Computer-implemented conversation method, communication system, computer product and computer-readable storage medium

Abstract

The invention relates to a computer-implemented conversation method that is improved in terms of the user experience, to a communication system for this purpose, to a computer product and to a computer-readable storage medium for implementing an automated conversation system. The present invention makes it possible for the first time to equip a chatbot with user experience to the effect that the outputs from the chatbot are equipped with personal and / or rhetorical and / or individual stylistic means depending on requirements.

Description

[0001] Description

[0002] Computer-implemented conversation method, communication system, computer product and computer-readable storage medium

[0003] The invention relates to a computer-implemented conversation method improved with regard to the user experience, a communication system therefor, a computer product and a computer-readable storage medium for implementing an automated conversation system.

[0004] A conversational system (also referred to as a "chatbot") is a computer-based system capable of conducting human-like conversations with users. Chatbots are often used in social media, websites, or instant messaging applications to provide users with information, answer questions, or offer services.

[0005] Chatbots can be programmed in a variety of ways to conduct human-like conversations. Some chatbots use rule-based systems, which use predefined rules and patterns to respond to user input. Other chatbots use machine learning to conduct human-like conversations and improve and / or adapt based on user input.

[0006] Chatbots can also be used in a variety of industries, such as customer service, e-commerce, and the entertainment industry. They can help improve the efficiency of business processes and provide users with a faster and easier way to respond to their inquiries and needs.

[0007] There are different ways a chatbot can be trained, depending on how it was programmed. A rule-based chatbot is typically programmed by a developer using predefined rules and patterns to respond to user input. These rules and patterns are built into the chatbot from the start and do not require further training.

[0008] A chatbot that uses machine learning to conduct human-like conversations is typically trained by providing it with large amounts of data, known as "training." This data includes typical user inputs and the chatbot's corresponding responses. The machine learning model is then used to identify patterns in this data and learn how to respond to user input.

[0009] There are different types of machine learning models that can be used in chatbots, such as neural networks and decision trees. Each model has its own strengths and weaknesses and may be better suited to certain use cases than others.

[0010] It is important to note that training a chatbot is a continuous process and that the chatbot may need further training to improve its performance and adapt to new user inputs and needs.

[0011] There is the chatbot GPT3, which has been activated for some time and which around 100 million users have already tried out. Based on gigantic amounts of text and with the help of thousands of human "trainers", the system has learned to give seemingly well-thought-out answers. But there are still many gaps, from incorrect salutations and misleading numbers to explicit insults by the chatbot. The more empathetic the user input, the less good the result, the chatbot's output, will be. For example, a user's input to write a sensitive message to end a relationship does not lead to the desired result because the chatbot is not trained to convey empathy.

[0012] On the other hand, an AI is quite capable of separating the objective content of an input from a personality style. For example, there are processors set up and configured to compose music in the classical style, even directly imitating Mozart, for example. An AI can also paint in the style of Edward Hopper and write in the style of Shakespeare if it is trained accordingly. On the other hand, chatbots can also be trained to respond in the typical intonation of Donald Trump or Jedi Master Yoda.

[0013] Currently, chatbot administrators offer two main functionalities. One is the input of at least 20 versions of the same question to trigger what is known as NLU (Natural Language Understanding) training of intents, e.g., "Can I have some ice cream?", "Do you have any ice cream?", "I really want ice cream," etc., to train the intent "The user wants ice cream." The others are hard-coded answers that the chatbot can give when such an intent is recognized, e.g., "Here you can find the nearest ice cream parlor."

[0014] Typically, this content is handed over to the data scientist as unstructured prose. It's also common for the content to be collected from various sources in various formats before this one-stop handover.

[0015] A chatbot administrator is someone responsible for managing and maintaining a chatbot. This may include programming, training, and updating the chatbot, depending on how the chatbot is configured and what type of functions it performs.

[0016] A chatbot administrator might also be responsible for monitoring the chatbot's performance and analyzing user inputs and responses to ensure the chatbot provides accurate and helpful answers. They might also be responsible for integrating the chatbot with other systems or applications, ensuring it functions properly and meets the company's needs.

[0017] In some cases, a chatbot administrator might also be responsible for communicating with users and customers when the chatbot is unable to answer their queries or assist them. In this case, the chatbot administrator might interact directly with users to answer their queries or help them resolve their issues.

[0018] Overall, the chatbot administrator is responsible for managing and maintaining the chatbot, ensuring it functions properly and provides helpful information and services to users.

[0019] A data scientist is a person who analyzes and understands large amounts of data. Data scientists use mathematics, statistics, machine learning, and other tools and techniques to identify and understand patterns and relationships in data. They often work with large amounts of structured and unstructured data, using specialized tools and techniques to process and analyze the data.

[0020] The work of a data scientist often involves analyzing data to support business decisions or improve processes. They can also make predictions or forecasts by identifying patterns in the data and presenting the results in a descriptive format. Data scientists often work closely with other professionals, such as software developers and business analysts, using their data analysis skills to solve complex problems and make decisions. To date, chatbots, whether using written and / or voice input, have operated neutrally; this means that a chatbot response cannot be assigned to a specific type, style, or mood.

[0021] However, this is disadvantageous because it lacks personalization and individualization, the so-called "user experience" for the user.

[0022] The object of the present invention is to create a conversation system which, in addition to the functionality of transmitting the user's intention and the correct answer, is equipped with a possibility of conveying a style, expressing empathy, and transmitting a personality format which is neither related to the chatbot administrator nor to the data scientist.

[0023] This object is achieved by the subject matter of the present invention as defined in the description, the figures and the claims.

[0024] Accordingly, the subject of the present invention is a computer-implemented conversation method comprising the following method steps:

[0025] 1 ) Provision of an interactive computer-implemented communication system with an input device, a processor for writing and / or speech recognition, a neural network with an AI for writing and / or speech evaluation and generation of initial results from the writing and / or speech recognition and their storage,

[0026] - 2 ) Input of a user into the interactive computer-implemented communication system

[0027] - 3 ) Processing of the user's input by an AI for writing and / or speech evaluation and generation of initial results from the writing and / or speech recognition, - 4 ) Carrying out an initial processing of the initial results by means of an AI set up and configured for this purpose in the form of a neutralization and / or removal of the stylistic elements from the voice and / or written input received from the user in order to formalize the input and / or anonymize the inputting user,

[0028] - 5 ) Generation of second results by the AI ​​based on the content of the user input in an anonymized and neutral form,

[0029] 6 ) Processing of the second results by the AI ​​to generate appropriate outputs, for example in the form of answers, and output them via a suitable output device .

[0030] Furthermore, the present invention provides an exemplary communication system comprising the following components:

[0031] - an input device such as a keyboard, a microphone, a camera,

[0032] - a first processor with a memory unit, which is set up and configured to carry out a writing and / or speech recognition, which has a neural network and an AI and can operate them and thus generates first results that can be stored in the connected first memory unit,

[0033] - a second processor which is set up and configured to firstly retrieve the first results from the first storage unit, to process them and to make them available to an AI which is set up and trained to neutralise the results, to remove the rhetoric and / or to free it from personalising stylistic elements and / or to insert them, wherein the results which are fed into this AI are either first, neutralised inputs from a user or third, neutral answers from the chatbot,

[0034] - wherein the AI ​​is set up and configured to equip the received neutral content in writing and / or speech with a predefined or AI-selectable style and / or rhetoric and / or personality attributes and / or to transform them accordingly and forward them to a third processor,

[0035] - wherein the third processor is set up and configured to perform speech and / or text generation by means of AI and to offer this acoustically, optically or both as output to the input received from the user via an output device of the communication system.

[0036] According to an advantageous embodiment of the invention, between the penultimate step of finding a suitable output and the last step of outputting a suitable answer, counter-speech and / or reaction by the communication system, the process step 7) takes place, by which the output generated by the AI ​​is supplemented with personalizing and / or other style elements.

[0037] For example, process step 7 can take place by carrying out processing of the output by the AI ​​in such a way that the output is rhetorically stylized by the AI ​​according to known rules and / or is typified by the choice of words, sentence length, sentence structure and / or is carried out in a typical personality profile stored in the AI ​​- for example in the style of Donald Trump.

[0038] The general finding of the invention is that an AI can separate general and personal stylistic elements from the core content of a user's input, i.e. it can subtract these during input and "expose" the content in a neutral way. The invention takes advantage of this and carries out the method when a user inputs something by having an AI carry out the neutralization and anonymization of an input and / or by having the AI ​​add any stylistic elements, typing features and / or elements for personalization and / or individualization when outputting something, for example in the subsequent answer. This creates a feeling of well-being for the user and a conversation between machine and user can be conducted in a much more open, immersive and / or thorough manner.So here not only the intonation of, for example, Jedi Master Yoda - because he is so peculiar - is made available to a chatbot, but also their way of thinking, their choice of words, their typical sentence structure, style and their individual approach to certain topics, questions, content, etc.

[0039] The training possibilities for the AI ​​processing performed in process step 7) to stylize the output are virtually unlimited. Any interview with a user can serve as the basis for generating training data. However, interviews, plays, and / or films collected in archives, which are available in audio and / or written form, can also be evaluated as a data pool.

[0040] Wherever empathetic speech, writing, and / or responses occur, data can be extracted for style and / or personality training of an AI. This data enables the appropriately trained AI, connected to the chatbot, to make the communication system's outputs appear emotional, empathetic, personal, and / or otherwise individual. This results in an improved user experience of the corresponding computer-implemented communication system.

[0041] For example, there are TextToSpeech machines that use "Speech Synthesis Markup Language." This technology makes it possible, for example, to make the audio output of a chatbot seem more emotional, although this does not necessarily mean that the resulting machine response has any empathy or emotional style.

[0042] The invention makes the previously common process step, in which data scientists use an algorithm to individually assign a personality to a chatbot's spoken and / or written response, superfluous because the AI ​​is trained to automatically select a personality and / or style for the user and to assign this individualization to its results, which are output, for example in the form of a spoken response from the chatbot.

[0043] In this way, the chatbot of the computer-implemented communication system can also be trained to be empathetic, because an emotional style of response conveys and / or suggests more empathy and understanding to the user. For example, in a further, optionally implemented process step, the user can enter a requirement profile for the style in which the output should be provided when making their input. This requirement profile can be created consciously and / or unconsciously, because, for example, the choice of words during input or simply through targeted questions conveys to the AI ​​which style or personality type the user prefers.

[0044] However, empathy in a response can be created by framing the response in Shakespearean style, because Shakespeare simply has an inherently empathetic language due to the many flowery depictions.

[0045] For example, scientifically well-researched linguistic and / or rhetorical stylistic devices can be used to enhance the expressiveness of a written or spoken text, with their effect on the reader / listener being precisely identifiable. Examples include metaphor, alliteration, hyperbole, oxymoron, antiphrasis, archaism, conversio, euphemism, tautology, climax, paraphrases, pleonasms, etc.

[0046] Rhetorical devices can be used to train the AI ​​and assigned to different personality profiles. This can influence the chatbot's choice of words, allowing a response to be developed in a specific style.

[0047] In addition to word choice, the AI ​​can also be trained to use a specific voice, intonation, and / or rhythm. This allows the chatbot's output to be given a specific style. Thus, virtually any output content and any output type—whether acoustic, optical, and / or visual—can be automatically provided with a literary, rhetorical, and / or personality-specific style by a suitably trained AI.

[0048] According to an advantageous embodiment of the invention, the user's reaction is recorded and fed into the system so that the AI ​​can learn from the reactions which type of chatbot is suitable for which type of user.

[0049] According to an advantageous embodiment of the invention, speech recognition in process step A) is carried out by means of an AI with NLP capability.

[0050] Natural Language Processing (NLP) is a branch of artificial intelligence that deals with the interaction between computers and humans using natural language. It involves the development of algorithms and models that can analyze, understand, and generate human language, including text and speech.

[0051] NLP capabilities refer to the ability of a computer system or software to perform tasks in the area of ​​natural language processing. These include, for example, language translation, text classification, sentiment analysis, and language generation.

[0052] In general, the goal of NLP is to enable computers to process, analyze, and understand human language in a way similar to humans, so that they can interact more effectively with people and perform tasks that would be difficult or impossible for humans to perform.

[0053] In the following, the invention is explained in more detail using a flowchart that represents the sequence of an exemplary embodiment of the disclosed method. - The process step 1) of providing an interactive computer-implemented communication system with an input device such as a microphone, a keyboard, a camera, a processor for character and / or speech recognition, a neural network with an AI for character and / or speech evaluation and generation of initial results from character and / or speech recognition as well as their storage,

[0054] - As well as the subsequent process step 2 ) of inputting a user into the interactive computer-implemented communication system via the input device;

[0055] - The process step 3 ) of processing the user's input by a suitable and appropriately configured processor with a neural network on which an AI for writing and / or speech evaluation and generation of initial results from the writing and / or speech evaluation and / or writing and / or speech recognition is running,

[0056] - Following this, process step 4) of carrying out a first processing of the first results by means of an AI set up and configured for this purpose in the form of a neutralization and / or removal of the stylistic elements from the voice and / or written input received from the user in order to formalize the input and / or anonymize the inputting user, as well as the following process step

[0057] - 5 ) the generation of second results by the AI ​​, which are based on the content of the user input in an anonymized and neutral form and finally the process step

[0058] 6) Processing of the second results by the AI ​​to generate appropriate outputs, for example in the form of answers, and to output them via a suitable output device, monitor, loudspeaker, etc.

[0059] Advantageously, the intermediate step 7 takes place between process step 5 and process step 6, in which the communication system carries out processing of the output by the AI ​​via the corresponding AI, which is trained as described above, in such a way that the output by the AI ​​is stylized and / or in a personality profile stored in the AI.

[0060] The present invention makes it possible for the first time to equip a chatbot with user experience in such a way that the chatbot's outputs are equipped with personal and / or rhetorical and / or individual stylistic devices depending on the requirements.

Claims

Patent claims 1 . Computer-implemented conversation method comprising the following steps: 1 ) Provision of an interactive computer-implemented communication system with an input device, a processor for writing and / or speech recognition, a neural network with an AI for writing and / or speech evaluation and generation of initial results from the writing and / or speech recognition and their storage, - 2 ) Input of a user into the interactive computer-implemented communication system - 3 ) Processing of the user's input by an AI for writing and / or speech evaluation and generation of initial results from the writing and / or speech recognition, - 4 ) Carrying out an initial processing of the first results by means of an AI set up and configured for this purpose in the form of a formalization, neutralization and / or removal of personalizing and / or other stylistic elements from the voice and / or written input received from the user in order to formalize the input and / or anonymize the inputting user, - 5 ) Generation of second results by the AI ​​based on the objective content of the user input in an anonymized and neutral form, 6 ) Processing of the second results by the AI ​​to generate appropriate outputs, for example in the form of answers, and output them via a suitable output device .

2. Method according to claim 1, in which before the process step 6, the processing of the second, neutral results by the AI ​​and generation of the corresponding output by the AI, there is an intermediate step 7) by the output generated by the AI ​​is supplemented with personalizing and / or other stylistic elements.

3. Method according to one of claims 1 or 2, in which the output is designed by an appropriately trained AI with rhetorical stylistic devices.

4. Method according to claim 3, in which metaphors are incorporated into the text as stylistic devices.

5. Method according to claim 3 or 4, in which word repetitions are incorporated into the text as a stylistic device.

6. Method according to one of claims 3 to 5, in which alliteration, hyperbole and / or oxymoron are incorporated into the text as stylistic devices.

7. Method according to one of claims 3 to 6, in which antiphrase, conversio and / or tautology are incorporated as stylistic devices.

8. Method according to one of claims 3 to 7, in which archaism, euphemism and / or climax are incorporated into the text as stylistic devices.

9. Method according to one of claims 3 to 8, in which paraphrases and / or pleonasms are incorporated into the text as stylistic devices.

10. Communication system comprising the following components: - an input device such as a keyboard, a microphone, a camera, - a first processor with a memory unit, which is set up and configured to carry out a character and / or speech recognition, which has and can operate a neural network and an AI and thus generates first results that can be stored in the connected first memory unit, - a second processor which is set up and configured to firstly retrieve the first results from the first storage unit, process them and make them available to an AI which is set up and trained to neutralise results, remove the rhetoric and / or free them from personalising stylistic elements and / or insert them, wherein the results fed into this AI are either first, neutralised inputs from a user or third, neutral answers from the chatbot, - wherein the AI ​​is set up and configured to equip the received neutral content in writing and / or speech with a predefined or AI-selectable style and / or rhetoric and / or personality attributes and / or to transform them accordingly and forward them to a third processor, - wherein the third processor is set up and configured to perform speech and / or text generation by means of AI and to offer this acoustically, optically or both as output to the input received from the user via an output device of the communication system.

11. Computer product suitable for carrying out the method according to one of claims 1 to 9.

12. A computer-readable storage medium comprising a computer program product according to claim 11.

Images