dissatisfaction? The method comprises the steps of carrying out a first treatment on the surface of the Answering: i want to return to the goods without someone else, feel ignored-! By way of "(the above-mentioned satisfaction dialogue log, there is satisfaction feedback about the user to the current service, such as unsatisfactory return of the goods. The dialogue management model is used for controlling the answer accuracy and the answer efficiency of the robot customer service when answering the user questions, the intention recognition model is used for recognizing what answer the user hopes to obtain, and the whole user communication service process, namely the construction of the whole dialogue system, can be completed according to the intention of the user and the dialogue management model. The satisfaction characteristic value extraction needs to be extracted and called from a dialogue flow model established by a satisfaction dialogue log, and specifically defines which satisfaction parameters are characteristic values, and needs to be determined by a merchant according to an actual application scene.

Specifically, the method comprises the following steps: generating a segmentation log by using a question-answer sequence, wherein the segmentation log comprises a plurality of question-answer pairs converted by the question-answer sequence; performing de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text; generating a merging log according to the vector text and the segmentation log; generating a session flow chart according to the combined log; and executing preset operation through the session flow chart.

According to another aspect of an embodiment of the present invention, there is provided an electronic device, including a processor and a memory; the memory stores computer readable instructions, and the processor is configured to execute the computer readable instructions, where the computer readable instructions execute a method for creating a dialog model.

By the method, the technical effect of objectively and efficiently establishing the dialogue system can be realized.

The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.

In the foregoing embodiments of the present invention, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.

Fig. 9 is a schematic hardware structure of a terminal device according to an embodiment of the present application. As shown in fig. 9, the terminal device may include an input device 90, a processor 91, an output device 92, a memory 93, and at least one communication bus 94. The communication bus 94 is used to enable communication connections between the elements. The memory 93 may comprise a high-speed RAM memory or may further comprise a non-volatile memory NVM, such as at least one magnetic disk memory, in which various programs may be stored for performing various processing functions and implementing the method steps of the present embodiment.

Alternatively, the processor 91 may be implemented as, for example, a central processing unit (Central Processing Unit, abbreviated as CPU), an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a controller, a microcontroller, a microprocessor, or other electronic components, and the processor 91 is coupled to the input device 90 and the output device 92 through wired or wireless connection.

Alternatively, the input device 90 may include a variety of input devices, for example, may include at least one of a user-oriented user interface, a device-oriented device interface, a programmable interface to software, a camera, and a sensor. Optionally, the device interface facing the device may be a wired interface for data transmission between devices, or may be a hardware insertion interface (such as a USB interface, a serial port, etc.) for data transmission between devices; alternatively, the user-oriented user interface may be, for example, a user-oriented control key, a voice input device for receiving voice input, and a touch-sensitive device (e.g., a touch screen, a touch pad, etc. having touch-sensitive functionality) for receiving user touch input by a user; optionally, the programmable interface of the software may be, for example, an entry for a user to edit or modify a program, for example, an input pin interface or an input interface of a chip, etc.; optionally, the transceiver may be a radio frequency transceiver chip, a baseband processing chip, a transceiver antenna, etc. with a communication function. An audio input device such as a microphone may receive voice data. The output device 92 may include a display, audio, etc.

In this embodiment, the processor of the terminal device may include functions for executing each module of the data processing apparatus in each device, and specific functions and technical effects may be referred to the above embodiments and are not described herein again.

Fig. 10 is a schematic hardware structure of a terminal device according to another embodiment of the present application. Fig. 10 is a diagram of one particular embodiment of the implementation of fig. 9. As shown in fig. 8, the terminal device of the present embodiment includes a processor 101 and a memory 102.

The processor 101 executes computer program code stored in the memory 102 to implement the methods of fig. 1-2 in the above-described embodiments.

The memory 102 is configured to store various types of data to support operations at the terminal device. Examples of such data include instructions for any application or method operating on the terminal device, such as messages, pictures, video, etc. The memory 102 may include a random access memory (random access memory, simply referred to as RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

Optionally, a processor 101 is provided in the processing assembly 100. The terminal device may further include: a communication component 103, a power supply component 104, a multimedia component 105, an audio component 106, an input/output interface 107 and/or a sensor component 108. The components and the like specifically included in the terminal device are set according to actual requirements, which are not limited in this embodiment.

The processing component 100 generally controls the overall operation of the terminal device. The processing assembly 100 may include one or more processors 101 to execute instructions to perform all or part of the steps of the methods described above. Further, the processing component 100 may include one or more modules that facilitate interactions between the processing component 100 and other components. For example, the processing component 100 may include a multimedia module to facilitate interaction between the multimedia component 105 and the processing component 100.

The power supply assembly 104 provides power to the various components of the terminal device. The power components 104 can include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power for the terminal devices.

The multimedia component 105 comprises a display screen between the terminal device and the user providing an output interface. In some embodiments, the display screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the display screen includes a touch panel, the display screen may be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensor may sense not only the boundary of a touch or slide action, but also the duration and pressure associated with the touch or slide operation.

The audio component 106 is configured to output and/or input audio signals. For example, the audio component 106 includes a Microphone (MIC) configured to receive external audio signals when the terminal device is in an operational mode, such as a speech recognition mode. The received audio signals may be further stored in the memory 102 or transmitted via the communication component 103. In some embodiments, the audio component 106 further comprises a speaker for outputting audio signals.

The input/output interface 107 provides an interface between the processing assembly 100 and peripheral interface modules, which may be click wheels, buttons, etc. These buttons may include, but are not limited to: volume button, start button and lock button.

The sensor assembly 108 includes one or more sensors for providing status assessment of various aspects for the terminal device. For example, the sensor assembly 108 may detect the open/closed state of the terminal device, the relative positioning of the assembly, the presence or absence of user contact with the terminal device. The sensor assembly 108 may include a proximity sensor configured to detect the presence of nearby objects in the absence of any physical contact, including detecting the distance between the user and the terminal device. In some embodiments, the sensor assembly 108 may also include a camera or the like.

The communication component 103 is configured to facilitate communication between the terminal device and other devices in a wired or wireless manner. The terminal device may access a wireless network based on a communication standard, such as WiFi,2G or 3G, or a combination thereof. In one embodiment, the terminal device may include a SIM card slot, where the SIM card slot is used to insert a SIM card, so that the terminal device may log into a GPRS network, and establish communication with a server through the internet.

From the above, the communication component 103, the audio component 106, the input/output interface 107, and the sensor component 108 in the embodiment of fig. 10 can be implemented as the input device in the embodiment of fig. 9.

In the several embodiments provided in the present application, it should be understood that the disclosed technology may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, for example, may be a logic function division, and may be implemented in another manner, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.

Claims

1. A method for building a dialog model, comprising:

generating a segmentation log by using a question-answer sequence, wherein the segmentation log comprises a plurality of question-answer pairs converted by the question-answer sequence;

performing de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text;

combining the vector text and the segmentation log to obtain a first combination result;

clustering the first combined result to generate a clustering result;

generating a merging log according to the first merging result and the clustering result;

generating a session flow chart according to the combined log;

executing preset operation through the session flow chart;

and constructing a dialogue model according to the conversation flow chart.

2. The method of claim 1, wherein prior to generating a segmentation log using the sequence of questions and answers, wherein the segmentation log comprises a number of question-answer pairs converted from the sequence of questions and answers, the method further comprises: and acquiring the question-answer sequence.

3. The method of claim 1, wherein performing the de-duplication and vectorization operations on the text in the segmentation log to obtain vector text comprises:

performing a de-duplication operation on the repeated dialogue text in the segmentation log;

and carrying out text vectorization operation on the dialogue text after the de-duplication operation is carried out, and generating the vector text.

4. The method of claim 1, wherein generating the merge log from the merge result and the cluster result comprises:

combining the first combination result with the clustering result to obtain a second combination result;

and generating the merging log according to the second merging result.

5. The method of claim 1, wherein generating a session flow diagram from the consolidated log comprises:

acquiring the number of the same session in the combined log;

sequencing the same session number according to a preset rule to obtain a sequencing result;

and generating the session flow chart according to the sequencing result.

6. The method of claim 1, wherein the preset operation comprises at least one of: sorting, merging and inserting.

7. The method for establishing the physical distribution customer service dialogue is applied to a physical distribution return customer service end and is characterized by comprising the following steps:

generating a return goods segmentation log by using a return goods question-answering sequence, wherein the return goods segmentation log comprises a plurality of return goods question-answering pairs converted by the return goods question-answering sequence;

performing de-duplication and vectorization operations on the text in the refund segmentation log to obtain a vector text;

combining the vector text and the return goods segmentation log to obtain a first combination result;

clustering the first combined result to generate a clustering result;

generating a return goods session flow chart according to the combined log;

executing preset operation through the goods returning session flow chart;

and constructing a stock return dialogue model according to the stock return dialogue flow chart.

8. A method for applying a dialog model, comprising:

obtaining dialogue model data;

extracting training features according to the dialogue model data;

training a dialogue management model and an intention recognition model according to the training characteristics;

Wherein the dialogue model data is created by the dialogue model creation method according to any one of claims 1 to 6.

9. The method of claim 8, wherein the obtaining dialog model data comprises:

obtaining a batch of dialogue models;

and acquiring batch dialogue model data according to the batch dialogue model.

10. The method of claim 8, wherein after the training of the dialog management model and the intent recognition model based on the training features, the method further comprises:

transmitting the trained dialogue management model and the trained intention recognition model to a dialogue system;

and completing the construction of the dialogue system according to the dialogue management model and the intention recognition model.

11. The satisfaction dialogue model application method is applied to customer service satisfaction feedback and is characterized by comprising the following steps:

acquiring satisfaction dialogue model data;

extracting training features according to the satisfaction dialogue model data;

wherein the satisfaction session model data is created by the session model creation method of any one of claims 1 to 6.

12. A dialog model building device, comprising:

the system comprises a segmentation log module, a segmentation log module and a query and answer processing module, wherein the segmentation log module is used for generating a segmentation log by utilizing a query and answer sequence, and the segmentation log comprises a plurality of query and answer pairs converted by the query and answer sequence;

the vector text module is used for carrying out de-duplication and vectorization operations on the text in the segmentation log to obtain a vector text;

the merging log module is used for generating a merging log according to the vector text and the segmentation log;

the generation module is used for generating a session flow chart according to the combined log;

the execution module is used for executing preset operation through the session flow chart;

the merge log module includes:

the first merging unit is used for merging the vector text and the segmentation log to obtain a first merging result;

the clustering unit is used for carrying out clustering operation on the first combined result to generate a clustering result;

the generation unit is used for generating the merging log according to the first merging result and the clustering result;

and the modeling module is used for constructing a dialogue model according to the conversation flow chart.

13. The apparatus of claim 12, wherein the apparatus further comprises: and the acquisition module is used for acquiring the question-answer sequence.

14. The apparatus of claim 12, wherein the vector text module comprises:

the duplicate removal unit is used for performing duplicate removal operation on the repeated dialogue text in the segmentation log;

and the vector unit is used for carrying out text vectorization operation on the dialogue text after the de-duplication operation is carried out, and generating the vector text.

15. The apparatus of claim 12, wherein the generating unit comprises:

the second merging unit is used for merging the first merging result and the clustering result to obtain a second merging result;

the generating unit is further configured to generate the merge log according to the second merge result.

16. The apparatus of claim 12, wherein the generating module comprises: an obtaining unit, configured to obtain the number of identical sessions in the merge log;

the ordering unit is used for ordering the same session number according to a preset rule to obtain an ordering result;

and the generating unit is used for generating the session flow chart according to the sequencing result.

17. The apparatus of claim 12, wherein the preset operation comprises at least one of: sorting, merging and inserting.

18. The utility model provides a commodity circulation customer service dialogue establishment device is applied to commodity circulation and moves back goods customer service end which characterized in that includes:

the execution module is used for executing preset operation through the return goods exchange session flow chart;

the merge log module includes:

the first merging unit is used for merging the vector text and the refund segmentation log to obtain a first merging result;

And the modeling module is used for constructing a stock return dialogue model according to the session flow chart.

19. A dialog model application apparatus, comprising:

the acquisition module is used for acquiring dialogue model data;

the extraction module is used for extracting training features according to the dialogue model data;

the session model data is session model data created by the session model creation means according to claims 12-17.

20. The apparatus of claim 19, wherein the acquisition module comprises:

the acquisition unit is used for acquiring a batch of dialogue models;

the acquisition unit is further used for acquiring batch conversation model data according to the batch conversation model.

21. The apparatus of claim 19, wherein the apparatus further comprises:

the sending module is used for sending the trained dialogue management model and the trained intention recognition model to a dialogue system;

and the construction module is used for completing the construction of the dialogue system according to the dialogue management model and the intention recognition model.

22. A satisfaction dialogue model application device applied to customer service satisfaction feedback, comprising:

wherein the satisfaction session model data is created by the session model creation means of any of claims 12-17.

23. A non-volatile storage medium, characterized in that the non-volatile storage medium comprises a stored program, wherein the program, when run, controls a device in which the non-volatile storage medium is located to perform the method of any one of claims 1 to 11.

24. An electronic device comprising a processor and a memory; the memory has stored therein computer readable instructions for executing the computer readable instructions, wherein the computer readable instructions when executed perform the method of any of claims 1 to 11.