Assessment of the statement in respect to the conversation configuration and the conversation guideline

Abstract

An automated method for generating a report (A) based on an assessment of a user input. A digital model for a virtual person (V) is configured by a large language model (LLM1). The model is based on a

Description

Independent of the grammatical term usage, individuals with male, female or other gender identities are included within the term. Field of the Invention The embodiments disclosed herein relate to an automated method for generating a report based on an assessment of a user input. A corresponding system, computer program product, and computer-readable storage medium are disclosed as well. Background of the Invention Training material for soft skill related trainings, especially communication training, conversation training, pitch training and feedback training and interpersonal interaction training, is of limited availability. E.g. in the area of professional team management absolute finesse and diplomacy is required to do the job well at no emotional cost of the team members and the team manager. Conversations themselves are the job. Methods for quantitative automatic computer-implemented capturing and assessing of a communication skill are not available. Accordingly, there is a need for a training and a quantitative assessment of a communication skill and / or interpersonal interaction skill. Summary Embodiments herein generally relate to a method for generating a report based on an assessment of a user input. The method comprises several steps. All steps are computerized and automated. The method comprises the step of configuring a digital model for a virtual person based on a configuration parameter by a first large language model. The configuration parameter defines a personal characteristic of the virtual person. The configuration parameter can be seen as a parameter used to prompt the first large language model. The configuration parameter defines a personal characteristic of the virtual person and thereby a personality of the virtual person. The personal characteristic results out of a backstory of the virtual person. Therefore, the prompt used by the first large language model to configure the digital model includes the backstory (life story and personality) of the virtual person. The configuration parameter defines personal characteristic of the virtual person comprising a personal age, and / or an accent, and / or a life habit, and / or a location of living, and / or a gender, and / or an ethnicity, each of the virtual person. Different variables for a configuration parameter allow to configure the virtual person according to different personalities and according to different training needs and certain user skills to train. A user profile with previous assessed user skills serves optionally to select the configuration parameter. The method further comprises the step of defining (loading and / or receiving and / or setting) a conversation topic and a conversation guideline. The conversation topic and / or the conversation guideline are defined after loading and / or receiving corresponding data and / or after setting and / or configuring the corresponding data. The said corresponding data comprise the conversation topic and / or the conversation guideline. The conversation topic (e.g. subject) especially comprises a sensitive topic and / or a problem discussion and / or an issue. The sensitive subject is to be understood as an emotional, a vulnerable, subject). Examples for conversation topics are timing (punctuality), reliability, honesty, openness, communication in general, interaction with team members, collaboration with team members, tidiness, accurateness of work results, education, and / or grooming. Examples are an employee or colleague being late for meetings and / or a maintenance worker not having done the due maintenance. Moreover, the user is enabled to choose between conversation configuration to make conversation even more interactive. The conversation configuration includes following parameters: • The conflict type of avatar (e.g. Competing, Accommodating, Avoiding, Compromising, Collaborating); • The user's conflict type (e.g. Competing, Accommodating, Avoiding, Compromising, Collaborating); • The avatar's personality type (e.g. red, white, blue or yellow - based on the Taylor Hartmann https: / / taylorhartman.com / assessment-information / ); • Dialect (e.g. English, Greek, Spanish, Portugese, Russian, Hungarian); • Origin (e.g. UK, Greece, Spain, Portugal, Russia, Hungary); • Culture Influence (e.g. give a rating from 1-5, where very low=1, very high=5); • Environment (e.g. Office or HomeOffice). The conflict type, personal type, origin and cultural influence settings are automatically transmitted to the main prompt to enable customizable conversation setup. The dialect settings (e.g. English, Greek, Spanish, Portugese, Russian, Hungarian) are automatically transmitted and trigger the particular accent during the conversation. The environment settings are automatically transmitted and trigger the particular 3D environment (e.g. Office or Home office) for the scene. The conversation guideline is received and / or loaded from a database. The conversation guideline comprises a requirement on how to give feedback and / or a requirement on a feedback technique, and / or a requirement on how to handle a conversation and / or a requirement on framing a perception and / or an effect of a perception and / or a recommendation. The conversation guideline especially comprises the requirement to follow the “perception-effect-recommendation-feedback-technique”, also known as the PER method. An example for a perception, an effect of a perception and a recommendation are: “I have noticed you are often late (perception), this creates chaos in the team (effect), so maybe you could leave your home 30 mins earlier (recommendation)? The method further comprises the step of generating a representation (e.g. a visualization and a sound, e.g. a voice output) comprising the virtual person (e.g. a representation of the virtual person) based on the digital model, wherein the virtual person represents the personal characteristic defined by the configuration parameter. The representation is also describable as a basis for a simulation of an interpersonal interaction providing one interaction person (the virtual person). In a further embodiment the generation of the virtual person for interaction with the real person (e.g. the user) based on the digital model is done by a 3D computer graphics engine (for example by “unreal engine”, a series of 3D computer graphics game engines developed by Epic Games) and a displaying unit, e.g. an output device, especially a displaying unit, e.g. a monitor, and / or a display, and / or a holographic display, and / or virtual reality (VR) headset, and / or glasses (e.g. virtual or augmented reality glasses, and / or a projector, and / or speakers. Generating the representation (or “visualization”) (e.g. displaying, presenting, outputting the visualisation) comprising the virtual person is of advantage as it makes the interaction between real person and the virtual person easier to execute. The virtual person is especially generated (displayed) in life-sized. This enables an improved and more realistic interaction between the real person and the virtual person. The method further comprises the step of receiving a user input comprising a statement, especially more than one statements. The user input is received via a user interface, especially via typing (textual input), speaking (sound input using a microphone), and additional via streaming (video input using a camera). Option for input devices comprises a motion controller, and / or a VR controller, and / or a gesture recognition system, and / or a motion capture system, and / or a microphone, and / or a keyboard and / or a mouse, and / or a touchscreen, and / or a biometric sensor. The statement especially comprises feedback. The feedback is especially on the said sensitive subject. In a further embodiment the statement especially comprises an assertion (the virtual person may disagree on), and / or an assumption (the virtual person may disagree on), and / or a question, and / or an open question, and / or a closed question. The user input comprising the statement is received as a voice input and / or a written text and / or a mimic and / or a gestic. The method further comprises the step of performing an assessment (e.g. an evaluation) of the statement (especially of more than one statements and / or especially of a further statement after a reply of the virtual person has been put out (see further embodiments)) in respect to the configuration parameter, the conversation topic, and the conversation guideline. The method further comprises the step of and generating a report comprising user assessment (on the user’s performance) and optionally advice (e.g. a recommendation) on how to improve framing of a statement and / or giving feedback and / or a recommendation, and / or framing a question and / or a perception, and / or questioning technique based on a result of the assessment. The method further comprises the step of and generating a report comprising user assessment on the intercultural approach and optionally advice (e.g. a recommendation) on howto improve framing of a statement and / or giving feedback and / or a recommendation, and / or framing a question and / or a perception, and / or questioning technique based on a result of the assessment. This report can be based on the well-known evaluation approaches such as Cultural Dimentions by Hofstede or / and similar frameworks, where each culture can be leveraged by LLM based on the pre-defined factors. Hofstede’s cultural dimensions can be used to help explain why certain behaviors are more or less common in different cultures. These behaviors can be injected to the LLM based on the chosen origin in the configuration and, the characteristics typical for a particular culture, will be triggered during the conversation. The method further comprises the step of outputting the report. The report is outputted via the user interface which is also used for displaying the virtual person. “Outputting the report” is to be understood as making the report available, e.g. readable and / or hearable and / or visually capturable (e.g. showing a “thumbs up) to the real person (the user). According to the embodiments the method supports the user with a simulation of a personal interaction, e.g. a role play, meaning that a real person meets a virtual person, and the virtual person represents a personal characteristic. The conversation topic is either pre-defined by the user or by the system. If the conversation topic is pre-defined by the system, the user is briefed accordingly. The user has to phrase a statement. The statement is assessed with respect to the configuration parameter, the conversation topic, and the conversation guideline. The report and the comprised assessment (and optionally the advice and / or recommendation) serve the user to improve their statement in repetition, e.g. a restart, of the training, e.g. a repetition of the whole method. Further, the training results can be stored in a user profile and reused in case of the training repetition. In a further embodiment the virtual person is configured as an avatar. An avatar is a virtual person or an artificial person in the form of a 3D high-poly asset with the integrated skeleton, which represents the embodiment of a person, especially a photorealistic representation with a digital personality. The skeleton of a 3D asset, often referred to as a "rig," is a hierarchical structure composed of interconnected bones or joints that allows for the animation and manipulation of a 3D model. This skeleton serves as an internal framework to which the 3D mesh is attached, enabling realistic movements and deformations. In other words, the virtual person can be represented as a digital person, especially with corresponding attributes. The attributes can be classified into two groups: Physical attributes (or parameters) and behavioural attributes (or parameters). The physical attributes and behavioural attributes serve as configuration parameters for the virtual person. Physical attributes comprise: • Height: The height of the character, which affects interactions with the environment and other characters. • Build: Body proportions, such as muscular, slim, athletic, etc. • Facial Features: Shape of the face, eyes, nose, mouth, and ears. • Hair Style and Colour: Type (straight, curly, etc.) and colour of hair. • Skin Tone: Colour and texture of the skin. • Age: Young adult, middle-aged, elderly, etc. • Gender: Male, female, or other. • Clothing: Style and type of clothing worn. Behavioural attributes comprise: • Personality Traits: Extroverted, introverted, adventurous, cautious, etc. • Skills and Abilities: Proficiency in various activities or fields (e.g., combat skills, technical skills). • Interests and Hobbies: Activities the character enjoys or pursues. • Behavioural Patterns: Typical reactions to situations, habits, mannerisms. In particular, the avatar can also perform movements of the face and body depending on the output signal for the output of the avatar. For example, when the avatar speaks, corresponding mouth movements can be issued. Thus, an extremely realistic communication between the real person and the virtual person, especially the avatar, can be realized. The prompting of a large language model combined with a photorealistic avatar and preceding brief for the user to follow enable a natural, unique, and memorable user experience that will ensure, seeing as the situation was “lived” (as opposed to consumed), resulting in that the learnings are manifested in the user’s long-term memory. According to a further embodiment, additional to the physical attributes and behavioural attributes which serve as a configuration parameter, the configuration parameter (CP) comprises a communication skill, and / or a social skill, and / or an interaction skill, (resulting in an interaction intensity of the virtual person with the real person) and / or a mindset parameter, and / or a (fictive) relationship type with the real person, and / or a duration they have known the real person, and / or a power balance to the real person and / or an emotional parameter each of the virtual person. According to this embodiment the configuration parameter comprises parameters which influence the communication and / or interaction capabilities of the virtual person. The emotional parameter especially defines how the virtual person reacts on a statement of the user, e.g. on the user input. The emotional parameter influences the representation of the virtual person by showing a certain emotion, e.g. a mood, e.g. a positive or negative body gestures and / or mimic, and / or an exclamation. According to a further embodiment the representation (e.g. a visualization and a sound, e.g. a voice output) generated based on the digital model further includes an environment (e.g. an office or an industrial / factory environment) in which the virtual person is represented (e.g. shown and / or visualized), a background sound, which is played addition (in the background) of a voice output of the virtual person. According to this embodiment the representation especially includes a distraction (especially in form of a background sound or a further representation), such as a phone call. This has the advantage of creating a more realistic scenario. According to a further embodiment the representation (e.g. a visualization and a sound, e.g. a voice output) comprising the virtual person comprises an emotional characteristic (e.g. a mood) (based on the configuration parameter) of the virtual person. According to a further embodiment the emotional characteristic depends on the user input comprising the statement. The emotional characteristic is especially a reaction of the virtual person on the statement of the user. According to this embodiment the representation of the virtual person is amended after the user input is received. Especially, if the PER method is followed by the user, the feedback is well accepted by the virtual person - depicted by positive body gestures (e.g. calm, nodding), mimic (e.g. smiling) and exclamations (e.g. understanding, agreeing). If the PER feedback guideline is not followed by the user, the virtual person reacts emotionally negative (e.g. angry) and presents negative gestures (e.g. aggressive) and answers. According to a further embodiment the method comprises the further step of outputting a briefing information (to the user), the briefing information comprising an information regarding the conversation topic, especially if the conversation topic was received and / or loaded from a database and not received from a user input and / or an information about the virtual person according to the conversation topic and / or an instruction in respect to the conversation guideline. The instruction in respect to the conversation guideline especially comprises an instruction on how to give feedback and / or an instruction on a feedback technique, and / or an instruction on how to handle a conversation and / or an instruction on framing a perception and / or an effect of a perception and / or a recommendation. An example for a perception, an effect of a perception and a recommendation are: “I have noticed you are often late (perception), this creates chaos in the team (effect), so maybe you could leave your home 30 mins earlier (recommendation)”? The step of this embodiment is especially performed before the step of receiving the user input comprising the statement, and after the step of defining (loading and / or receiving and / or setting) the conversation topic (T) and a conversation guideline. The step of this embodiment is especially performed in parallel (e.g. slightly before or slightly after) the step of generating the representation (e.g. a visualization and a sound, e.g. a voice output) comprising the virtual person based on the digital model. According to a further embodiment the method comprises the further step of forwarding the statement to a second large language model, wherein the second large language model is arranged to perform an analysis (arrange to analyse) of the statement in respect to the configuration parameter (and thereby in respect to the personal characteristic represented by the virtual person and defined by the configuration parameter), and wherein the second large language model is arranged to configure (e.g. to generate) a reply based on a result of the analysis. Therefore, the reply is based on the received statement from the user and the configuration parameter of the virtual person. The reply is to be seen as a reply to the statement according to the configuration parameter and the personal characteristics of the virtual person. As an embodiment, for the second large language model, instead of using a different large language (LLM) model, the first large language model and / or the third large language model (described below) are used again. This means that using only one large language model for the disclosed method and all its embodiments is possible (but not necessary). In this embodiment the first large language model, the second large language model, and the third large language model are identical and only on entity. The second large language model is also to be understand as a progression (ongoing development) of the first large language model, e.g. the first large language model is trained and amended during the simulation (execution of the disclosed method) and the second large language model is created. E.g. there is a change in the first large language model over, this is described by using the term “second large language model”. The same applies for the third language model in relation to the first and / or the second large language model, respectively. In this embodiment the same (identical) LLM is used during the whole training, so there is one LLM and the prompt several times inject. The prompt injection is having the following sequence: 1. Inject the first prompt with the virtual personality description and instructions to execute the role (this info is static and stored in the backend). 2. LLM interacts with the user (this info is dynamic and is accessible via UI - user can enter input and get output). 3. Inject evaluation prompt to perform conversation analysis (this info is static and stored in the backend). 4. LLM generates output and sends it to the user (this info is dynamic and comes in form of a textual summary). At any point of time, the same LLM is used and given instructions. According to a further embodiment the method comprises the further step of outputting the reply via the virtual person (e.g. a voice output by the virtual person). The reply is especially outputted by a voice output (plus corresponding visual speaking animation, especially plus mimics, gestic) of the virtual person. This step is carried out especially by a 3D computer graphics engine and a displaying unit, e.g. an output device, especially a monitor, and / or glasses (e.g. virtual reality glasses). The generated representation of the virtual person can be seen as a user interface and / or a user interface showing the representation. According to this embodiment, the virtual person, e.g. an avatar welcomes statements and / or questions and answers, e.g. replies, according to the configuration parameter, e.g. the prompted backstory and motivation. According to an embodiment, the second LLM receives the statement, e.g. a question, and configures (generates) answer in text form based on the pre-defined prompt (the configuration parameter. An option for a pre-defined prompt is: “Pretend to be a Posy Bean. She is a motivated marketing specialist. Topic of feedback: being late for meeting. Guideline: Follow PER framework perception, effect, recommendation”. Text answer (reply) configured by the second LLM is transcribed into audio file in .wav format with pre-configured voice accent (according to the configuration parameter). The .wav file is downloaded on the local visualization machine and played together with a lip-sync animation of the representation of the virtual person. According to a further embodiment the steps of “receiving the user input comprising the statement”, “forwarding the statement to the second large language model”, and “outputting the reply via the virtual person” are repeated for a user input comprising a further statement and for a further reply, respectively. According to this embodiment, the method suggests a dialogue with several statements, from the user and several reactions, e.g. replies, from the virtual person. According to this embodiment an interaction and / or conversation and / or an interview between the virtual person and a real person, e.g. a user, is provided, e.g. simulated. The method provides a training possibility to practice interpersonal interaction and / or conversation. In the assessment the statement and the further statement are assessed in respect to the in respect to the configuration parameter, the conversation topic, and the conversation guideline (e.g. in the same way as disclosed for the statement above). According to a further embodiment the assessment is performed by a third large language model, wherein the third large language model is arranged to assess (doing an assessment) the statement in respect to configuration parameter, (e.g. in relation to the represented personal characteristic), the conversation topic, and the conversation guideline and wherein the third large language model is arranged to generate the report based on a result of the assessment. According to this embodiment, a third LLM generates an overall conversation summary, the result of the assessment and the report, based on a pre-defined evaluation prompt. Especially, it is assessed by the third large language model if the statement is coherent with the conversation topic and / or the conversation guideline (e.g. if the PER method was followed by the user). Further, if more than one statements are received, it is especially assessed if the more than one statements are coherent within themselves. Further, especially it is assessed how well the statement corresponds to the configuration parameter (and therefore to the personal characteristic presented to the user), e.g. if the statement is personalized for the virtual person. As an embodiment, for the third large language model, instead of using a different large language model, the first large language model and / or the second large language model are used again. This means that, as described in detail above, using only one large language model for the disclosed method and all its embodiments is possible (but not necessary). According to a further embodiment the configuration parameter, and / or the conversation topic are received and / or loaded from a database, especially with a corresponding user profile and / or via a user (U) interface. According to this embodiment the configuration parameter, especially several configuration parameters, is / are received / loaded from the database and have been pre-defined by a developer of the method, e.g. a software solution. As an alternative or additional the configuration parameter is received via a user interface (e.g. the configuration parameters are completely or partly defined by a user). According to this embodiment the configuration parameter, especially several configuration parameters, is / are not only received / loaded from the database and have not only been pre-defined by a developer of the method, e.g. a software solution. At least part of the configuration parameters have been defined by a user. With the corresponding user profile, a previous report of a previous training (e.g. the execution of the described method) is storable. A configuration parameter is selectable according to the corresponding user profile, e.g. a virtual person according to certain user skills is configurable. Further, according to this embodiment the conversation topic is received / loaded from the database and have been pre-defined by a developer of the method, e.g. a software solution. As an alternative or additional the conversation topic is received via a user interface (e.g. the conversation topic is completely or partly defined by a user). According to a further embodiment the personal characteristic is represented by the virtual person in a voice output, and / or a spoken sentence, (especially with a corresponding speaking animation (e.g. an animation of mouth and lips of the virtual person), corresponding voice dialect (e.g. Spanisch, Portugese, Hungarian, Russian) and / or a description of a skill, and / or a mimic, and / or a gestic, and / or a movement. This means that the representation of the virtual person includes a voice output, and / or a spoken sentence, and / or a mimic, and / or a gestic, and / or a movement each of the virtual person. Further representation, e.g. visual output, options for personal characteristics of the virtual person include a facial animation, interactive animations, lip-syncing, and interactive eye contact. With “facial animation” the virtual person, also a “character” is arranged to be able to display a wide range of facial expressions, enabling realistic emotion portrayal. With “interactive animations”, the virtual person, especially an avatar, is arranged to be animated to interact with a user (real person) and to provide a more immersive and interactive experience. “Lip syncing” synchronizes a character's lip movement with spoken dialogue, enhancing the realism and immersion of the character's speech. With “interactive eye contact’ the virtual person, especially an avatar, is arranged to be able to dynamically react to the user's gaze, creating a more engaging and interactive environment. According to a further embodiment the reply and / or the report is outputted by a voice output (plus corresponding speaking animation, especially plus mimics, gestic) of the virtual person, especially advantageous for the reply, and / or as text on a displaying unit, e.g. a screen, especially advantageous for the report. According to a further embodiment a system for generating a report based on an assessment of a user input is disclosed. The system comprises a computational device (electronic computation unit (e.g. a processor)) and an output device (e.g. a displaying unit, e.g. a monitor, and / or glasses (e.g. virtual reality glasses), wherein the system is arranged to execute a method according to this disclosure. According to a further embodiment a computer program product is disclosed. The computer program product comprising instructions which, when the program is executed by the computational device, cause the computational device to carry out the steps of the method according to this disclosure. According to a further embodiment a computer-readable storage medium is disclosed. The computer-readable storage medium comprising instructions which, when executed by the computational device, cause the computational device to carry out the steps of the method according to this disclosure. It is to be understood that the elements and features recited in the appended claims may be combined in different ways to produce new claims that likewise fall within the scope of the present invention. Thus, whereby the dependent claims appended below depend from only a single independent or dependent claim, it is to be understood that these dependent claims can, alternatively, be made to depend in the alternative from any preceding or following claim, whether independent or dependent, and that such new combinations are to be understood as forming a part of the present specification. While the present invention has been described above by reference to various embodiments, it should be understood that many changes and modifications can be made to the described embodiments. It is therefore intended that the foregoing description be regarded as illustrative rather than limiting, and that it be understood that all equivalents and / or combinations of embodiments are intended to be included in this description. Brief Description of the Drawing Fig. 1 shows a system according to the disclosure. Detailed Description of the Drawing Fig. 1 shows a system for generating a report based on an assessment of a user input is disclosed. The system comprises a computational device (electronic computation unit (e.g. a processor)) and an output device (e.g. a displaying unit, e.g. a monitor, and / or glasses (e.g. virtual reality glasses), wherein the system is arranged to execute a method as follows. Step S1: Configuring S1 a digital model for a virtual person V based on a configuration parameter CP by a first large language model LLM1, wherein the configuration parameter CP defines a personal characteristic C of the virtual person V. Step S2: Defining S2 a conversation topic T and a conversation guideline G. Wherein the configuration parameter CP, and / or the conversation topic T are received and / or loaded from a database DB, especially with a corresponding user U profile and / or via a user U interface. Step S3a: Generating S3a a representation P comprising the virtual person V based on the digital model, wherein the virtual person V represents the personal characteristic C defined by the configuration parameter CP, wherein the representation P comprising the virtual person V comprises an emotional characteristic EC of the virtual person V, wherein the emotional characteristic depends EC on the user U input comprising the statement F. Step S3b: Outputting S3b a briefing information, the briefing information comprising an information regarding the conversation topic T, especially if the conversation topic T was received and / or loaded from a database DB, and / or an information about the virtual person V according to the conversation topic and / or an instruction in respect to the conversation guideline G. Step S4: Receiving S4 a user U input comprising a statement F. Step S5: Forwarding S5 the statement F to a second large language model LLM2, wherein the second large language model LLM2 is arranged to perform an analysis of the statement F in respect to the configuration parameter CP, and wherein the second large language model LLM2 is arranged to configure a reply R based on a result of the analysis. Step S6: Outputting S6 the reply R via the virtual person V, wherein the reply R by a voice output of the virtual person V. Step S7: Performing an assessment S7 of the statement F in respect to the configuration parameter CP, the conversation topic T, and the conversation guideline G, wherein the assessment S7 is performed by a third large language model LLM3, wherein the third large language model LLM3 is arranged to assess the statement F in respect to configuration parameter CP, the conversation topic T, and the conversation guideline G and wherein the third large language model LLM3 is arranged to generate the report A based on a result of the assessment. Step S8: Generating S8 a report A based on a result of the assessment. Step S9: Outputting S9 the report A, wherein the report A is outputted S9 as text on a displaying unit DU. Wherein the steps of receiving S4 the user U input comprising the statement F, forwarding S5 the statement F to the second large language model LLM2, and outputting S6 the reply R via the virtual person V are repeated, indicated by the circle-shaped arrows, for a user U input comprising a further statement F2 and for a further reply R2, respectively. Although the invention has been explained in relation to its advantageous embodiment(s) as mentioned above, it is to be understood that many other possible modifications and variations can be made without departing from the scope of the present invention. It is, therefore, contemplated that the appended claim or claims will cover such modifications and variations that fall within the true scope of the invention.

Claims

1. An automated method for generating a report (A) based on an assessment of a user input, the method comprising the steps:Configuring (S1) a digital model for a virtual person (V) based on a configuration parameter (CP) by a first large language model (LLM1), wherein the configuration parameter (CP) defines a personal characteristic (C) of the virtual person (V),- defining (S2) a conversation topic (T) and a conversation guideline (G),- generating (S3a) a representation (P) comprising the virtual person (V) based on the digital model, wherein the virtual person (V) represents the personal characteristic (C) defined by the configuration parameter (CP),- receiving (S4) a user (U) input comprising a statement (F),- performing an assessment (S7) of the statement (F) in respect to:o the configuration parameter (CP),o the conversation topic (T), ando the conversation guideline (G)and generating (S8) a report (A) based on a result of the assessment, and- outputting (S9) the report (A).

2. Method according to claim 1, wherein the configuration parameter (CP) comprises:- a communication skill, and / or- a social skill, and / or- an interaction skill, and / or- a mindset parameter, and / or- a relationship type with the real person, and / or- a duration they have known the real person, and / or- a power balance to the real person and / or- an emotional parameter each of the virtual person.

3. Method according to one of the previous claims, wherein the representation (P) further includes:- an environment in which the virtual person is represented,- a background sound, which is played addition (in the background) of a voice output of the virtual person.

4. Method according to one of the previous claims,wherein the representation (P) comprising the virtual person (V) comprises an emotional characteristic (EC) of the virtual person (V).

5. Method according to claim 4,wherein the emotional characteristic (EC) depends on the user (U) input comprising the statement (F).

6. Method according to one of the previous claims,comprising the further step of outputting (S3b) a briefing information, the briefing information comprising:- an information regarding the conversation topic (T), especially if the conversation topic (T) was received and / or loaded from a database (DB), and / or- an information about the virtual person (V) according to the conversation topic and / or- an instruction in respect to the conversation guideline (G).

7. Method according to one of the previous claims, comprising the further steps of:- forwarding (S5) the statement (F) to a second large language model (LLM2), wherein the second large language model (LLM2) is arranged to perform an analysis of the statement (F) in respect to the configuration parameter (CP), and wherein the second large language model (LLM2) is arranged to configure a reply (R) based on a result of the analysis, and- outputting (S6) the reply (R) via the virtual person (V).

8. Method according to claim 7, wherein the steps of:- receiving (S4) the user (II) input comprising the statement (F),- forwarding (S5) the statement (F) to the second large language model (LLM2), and- outputting (S6) the reply (R) via the virtual person (V)are repeated for a user (U) input comprising a further statement (F2) and for a further reply (R2), respectively.

9. Method according to one of the previous claims,wherein the assessment (S7) is performed by a third large language model (LLM3),wherein the third large language model (LLM3) is arranged to assess the statement (F) in respect to:- configuration parameter (CP),- the conversation topic (T), and- the conversation guideline (G)and wherein the third large language model (LLM3) is arranged to generate the report (A) based on a result of the assessment.

10. Method according to one of the previous claims, wherein:- the configuration parameter (CP), and / or- the conversation topic (T) are received and / or loaded:- from a database (DB), especially with a corresponding user (U) profile and / or- via a user (U) interface.

11. Method according to one of the previous claims,wherein the personal characteristic (C) is represented (S3a) by the virtual person (V) in:o a voice output, and / oro a spoken sentence, and / oro a description of a skill, and / oro a mimic, and / oro a gestic, and / or o a movement.

12. Method according to one of the previous claims,wherein the reply (R) according to claim 7 or 8 and / or the report (A) is outputted (S6, S9):- by a voice output of the virtual person (V), and / or- as text on a displaying unit (DU).

13. System for generating a report (A) based on an assessment of a user input, comprising a computational device (CU) and an output device (DU), wherein the system is arranged to execute a method according to one of the claims 1 to 12.

514. A computer program product comprising instructions which, when the program is executed by the computational device, cause the computational device to carry out the steps of the method according to one of the claims 1 to 12.10 15. A computer-readable storage medium comprising instructions which, when executed by thecomputational device, cause the computational device to carry out the steps of the method according to one of the claims 1 to 12.

Images