CN117744801A - Conversation method, device and equipment of psychology big model based on artificial intelligence - Google Patents
Conversation method, device and equipment of psychology big model based on artificial intelligence Download PDFInfo
- Publication number
- CN117744801A CN117744801A CN202311817029.4A CN202311817029A CN117744801A CN 117744801 A CN117744801 A CN 117744801A CN 202311817029 A CN202311817029 A CN 202311817029A CN 117744801 A CN117744801 A CN 117744801A
- Authority
- CN
- China
- Prior art keywords
- model
- psychological
- input information
- current input
- scene
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 22
- 230000004044 response Effects 0.000 claims abstract description 58
- 238000004590 computer program Methods 0.000 claims description 13
- 238000005516 engineering process Methods 0.000 description 13
- 238000012549 training Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 230000008451 emotion Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 230000014509 gene expression Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000000306 recurrent effect Effects 0.000 description 4
- 230000003993 interaction Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000009223 counseling Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002996 emotional effect Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000004984 smart glass Substances 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The application discloses a conversation method, device and equipment of a psychology big model based on artificial intelligence. The psychological large model comprises a scene recognition model, a psychological strategy response model and a generated large model, and the method comprises the steps of obtaining current input information; identifying a psychological scene of the current input information based on the scene identification model; inputting the psychological scene and the current input information to the psychological strategy response model to generate a first prompt word under the condition that the psychological scene is identified; and inputting the current input information and the first prompt word into the generation type large model to generate target reply information. The user's input may be psychologically scene identified and then a prompt word corresponding to the reply strategy generated based on the hit psychology scene, such that the generated large model generates a professional reply that is more psychologically conformable based on the prompt word and the current input information.
Description
Technical Field
The present application relates to the field of large model technology, and more particularly, to a dialogue method, dialogue apparatus, computer device, and non-volatile computer-readable storage medium for an artificial intelligence-based psychology large model.
Background
After the large model is formed, the question answering mode of the large model tends to answer questions in a one-to-one mode, and the specific teaching (i.e. finishing) is not performed. It is difficult to conduct multiple rounds of conversations, which by default will give knowledge once, completing the round of conversations.
In the prior art, psychological problems are typically performed manually by psychological practitioners (e.g., psychological doctors, psychological consultants, etc.). However, the number of psychological practitioners is small and not all users with demand can be satisfied in time.
Therefore, how to provide a large model for automatically and professionally aiming at psychological problems to meet the consultation requirement of psychological problems is a technical problem to be solved urgently by those skilled in the art.
Disclosure of Invention
The embodiment of the application provides a conversation method, a conversation device, computer equipment and a non-volatile computer readable storage medium of a psychology big model based on artificial intelligence, which can be used for identifying psychology scenes of user input and then generating prompt words corresponding to a reply strategy based on hit psychology scenes, so that the generated big model generates professional replies which are more fit with psychology based on the prompt words and current input information.
The dialogue method based on the artificial intelligence psychology big model comprises the steps of obtaining current input information; identifying a psychological scene of the current input information based on the scene identification model; inputting the psychological scene and the current input information to the psychological strategy response model to generate a first prompt word under the condition that the psychological scene is identified; and inputting the current input information and the first prompt word into the generation type large model to generate target reply information.
In certain embodiments, the psychology-big model further comprises a generic strategy response model, the method further comprising:
inputting the field current input information to the general strategy response model to generate a second prompt word under the condition that the psychological scene is not recognized;
and inputting the current input information and the second prompt word into the large generation model to generate intermediate dialogue information.
In certain embodiments, the psychographic large model further comprises an intent recognition model, the method further comprising:
identifying an intent of the current input information based on the intent identification model;
said inputting the field current input information to the generic policy response model to generate a second cue word without identifying the psychological scene, comprising:
in the event that the psychological scenario is not identified, the current input information and the intent are input to the generic policy response model to generate the second cue word.
In some implementations, the inputting the current input information and the intent to the generic policy response model to generate the second hint word includes:
the current input information, the intention and the historical dialog memory are input into a preset general strategy response model to generate a second prompt word, wherein the historical dialog memory comprises at least one of historical dialog information and historical intention.
In some embodiments, the psychological policy response model and the general policy response model are both classification models, the psychological policy response model is established according to a preset psychological scene and a corresponding preset prompt word, and the general policy response model is established according to preset input information and a corresponding preset prompt word.
In some embodiments, the inputting the current input information and the first prompt word into the generative large model to generate target reply information includes:
and inputting the current input information, the first prompt word and a history dialogue memory to a preset generation type large model so as to generate the target reply information, wherein the history dialogue memory comprises one or more input information before the current input information.
In some embodiments, the obtaining the current input information includes:
and receiving text information and/or voice information input by a user to acquire the current input information.
The dialogue apparatus of the present embodiment is applied to an artificial intelligence-based psychology big model including a scene recognition model, a psychological strategy response model, and a generative big model, and the apparatus includes:
the acquisition module is used for acquiring current input information;
the identification module is used for identifying the psychological scene of the current input information based on the scene identification model;
the prompt generation module is used for inputting the psychological scene and the current input information into the psychological strategy response model under the condition that the psychological scene is identified so as to generate a first prompt word;
and the reply module is used for inputting the current input information and the first prompt word into the generation type large model so as to generate target reply information.
The computer device of the present application embodiment includes a processor, a memory, and a computer program, where the computer program is stored in the memory and executed by the processor, and the computer program includes instructions for executing the dialogue method of any of the foregoing embodiments.
The non-transitory computer readable storage medium of the present embodiments includes a computer program that, when executed by a processor, causes the processor to perform the dialogue method of any of the above embodiments.
According to the conversation method, the conversation device, the computer equipment and the computer readable storage medium based on the artificial intelligence psychological big model, after the current input information of the user is obtained, whether the input information of the user is one of preset psychological scenes is firstly identified based on the preset scene identification model, after the psychological scenes are identified, prompt words corresponding to the psychological scenes can be generated according to the psychological policy response model, and the generated big model is different in response to the same problem under the intervention of different prompt words, so that professional psychological responses matched with the identified psychological scenes can be output through inputting the prompt words and the current input information into the generated big model, and the consultation requirement of the user on the psychological problems is met.
Additional aspects and advantages of embodiments of the application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of embodiments of the application.
Drawings
The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a schematic illustration of an application scenario of a dialog method of some embodiments of the present application;
FIG. 2 is a flow diagram of a dialog method of some embodiments of the present application;
FIG. 3 is a flow diagram of a dialog method of some embodiments of the present application;
FIG. 4 is a flow diagram of a dialog method of some embodiments of the present application;
FIG. 5 is a flow diagram of a dialog method of some embodiments of the present application;
FIG. 6 is a block diagram of a dialog device according to some embodiments of the present application;
FIG. 7 is a schematic structural diagram of a computer device according to some embodiments of the present application;
FIG. 8 is a schematic diagram of a connection state of a non-volatile computer readable storage medium and a processor according to some embodiments of the present application.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are exemplary only for explaining the embodiments of the present application and are not to be construed as limiting the embodiments of the present application.
To facilitate an understanding of the present application, the following description of terms appearing in the present application will be provided:
1. artificial intelligence (Artificial Intelligence, AI), is a theory, method, technique, and application system that simulates, extends, and extends human intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, obtains knowledge, and uses the knowledge to obtain optimal results. In other words, artificial intelligence is an integrated technology of computer science that attempts to understand the essence of intelligence and to produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence, i.e. research on design principles and implementation methods of various intelligent machines, enables the machines to have functions of sensing, reasoning and decision. The artificial intelligence technology is a comprehensive subject, and relates to the technology with wide fields, namely the technology with a hardware level and the technology with a software level. Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions. The technical scheme provided by the embodiment of the application mainly relates to natural language processing technology in artificial intelligence, machine learning/deep learning and the like.
2. Large model: the main use mode is a mode for generating answers to questions of users through characters based on a large-scale generation model of a large amount of data training. Because of the huge training data and model parameters, the knowledge of each field can be answered to answer the related questions. The large model currently supports active intervention of its generation content by writing a prompt, i.e. different answers appear to different shortts for the same question.
Before the large model shows extremely strong knowledge, the psychological class of conversational robots is basically 2 types (such as type a or type B, or a and B in combination (i.e. a cannot deal with the problem to B, and then returns to a continuous flow)):
type a: task-type robots. The robot collects the information of the user through dialogue according to fixed logic design, the questions and the answers are relatively fixed, the dialogue process is more like that of questionnaire investigation, and a designed form is filled up. The problem with this approach is that the dialog logic is too fixed and it is difficult to respond correctly when the user's problem is out of recognition range. Meanwhile, the robot is not suitable for talking with multiple times, and even if the robot is chatting on different topics, the user can feel the robot to be hard to chatting.
Type B: the boring robot through finishing. The chat robot generally throws the collected dialogue corpus into a model for training, and the model has the dialogue simulation function. If the corpus given to the model is a psychology-related dialogue, the model is a chat robot capable of performing psychology-based dialogue. finishing refers to training the dialogue ability of a model through a large amount of data, and then adding a psychological correlation dialogue to do the enhanced training of the class. The method has the problem that the chat content and trend are difficult to control, and the effect is completely determined by the corpus during training.
In summary, after the large model is formed, although the model itself carries knowledge in the psychological aspect, the large model answers questions in a more aggressive manner, without specific teaching (i.e., finishing). It is difficult to conduct multiple rounds of conversations, which by default will give knowledge once, completing the round of conversations.
Thus, for the same problem, the large model may give a general reply or give a reply in one of the fields, and it may be difficult to meet the requirement of the user for consulting the psychological problem, and accurately make a professional reply of the psychology.
In order to solve the above technical problems, an embodiment of the present application provides a dialogue method.
An application scenario of the technical solution of the present application is described first, as shown in fig. 1, which is a schematic application scenario of a dialogue method provided in an embodiment of the present application, where the application scenario relates to a terminal device 110 and a server 120, and the terminal device 110 may communicate with the server 120.
Fig. 1 illustrates one terminal device 110 and one server 120 as an example, and may include other numbers of terminal devices and servers in practice, which are not limited in this embodiment of the present application.
In some implementations, the server 120 in fig. 1 may be an independent physical server, or may be a server cluster or a distributed system formed by a plurality of physical servers, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, and basic cloud computing services such as big data and artificial intelligence platforms. The embodiments of the present application are not limited in this regard.
In some implementations, the terminal device 110 as shown in fig. 1 may be installed with an application client that, when running in the terminal device 110, may interact with the server 120. The client may specifically include, for example: psychological-related applications, browser clients, vehicle-mounted clients, smart home clients, game clients, multimedia clients (e.g., video clients), social clients, and information-based clients (e.g., news clients), all of which can run psychological counseling services.
Optionally, in the embodiment of the present application, the terminal device 110 may be a device with rich man-machine interaction modes, having capability of accessing to the internet, generally carrying various operating systems, and having a relatively strong processing capability. The terminal device 110 may be a smart phone, smart glasses, a handheld terminal, a smart television, a tablet computer, a vehicle-mounted terminal, etc., but is not limited thereto.
In one implementation manner, the server 120 and the terminal device 110 may perform the session method provided in the embodiments of the present application in an interactive manner, or the terminal device 110 or the server 120 may perform the session method provided in the embodiments of the present application.
The dialogue method of the present application will be described in detail below:
referring to fig. 2, an embodiment of the present application provides a dialogue method based on an artificial intelligence psychological large model, where the psychological large model includes a scene recognition model, a psychological strategy response model, and a generation type large model, and the dialogue method includes:
step 011: acquiring current input information;
specifically, the user may input the current input information through voice input, text box input. If the user opens the psychological consultation service, the voice of the psychological problem to be consulted is input through speaking, and the voice is converted into a text through voice recognition, wherein the text is the current input information. For another example, after the user opens the psychological consultation service, a text is input through a preset text box, and the text is the current input information.
Optionally, since the user may not be able to clearly recognize the psychological questions to be consulted by himself/herself through text, the psychological consultation service may also provide a plurality of preset psychological questions, and the user may select one of the psychological questions as the current input information through the selection operation, thereby inputting more accurate psychological questions.
Step 012: based on the scene recognition model, recognizing a psychological scene of the current input information;
the scene recognition model is a model for recognizing a psychological scene to which the input information belongs. The scene recognition model is constructed based on a large number of expected and preset psychological scenes, and can accurately recognize the psychological scene of the psychological problem of the user through input information, so that the subsequent generation type large model is assisted to accurately reply.
Wherein, each psychological scene is arranged and generalized by professional psychologists.
Alternatively, the scene recognition model may be a BERT model (all Bidirectional Encoder Representations from Transformers).
Specifically, after the current input information is acquired, the current input information is input into a scene recognition model, and the psychological scene of the current input information is recognized through the scene recognition model, so that the psychological scene of the current input information is output.
As shown in table 1 below, for psychological scenarios corresponding to different input information, it is understood that table 1 is only an example, and not all psychological scenarios are listed.
| Currently entered information | Psychological scenes |
| Input information 1 | Scene 1 |
| Input information 2 | Scene 2 |
TABLE 1
Step 013: inputting the psychological scene and the current input information into a psychological strategy response model under the condition that the psychological scene is identified so as to generate a first prompt word;
the psychological strategy response model is a model for generating corresponding prompt words according to psychological scenes and input information. The psychological strategy model can be established according to preset prompt words which correspond to the psychological scene and the user information together, and the prompt words are different from different user information under the same psychological scene; under the same user information, different psychological scenes and prompt words are different.
Each psychological scene, user information and corresponding preset prompt words are sorted and summarized by professional psychologists, the professionality is high, the generation accuracy of the prompt words can be improved, and the follow-up large model can generate more professional psychological replies.
Alternatively, the psychological policy response model may be a recurrent neural network (Gated Recurrent Unit), GRU, model.
Alternatively, the psychological policy response model may also be a knowledge graph. Knowledge maps are essentially a knowledge base of the Semantic Network (Semantic Network).
Specifically, in the case that the psychological scene is identified by the scene identification model, the psychological policy response model may generate the first prompt word according to the psychological scene and the current input information.
When the first prompting word is generated, the psychological strategy response model firstly acquires user information from the current input information, and then determines the final first prompting word according to the psychological scene and the user information.
Examples of hints corresponding to different physiological scenarios and different user information are shown in tables 2 and 3 below.
It will be appreciated that different psychological scenarios, corresponding user information, different psychological scenarios and user information, and corresponding prompt words are also different.
TABLE 2
TABLE 3 Table 3
Step 014: and inputting the current input information and the first prompt word into the generation type large model to generate target reply information.
Specifically, after the first prompt word is obtained, the current input information and the first prompt word can be input into the generation type large model, and the generation type large model fully considers the first prompt word when replying, so that psychological professional replying meeting the requirement of the first prompt word is given when replying to the current input information.
In this way, by accurately identifying the psychological scene and combining the user information in the current input information, an accurate first prompt word is generated to intervene in the reply of the generated large model to the current input information, so that the professional psychological reply which meets the psychological problem consultation requirement of the user is generated.
According to the conversation method based on the artificial intelligence psychological large model, after the current input information of the user is obtained, whether the input information of the user is one of preset psychological scenes is firstly identified based on the preset scene identification model, after the psychological scenes are identified, prompt words corresponding to the psychological scenes can be generated according to the psychological policy response model, and the generated large model replies different questions under the intervention of different prompt words, so that professional psychological replies matched with the identified psychological scenes can be output through inputting the prompt words and the current input information into the generated large model, and the consultation requirement of the user on the psychological questions is met.
Referring to fig. 3, in some embodiments, the psychology-big model further comprises a generic strategy response model, the method further comprising:
step 015: under the condition that the psychological scene is not recognized, inputting current field input information into a general strategy response model to generate a second prompt word;
step 016: the current input information and the second prompt word are input to the generative large model to generate intermediate dialogue information.
The general strategy response model is a model for generating corresponding prompt words according to input information. The general strategy response model is mainly used for ensuring smooth conversation, enriching the diversity of reply contents and mining the requirements of users. The general strategy response model can be obtained through training according to a large amount of input information and corresponding prompt words.
The general strategy response model can also memorize a large amount of real dialogue corpus during training, and the whole execution process predicts the most probable reply logic of the next round through the trained model according to the input content.
Alternatively, the generic policy response model may be a recurrent neural network (Gated Recurrent Unit), GRU, model.
In particular, it may be difficult for a user to input accurate and complete information in the current input information at one time based on the difference in expressive power, so that the scene recognition model may not recognize the psychological scene.
At this time, in order to ensure that the dialogue normally proceeds and that the psychological scene can be identified later, the second prompt word can be generated through the general strategy response model, then the second prompt word and the current input information are input into the large generation model, so that intermediate dialogue information is generated, the intermediate dialogue information at this time is generally a reply of interaction with the user, and the user is guided to input more information, so that the psychological scene can be identified conveniently.
For example, when the current input information relates to emotion expressions, a prompt word of "open question for emotion" may be generated: thereby generating an intermediate reply message inviting the principal to clarify his or her emotion, knowing the principal's true perception using an open query.
For another example, when the current input information relates to a viewpoint expression, a prompt word of "open question for ideas" may be generated, thereby generating intermediate reply information inviting the exploring party's ideas.
For another example, when the current input information relates to emotion expression a plurality of times, the high probability can determine that the emotion of the user is quite unstable, a prompt word for "pacifying" can be generated, and intermediate reply information capable of pacifying the emotion of the user is generated through the prompt word for "pacifying".
Optionally, inputting the current input information and the second prompt word into a preset generation type big model to generate the intermediate dialogue information includes: the current input information, the second prompt word and the history dialogue memory are input into a preset generation type large model to generate intermediate dialogue information, and the history dialogue memory comprises the history dialogue information.
Specifically, the psychological consultation of the user can give the final target reply information only by one round, multiple rounds of dialogue may be needed, and the final target reply information is regenerated after more sufficient information is obtained.
Thus, the relevant information of each input information of the user can be recorded, so that a history dialogue memory is generated, and the history dialogue information can comprise one or more input information before the current input information and can also comprise reply information corresponding to each input information.
When the generated large model replies the middle reply information, the current input information, the second prompt word and the historical dialogue information are considered at the same time, so that more accurate reply is performed, repeated replies can be avoided, and if the information already given in the previous reply information is not replied.
Alternatively, the historical dialog information may include the first N rounds of dialog information. Wherein, N can be determined according to the reply efficiency and the storage limit of the psychological consultation service, and the faster the reply efficiency is, the smaller N is, and the smaller the storage is, the smaller N is.
Optionally, when the psychological scene is identified, if the current input information or the history dialogue memory lacks the user information required by the psychological scene, the target reply information for inquiring the user information needs to be generated first, so that the user information is obtained by inquiring the user, and then the final professional psychological reply is generated according to the psychological scene and the corresponding user information.
Referring to fig. 4, in some embodiments, the psychographic large model further includes an intent recognition model, the method further comprising:
step 017: identifying the intention of the current input information based on the intention identification model;
step 015: in the event that no psychological scenario is identified, inputting field current input information to the generic strategic response model to generate a second cue word, comprising:
step 0151: in the event that no psychological scenario is identified, current input information and intent are input to the generic policy response model to generate a second cue word.
Wherein the intention recognition model is a model that recognizes the intention of the user in the current input information. The intention recognition model is constructed based on a large number of expected and preset various intentions, and the intention recognition model can accurately recognize the intention of a user through input information, so that the follow-up generation type large model is assisted to accurately reply.
Specifically, the intention recognition is carried out on the current input information through the intention recognition model, and the recognized intention is combined with the current input information and is input into the general strategy response model, so that the accuracy of the second prompt word generated by the general strategy response model is improved.
As shown in table 4 below, examples of intents corresponding to different input information.
| Inputting information | Intent of |
| Input information 1 | Intent 1 |
| Input information 2 | Intent 2 |
TABLE 4 Table 4
Optionally, inputting the current input information and the intent to the generic policy response model to generate a second hint word includes:
the current input information, the intention and the historical dialog memory are input into a preset general strategy response model to generate a second prompt word, and the historical dialog memory comprises at least one of the historical dialog information and the historical intention.
Specifically, when the second prompt word is generated, the history dialogue memory may be considered to obtain a more accurate second prompt word.
The historical dialog memory may include at least one of both historical dialog information and historical intent corresponding to each of the previous input information.
When the general strategy response model generates the second prompt word, the current input information, the current intention and the history dialogue memory are simultaneously considered, so that the second prompt word is generated more accurately.
The user's sentences for the previous rounds have a large number of emotional expressions, while the most recent cue words have no "pacifying", which predicts a higher probability of pacifying. If the user is stating something all the time in the previous rounds, the probability of predicting an open question for emotion is higher.
Alternatively, the historical dialog memory may include the first N rounds of dialog information and the first M historical intents. Wherein, N and M can be determined together according to the reply efficiency and the storage limit of the psychological consultation service, and the faster the reply efficiency is, the smaller N and M are, the smaller the storage is, and the smaller N and M are.
For example, the history dialogue memory may record the first 6 rounds of dialogue information (e.g., only input information, or reply information containing input information and input information correspondence), whereas the history dialogue memory may record the history intentions corresponding to the first 20 input information because the intent information is generally less in content.
Referring to fig. 5, in certain embodiments, step 014: inputting the current input information and the first prompt word into the generation type big model to generate target reply information, wherein the method comprises the following steps:
step 0141: the current input information, the first prompt word and the history dialogue memory are input into a preset generation type big model to generate target reply information, and the history dialogue memory comprises one or more input information before the current input information.
Specifically, the psychological consultation of the user can give the final target reply information only by one round, multiple rounds of dialogue may be needed, and the final target reply information is regenerated after more sufficient information is obtained.
Thus, the relevant information of each input information of the user can be recorded, so that a history dialogue memory is generated, and the history dialogue information can comprise one or more input information before the current input information and can also comprise reply information corresponding to each input information.
When the generated large model replies the target reply information, the current input information, the first prompt word and the historical dialogue information are considered at the same time, so that more accurate reply is performed, repeated reply can be avoided, and if the information already given in the previous reply information is not replied.
Alternatively, the historical dialog information may include the first N rounds of dialog information. Wherein, N can be determined according to the reply efficiency and the storage limit of the psychological consultation service, and the faster the reply efficiency is, the smaller N is, and the smaller the storage is, the smaller N is.
Referring to fig. 6, to facilitate better implementation of the dialogue method of the present embodiment, the present embodiment further provides a dialogue device 10, which is applied to an artificial intelligence-based psychology big model, wherein the psychology big model includes a scene recognition model, a psychological strategy response model, and a generative big model. The dialogue device 10 comprises an acquisition module 11 for acquiring current input information; the identification module 12 is used for identifying the psychological scene of the current input information based on the scene identification model; the prompt generation module 13 is configured to input the psychological scene and the current input information to the psychological policy response model to generate a first prompt word when the psychological scene is identified; the reply module 14 is used for inputting the current input information and the first prompt word into the generative large model to generate target reply information.
The prompt generation module 13 is further configured to input field current input information to the universal policy response model to generate a second prompt word if the psychological scene is not identified; the current input information and the second prompt word are input to the generative large model to generate intermediate dialogue information.
The recognition module 15 is also used to recognize the intention of the current input information based on the intention recognition model. The hint generation module 13 is further configured to input the current input information and the intention to the generic policy response model to generate a second hint word if the psychological scene is not identified.
The prompt generation module 13 is specifically configured to input current input information, intention, and history dialogue memory to a preset general policy response model, where the history dialogue memory includes at least one of history dialogue information and history intention, to generate a second prompt word.
The prompt generation module 13 is specifically configured to input the current input information, the first prompt word, and the history dialogue memory to a preset generation type large model, so as to generate the target reply information, where the history dialogue memory includes one or more input information before the current input information.
The obtaining module 11 is configured to receive text information and/or voice information input by a user, so as to obtain current input information.
The dialog device 10 has been described above in connection with the figures from the point of view of functional modules, which can be realized in hardware, instructions in software, or a combination of hardware and software modules. Specifically, each step of the method embodiments in the embodiments of the present application may be implemented by an integrated logic circuit of hardware in a processor and/or an instruction in software form, and the steps of the method disclosed in connection with the embodiments of the present application may be directly implemented as a hardware encoding processor or implemented by a combination of hardware and software modules in the encoding processor. Alternatively, the software modules may be located in a well-established storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, and the like. The storage medium is located in a memory, and the processor reads information in the memory, and in combination with hardware, performs the steps in the above method embodiments.
Referring to fig. 7, a computer device 100 of an embodiment of the present application includes a processor 20, a memory 30, and a computer program, wherein the computer program is stored in the memory 30 and executed by the processor 20, and the computer program includes instructions for executing the dialogue method of any of the embodiments described above.
Alternatively, the computer device 100 may be any device having image processing capabilities, such as a server or terminal device (e.g., a cell phone, tablet, display device, notebook, smart watch, head display device, game console, etc.).
Alternatively, the computer device 100 may be a combined system of a terminal device and a server, where the terminal device is used to obtain current input information, and the psychology big model may be deployed to at least one of the terminal device and the server, for example, the psychology big model is deployed to the server. After the terminal equipment acquires the current input information, the current input information is sent to the server, and then the psychological big model in the server processes the current input information, outputs target reply information and sends the target reply information to the terminal equipment for display.
Referring to fig. 8, the embodiment of the present application further provides a computer readable storage medium 300, on which a computer program 310 is stored, where the computer program 310, when executed by the processor 320, implements the steps of the dialogue method of any of the above embodiments, which is not described herein for brevity.
In the description of the present specification, reference to the terms "certain embodiments," "in one example," "illustratively," and the like, means that a particular feature, structure, material, or characteristic described in connection with the embodiments or examples is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and further implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present application.
While embodiments of the present application have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the present application, and that variations, modifications, alternatives, and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the present application.
Claims (10)
1. A method of dialogue based on an artificial intelligence psychology big model, wherein the psychology big model comprises a scene recognition model, a psychological strategy response model, and a generative big model, the method comprising:
acquiring current input information;
identifying a psychological scene of the current input information based on the scene identification model;
inputting the psychological scene and the current input information to the psychological strategy response model to generate a first prompt word under the condition that the psychological scene is identified;
and inputting the current input information and the first prompt word into the generation type large model to generate target reply information.
2. The dialog method of claim 1 wherein the psychology-big model further comprises a generic policy response model, the method further comprising:
inputting the field current input information to the general strategy response model to generate a second prompt word under the condition that the psychological scene is not recognized;
and inputting the current input information and the second prompt word into the large generation model to generate intermediate dialogue information.
3. The dialog method of claim 2 wherein the psychology big model further comprises an intent recognition model, the method further comprising:
identifying an intent of the current input information based on the intent identification model;
said inputting the field current input information to the generic policy response model to generate a second cue word without identifying the psychological scene, comprising:
in the event that the psychological scenario is not identified, the current input information and the intent are input to the generic policy response model to generate the second cue word.
4. A dialog method according to claim 3 wherein the entering the current input information and the intent into the generic policy response model to generate the second prompt word comprises:
the current input information, the intention and the historical dialog memory are input into a preset general strategy response model to generate a second prompt word, wherein the historical dialog memory comprises at least one of historical dialog information and historical intention.
5. The dialog method of claim 2 wherein the psychological policy response model is established based on a preset psychological scenario and corresponding preset prompt words, and the general policy response model is established based on preset input information and corresponding preset prompt words.
6. The dialog method of claim 1 wherein the inputting the current input information and the first prompt word into the generative large model to generate target reply information comprises:
and inputting the current input information, the first prompt word and a history dialogue memory to a preset generation type large model so as to generate the target reply information, wherein the history dialogue memory comprises one or more input information before the current input information.
7. The dialog method of claim 1, wherein the acquiring of the current input information includes:
and receiving text information and/or voice information input by a user to acquire the current input information.
8. A dialog device for application to an artificial intelligence based psychology big model including a scene recognition model, a psychological strategy response model, and a generative big model, the device comprising:
the acquisition module is used for acquiring current input information;
the identification module is used for identifying the psychological scene of the current input information based on the scene identification model;
the prompt generation module is used for inputting the psychological scene and the current input information into the psychological strategy response model under the condition that the psychological scene is identified so as to generate a first prompt word;
and the reply module is used for inputting the current input information and the first prompt word into the generation type large model so as to generate target reply information.
9. A computer device, comprising:
a processor, a memory; and
A computer program, wherein the computer program is stored in the memory and executed by the processor, the computer program comprising instructions for performing the dialog method of any of claims 1 to 7.
10. A non-transitory computer readable storage medium containing a computer program which, when executed by a processor, causes the processor to perform the dialog method of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311817029.4A CN117744801A (en) | 2023-12-25 | 2023-12-25 | Conversation method, device and equipment of psychology big model based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311817029.4A CN117744801A (en) | 2023-12-25 | 2023-12-25 | Conversation method, device and equipment of psychology big model based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117744801A true CN117744801A (en) | 2024-03-22 |
Family
ID=90283018
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311817029.4A Pending CN117744801A (en) | 2023-12-25 | 2023-12-25 | Conversation method, device and equipment of psychology big model based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117744801A (en) |
-
2023
- 2023-12-25 CN CN202311817029.4A patent/CN117744801A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101925440B1 (en) | Method for providing vr based live video chat service using conversational ai | |
| CN111930940B (en) | Text emotion classification method and device, electronic equipment and storage medium | |
| CN106649704B (en) | Intelligent conversation control method and system | |
| CN112328849B (en) | User portrait construction method, user portrait-based dialogue method and device | |
| CN107807734B (en) | Interactive output method and system for intelligent robot | |
| CN110808038A (en) | Mandarin assessment method, device, equipment and storage medium | |
| CN113761156A (en) | Data processing method, device and medium for man-machine interaction conversation and electronic equipment | |
| CN110489519B (en) | Session method based on session prediction model and related products | |
| CN111611365A (en) | Flow control method, device, equipment and storage medium of dialog system | |
| CN114048299A (en) | Dialogue method, apparatus, device, computer-readable storage medium, and program product | |
| CN112307166B (en) | Intelligent question-answering method and device, storage medium and computer equipment | |
| CN117271745A (en) | Information processing method and device, computing equipment and storage medium | |
| CN111767386A (en) | Conversation processing method and device, electronic equipment and computer readable storage medium | |
| CN112820265B (en) | Speech synthesis model training method and related device | |
| CN117744801A (en) | Conversation method, device and equipment of psychology big model based on artificial intelligence | |
| CN113360625B (en) | Intelligent dialogue marketing customer acquisition method and system based on NLP | |
| CN110456920A (en) | Semantic analysis-based content recommendation method and device | |
| CN114519347A (en) | Method and device for generating conversation content for language and vocabulary learning training | |
| CN113806620B (en) | Content recommendation method, device, system and storage medium | |
| CN114757206A (en) | Consultation service method, system and storage medium | |
| KR20210098623A (en) | Method and system for mimicking tone and style of real person | |
| CN112331201A (en) | Voice interaction method and device, storage medium and electronic device | |
| CN112466435A (en) | Psychological coaching scheme determination method and device, storage medium and electronic device | |
| KR100520426B1 (en) | The method for teaching a chatting avatar | |
| KR102268005B1 (en) | Emotional Artificial Intelligence Curation System and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |