Method and system for generating relevant prompt words for science and technology text based on large model

Abstract

The present invention discloses a method and system for generating prompt words related to science and technology texts based on a large model. The method and system determine technical subject words, background words and key words by analyzing simple prompt words input by a user, thereby determining the technical subject, background and key points of the science and technology texts desired by the user. Then, based on the current simple prompt words, science and technology prompt words including the latest technical branches, innovative concepts, potential problems and improvement effects are derived to further improve the prompt words. The improved prompt words are input into a second large model to generate a science and technology text, thereby improving the professionalism and information content of the text and meeting the high-standard writing requirements in the field of science and technology.

Description

Technical Field

[0001] The present invention belongs to the technical field of large models, and in particular relates to a method and system for generating prompt words related to science and technology texts based on large models. Background Art

[0002] Prompts are inputs from users to large language models, which guide the model to generate text of a specific theme or style. Prompts can be a question, a sentence, or a paragraph, which sets the tone, content, and direction of text generation. Prompts play a very important role in the text generation process. A good prompt can improve the quality of generated text and make it more relevant and useful.

[0003] In the context of the current rapid development of science and technology, the writing of scientific and technological texts such as patent applications, industrial research reports, and project proposals requires extremely high professionalism and information. However, when users prompt the big model, they often input too simple inputs, lacking clear technical subject terms, background terms, and key terms, resulting in the generated text being difficult to meet the needs of professionalism and depth. In addition, although there are tools for big models to assist in writing texts, they fail to effectively parse the key information in the user's simple prompt words, making it impossible to generate high-quality technical texts. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present application provides a method and system for generating scientific and technological texts and related prompt words based on a large model.

[0005] Receiving a simple prompt word input by a user, analyzing the simple prompt word using the first large model, and identifying and extracting a subject word, a background word, and a key word therefrom;

[0006] Determine the technical theme according to the subject words, and list the main technical branches, components and innovative concepts covered based on the determined technical theme. The innovative concept refers to the technical principle, application scenario and system architecture extracted based on the subject words;

[0007] Perform technical background analysis based on the background words, and determine the technical background, innovation direction background, and time background based on the analysis results. The innovation direction background refers to the technology development trends, industry pain points, and market demands extracted based on the background words;

[0008] Perform technical effect analysis based on the key words, and determine the key technical features and technical performance indicators related to the technical subject based on the analysis results. The key technical features refer to the core attributes that are indispensable for the technology to achieve its functions and are different from other technologies. The technical performance indicators are specific parameters for quantifying the key technical features, which are used to measure the effectiveness and effects of the technology implementation;

[0009] The relevant information analyzed from the subject words, the background words and the key words is combined to generate a final version of prompt words that matches the format and style of the science and technology text.

[0010] In some optional implementations of some embodiments, the simple prompt word received from the user is analyzed by using the first large model to identify and extract the subject words, background words and key words, which includes a subject word extraction step, and the subject word extraction step is:

[0011] Leveraging the text classification capabilities of the first model, simple prompt words entered by users are matched with pre-built technical domain vocabulary libraries;

[0012] The word frequency-inverse document frequency algorithm is used to identify the words with the highest relevance to the technical topic among the simple prompt words as the subject words.

[0013] In some optional implementations of some embodiments, the step of receiving a simple prompt word input by a user, analyzing the simple prompt word using the first large model, identifying and extracting a subject word, a background word and a key word therefrom, further includes a background word analysis step, wherein the background word analysis step is:

[0014] Use the textual implication relationship analysis technology of the first model to identify sentences describing the background of technological development from simple prompt words;

[0015] By using the time expression recognition technology, the time background in the simple prompt words is extracted, and the extracted time background is combined with the technology development background sentence to form a complete description of the technology development background.

[0016] In some optional implementations of some embodiments, the step of receiving a simple prompt word input by a user, analyzing the simple prompt word using a first large model, identifying and extracting a subject word, a background word, and a key word, further includes a key word determination step, wherein the key word determination step is:

[0017] Using the sentiment analysis and keyword extraction technology of the first model, positive adjectives and nouns in the simple prompt words are identified as potential technical effects and improvement points;

[0018] The word vector technology is used to compare the similarity between the words in the simple prompt words and the known technical advantage words, so as to determine the key words.

[0019] In some optional implementations of some embodiments, the determining of the technical theme according to the subject word and listing of the main technical branches, components and innovative concepts covered based on the determined technical theme include:

[0020] After obtaining the subject terms, the deep learning algorithm of the first model is used to conduct extensive searches and analyses of relevant technical literature, patent databases, and industry reports to determine the technical topics;

[0021] Using the cluster analysis function of the first model, the technologies related to the identified technical topics are classified according to functions, application fields and implementation methods, and the main technical branches are sorted out. Each technical branch is described in detail, including working principles, technical characteristics and current research progress;

[0022] According to the main technical branch, each key component of the main technical branch is identified through the identification function of the first model, and the key components are listed, and each key component is elaborated in detail to explain its role and mutual relationship in the overall technical framework, wherein the identification function is used to identify the key components from the technical subject;

[0023] Through the discovery function of the first model, the latest scientific research papers, technical blogs and industry trends are tracked, innovative concepts and technological trends related to technical topics are explored, how the innovative concepts are applied to technical topics is analyzed, and their impact on the industry and potential commercial value are predicted. The discovery function is used to explore innovative concepts related to technical topics from the latest materials.

[0024] In some optional implementations of some embodiments, performing technical background analysis according to the background words and determining the technical background, innovation direction background and time background based on the analysis results include:

[0025] The natural language processing technology and background analysis function of the first model are used to perform semantic understanding and association analysis on background words, thereby depicting the macro-technical background of the technical theme. The analysis content includes the limitations of existing technologies, bottlenecks in industry development, and changing trends in relevant policies and market demand. The background analysis function is used to output a detailed description of the technical background.

[0026] Based on background analysis and combined with the problem identification function of the first model, the potential application scenarios of the technical theme in the future are predicted, and currently unresolved or newly emerging technical problems are identified. The classification algorithm in machine learning is used to train a large amount of industry data to distinguish and extract key application scenarios and problem points. Among them, the problem identification function is used to analyze and output potential application scenarios and problems to be solved from the technical background.

[0027] In some optional implementations of some embodiments, the performing of technical effect analysis according to the key words and determining the key technical features and technical performance indicators related to the technical subject based on the analysis results include:

[0028] The key word analysis function of the first model is used to deeply analyze the key words provided by the user to identify key features, performance indicators or user focus related to the technical topic, wherein the key word analysis function is used to output key points of technical improvement;

[0029] By mining the semantics of key words, combined with industry standards and user needs, we can evaluate the potential value and market impact of technology improvements;

[0030] Based on the key features, performance indicators or user focus related to the technical topic, the first model is used to automatically recommend technical improvement directions;

[0031] A decision tree algorithm is used to analyze historical patent data and industry reports to assist in determining the specific goals and expected effects of technological improvements.

[0032] In some optional implementations of some embodiments, the generating of a final version of prompt words matching the format and style of a scientific and technological text by combining the relevant information analyzed from the subject words, background words and key words includes:

[0033] Based on the technical theme, background analysis, key improvement points and expected effects, a final version of the prompt words is constructed that includes keywords, technical parameters, innovations, application scenarios and target effects.

[0034] In the second aspect, the present application proposes a method for generating scientific and technological texts, comprising:

[0035] The final version of the prompt words described in any of the above methods is input into the second largest model, processed by natural language processing technology and combined with the specific format and style requirements of scientific and technological texts to generate corresponding scientific and technological texts.

[0036] In the third aspect, the present application proposes a system for generating prompt words related to science and technology texts based on a large model, comprising: a simple prompt word classification and extraction module, a subject word application analysis module, a background word application analysis module, a key word application analysis module and a final prompt word generation module, wherein the simple prompt word classification and extraction module is used to: receive simple prompt words input by a user, analyze the simple prompt words using a first large model, and identify and extract subject words, background words and key words therefrom;

[0037] The subject word application analysis module is used to: determine the technical subject according to the subject word, and list the main technical branches, components and innovative concepts covered based on the determined technical subject;

[0038] The background word application analysis module is used to: perform technical background analysis according to the background words, and determine the technical background, innovation direction background and time background based on the analysis results;

[0039] The key word application analysis module is used to: perform technical effect analysis based on the key words, and determine the key technical characteristics and technical performance indicators related to the technical subject based on the analysis results;

[0040] The final prompt word generation module is used to generate a final version of prompt words that matches the format and style of the science and technology text by combining the relevant information analyzed from the subject words, background words and key words.

[0041] Beneficial effects of the present invention:

[0042] By parsing the simple prompt words input by the user, we determine the technical subject words, background words and key words, and then determine the technical theme, background and key points of the scientific and technological text that the user wants. Based on the current simple prompt words, we derive scientific and technological prompt words that include the latest technical branches, innovative concepts, potential problems and improvement effects to further improve the prompt words. The improved prompt words are input into the second largest model to generate the scientific and technological text, so as to improve the professionalism and information content of the text and meet the high-standard writing needs in the field of science and technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is the overall flow chart of the present invention.

[0044] Figure 2 This is the structural principle diagram of this system. DETAILED DESCRIPTION

[0045] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein; on the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0046] In the first aspect, the present application proposes a method for generating scientific and technological texts and related prompt words based on a large model, such as Figure 1 As shown, the following steps are included:

[0047] S100: receiving a simple prompt word input by a user, analyzing the simple prompt word using a first large model, and identifying and extracting a subject word, a background word, and a key word therefrom;

[0048] After receiving the simple prompt words input by the user, the text of the input simple prompt words is analyzed using natural language processing technology. First, the input text is deeply understood through the first large model, and the core theme-related words, i.e., subject words, are identified and extracted. These subject words can clearly indicate the technical field and core content of the scientific and technological text that the user needs to write.

[0049] In some optional implementations of some embodiments, the simple prompt word received from the user is analyzed by using the first large model to identify and extract the subject words, background words and key words, which includes a subject word extraction step, and the subject word extraction step is:

[0050] S101: using the text classification function of the first model, matching the simple prompt words input by the user with the pre-built technical field vocabulary library;

[0051] S102: Using a word frequency-inverse document frequency algorithm, identify the words with the highest relevance to the technical topic among the simple prompt words as subject words.

[0052] First, the built-in text classification function of the first model is used to accurately match the simple prompt words entered by the user with a pre-built and continuously updated technical field vocabulary library. The key to this step is that the vocabulary library contains a large number of professional words in the field of science and technology. These words are classified and labeled to help us quickly locate the technical related words in the prompt words. Next, this solution uses the term frequency-inverse document frequency (TF-IDF) algorithm to evaluate the importance of each word. The TF-IDF algorithm is a statistical method that can determine the distinctiveness of a word for a specific text collection based on the frequency of the word in the text (TF) and the distribution in the entire document collection (IDF). The algorithm is used in this step, in which logarithmic (log) smoothing is used to help reduce the impact of high-frequency words on the text. It can be understood that the logarithmic function can map a large range of values ​​to a smaller range, effectively reducing the impact of high-frequency words, making the ground frequency more influential in the calculation. The following are the specific implementation steps:

[0053] S1021: Segment the simple prompt words input by the user to obtain a series of words.

[0054] S1022: Calculate the TF value of each word, that is, the number of times the word appears in the prompt word divided by the total number of all words in the prompt word.

[0055] S1023: Calculate the IDF value of each word, that is, log (total number of documents / number of documents containing the word).

[0056] S1024: Calculate the TF-IDF score of each word by combining the TF and IDF values. The higher the score, the more likely the word is to be a topic word.

[0057] In some optional implementations of some embodiments, the step of receiving a simple prompt word input by a user, analyzing the simple prompt word using the first large model, identifying and extracting a subject word, a background word and a key word therefrom, further includes a background word analysis step, wherein the background word analysis step is:

[0058] S103: using the text implication relationship analysis technology of the first model to identify sentences describing the background of technological development from simple prompt words;

[0059] S104: using time expression recognition technology to extract the time background in the simple prompt word, and combining the extracted time background with the technology development background statement to form a complete description of the technology development background.

[0060] First, the text implication relationship analysis technology of the first model is used to deeply interpret the simple prompt words entered by the user. Text implication relationship analysis is a natural language processing technology that can identify the logical relationship between sentences, such as cause-effect relationship, conditional relationship, etc. In this step, the technology is used to identify the technology development background information implied in the prompt words, that is, information that is not directly mentioned but can be inferred from the context.

[0061] Next, we use time expression recognition technology to extract the time context in the prompt words. Time expression recognition involves the recognition of time words and the relative relationship between time in the text, which helps to construct a timeline of technological development and provide accurate historical and technological background for scientific and technological texts.

[0062] The following are the specific implementation steps:

[0063] S1031: Segment the simple prompt words input by the user into sentences to obtain independent sentence units.

[0064] S1032: Apply text implication relationship analysis technology to perform logical relationship analysis on each sentence and identify sentences that describe the background of technological development.

[0065] S1041: Perform time expression recognition on the identified background sentence to extract the specific time point or time period.

[0066] S1042: Combine the extracted time context with the background sentences to form a complete description of the technology development background.

[0067] Time expressions can be identified and extracted by various means. For example, the date format is year-month-day, year / month / day, month / day, etc. The time format usually contains hours, minutes and seconds, that is, xx:xx:xx; the date and time format is a combination of the above two. For the parsing process of time expressions, you can first identify the time expression format, then use appropriate separators to split the string into various parts, and then convert each part into a number or time unit corresponding to the range, verify whether the result of the above conversion is logical, and then use the logical result as the result of extracting the timeline.

[0068] In some optional implementations of some embodiments, the step of receiving a simple prompt word input by a user, analyzing the simple prompt word using a first large model, identifying and extracting a subject word, a background word, and a key word, further includes a key word determination step, wherein the key word determination step is:

[0069] S105: using the sentiment analysis and keyword extraction technology of the first large model to identify positive adjectives and nouns in the simple prompt words as potential technical effects and improvement points;

[0070] S106: Using word vector technology, compare the similarity between the words in the simple prompt words and the known technical advantage words, so as to determine the key words.

[0071] 1. Sentiment Analysis:

[0072] Sentiment analysis is a natural language processing technique used to identify and extract subjective information from text, that is, the author's emotional inclination towards a certain topic or the text as a whole. In this solution, sentiment analysis can be performed on simple prompt words input by users to identify words with positive emotional inclinations.

[0073] Sentiment Classification: Using the sentiment classifier of the first model, each word in the prompt is classified as positive, negative, or neutral.

[0074] Extract positive words: From the results of sentiment classification, filter out all adjectives and nouns marked as positive.

[0075] 2. Keyword extraction:

[0076] After extracting the positive words, keyword extraction techniques are then used to further determine the importance of these words.

[0077] Calculate the importance score: For each positive word, calculate its importance score in the entire prompt word. The TF-IDF algorithm or other statistics-based algorithms can be used.

[0078] Sorting and filtering: Sort the words according to their importance scores and filter out the words with the highest scores as potential focus words.

[0079] 3. Word vector technology:

[0080] In order to accurately determine the key words, we use word embedding technology to compare the similarity between the words in the prompt words and the known technical advantage words.

[0081] Word vector representation: Use the word vector model of the first large model to convert each word in the prompt word into its corresponding word vector representation.

[0082] Similarity Calculation: Calculate the cosine similarity between the word vector of each positive word and the word vector of each word in the pre-defined technical advantage word set.

[0083] Determine key words: Select the words with the highest similarity as key words, because these words are most likely to represent the key points of technical improvement.

[0084] Assume that the prompt word entered by the user is "Develop an intelligent robot to improve the efficiency of the production line". Through the above process, the following information is extracted: - Subject words: intelligent robot, production line - Background words: low efficiency of production line, automation demand - Key words: efficiency improvement, intelligent transformation.

[0085] S200: Determine a technical theme according to the subject word, and list the main technical branches, components and innovative concepts covered by the determined technical theme based on the determined technical theme, wherein the innovative concept refers to the technical principle, application scenario and system architecture extracted based on the subject word;

[0086] 1. Technical principles: Innovative concepts can be improvements on existing technical principles or completely new technical principles. These technical principles are closely related to the technical subject matter. For example, in the field of artificial intelligence, innovative concepts may include new neural network structures, optimization algorithms, etc.

[0087] 2. Application scenarios: Innovative concepts can propose new solutions for application scenarios of specific technology themes. For example, in the field of autonomous driving, an innovative concept may be a new sensor technology used to improve the vehicle's perception ability in adverse weather conditions. 3. Technology fusion: Innovative concepts can be cross-disciplinary or cross-field technology fusions that bring new development opportunities to technology themes. For example, the combination of bioinformatics and artificial intelligence has brought new breakthroughs in gene editing technology.

[0088] 3. System architecture: The innovative concept can be the optimization or reconstruction of the system architecture of existing technical themes. For example, in the field of cloud computing, the innovative concept may be a new distributed storage technology to improve the efficiency of data storage and processing.

[0089] In some optional implementations of some embodiments, the determining of the technical theme according to the subject word and listing of the main technical branches, components and innovative concepts covered based on the determined technical theme include:

[0090] After obtaining the subject terms, the deep learning algorithm of the first model is used to conduct extensive searches and analyses of relevant technical literature, patent databases, and industry reports to determine the technical topics;

[0091] Using the cluster analysis function of the first model, the technologies related to the identified technical topics are classified according to functions, application fields and implementation methods, and the main technical branches are sorted out. Each technical branch is described in detail, including working principles, technical characteristics and current research progress;

[0092] According to the main technical branch, each key component of the main technical branch is identified through the identification function of the first model, and the key components are listed, and each key component is elaborated in detail to explain its role and mutual relationship in the overall technical framework, wherein the identification function is used to identify the key components from the technical subject;

[0093] Through the discovery function of the first model, the latest scientific research papers, technical blogs and industry trends are tracked, innovative concepts and technological trends related to technical topics are explored, how the innovative concepts are applied to technical topics is analyzed, and their impact on the industry and potential commercial value are predicted. The discovery function is used to explore innovative concepts related to technical topics from the latest materials.

[0094] In order to precisely determine the technical subject, the following implementation steps are taken:

[0095] 1. Literature search: Using the search function of the first model, a wide range of literature searches are conducted on the technical fields related to the prompt words entered by the user, including but not limited to patent databases, scientific paper libraries, industry reports, etc.

[0096] 2. Content analysis: Conduct in-depth analysis of the retrieved literature content and use text mining technology to extract keywords, abstracts, and technical category information.

[0097] 3. Topic modeling: Apply topic modeling algorithms (such as latent Dirichlet allocation, LDA) to perform cluster analysis on the extracted information and identify potential technical topics.

[0098] 4. Topic Verification: Verify and adjust the identified technical topics through expert systems or manual review to ensure their accuracy and relevance.

[0099] After determining the technical topic, the following are the steps to subdivide the technical branches:

[0100] 1. Technology branch identification: Use the cluster analysis function of the first model to classify technologies according to functions, application fields, and implementation methods based on the literature content under the technology theme.

[0101] 2. Branch description: For each identified technical branch, describe in detail its working principle, technical characteristics and current research progress.

[0102] 3. List of key components: For each technical branch, identify and list the key components.

[0103] In a specific embodiment, taking the "intelligent robot" technology theme as an example, the first large model is used to cluster the technical literature related to "intelligent robots" to determine the main technical branches such as "perception system", "decision planning" and "actuator". Then the key components of the intelligent robot are listed, for example, the perception system includes sensors such as cameras and radars; decision planning involves algorithms such as path planning and task scheduling; and the actuators include motors and servo systems. Finally, by tracking the latest scientific research progress, innovative concepts such as "deep reinforcement learning" and "multimodal perception fusion" are discovered, and their application potential in the field of intelligent robots is explored.

[0104] S300: performing a technical background analysis based on the background words, and determining the technical background, innovation direction background, and time background based on the analysis results, wherein the innovation direction background refers to the technology development trends, industry pain points, and market demands extracted based on the background words;

[0105] 1.Technology development trends:

[0106] Background words: such as "artificial intelligence", "5G", "blockchain", etc. Innovation direction background: These background words indicate the current hot spots and future trends of technological development. For example, "artificial intelligence" may point to innovations in sub-fields such as deep learning and natural language processing; "5G" may be related to the application and optimization of technologies such as the Internet of Things and edge computing.

[0107] 2. Market demand and industry pain points:

[0108] Background words: such as "inefficiency", "high cost", "poor user experience", etc.

[0109] Innovation Direction Context: These words point out problems in the current market or industry, thus guiding innovation directions to solve these problems. For example, "inefficiency" may lead to the innovation direction of developing automated tools or processes.

[0110] In some optional implementations of some embodiments, performing technical background analysis according to the background words and determining the technical background, innovation direction background and time background based on the analysis results include:

[0111] The natural language processing technology and background analysis function of the first model are used to perform semantic understanding and association analysis on background words, thereby depicting the macro-technical background of the technical theme. The analysis content includes the limitations of existing technologies, bottlenecks in industry development, and changing trends in relevant policies and market demand. The background analysis function is used to output a detailed description of the technical background.

[0112] Based on background analysis and combined with the problem identification function of the first model, the potential application scenarios of the technical theme in the future are predicted, and currently unresolved or newly emerging technical problems are identified. The classification algorithm in machine learning is used to train a large amount of industry data to distinguish and extract key application scenarios and problem points. Among them, the problem identification function is used to analyze and output potential application scenarios and problems to be solved from the technical background.

[0113] In a specific embodiment, taking the "intelligent robot" technology theme as an example, the natural language processing technology and background analysis function of the first large model are used to analyze the background words related to "intelligent robots", such as "aging society" and "labor shortage", to understand the technical needs and social development trends behind these background words. Then, the SVM algorithm is used to train the industry data, and the problem identification function is used to identify the potential application scenarios of intelligent robots in the fields of medical care, warehousing and logistics. At the same time, by analyzing the limitations of existing technologies, such as "lack of naturalness of human-computer interaction" and "poor adaptability to complex environments", the direction for subsequent technological innovation is pointed out.

[0114] S400: performing technical effect analysis according to the key words, and determining technical key features and technical performance indicators related to the technical subject based on the analysis results, wherein the technical key features refer to the core attributes of the technology that are indispensable and different from other technologies when realizing its functions, and the technical performance indicators are specific parameters for quantifying the technical key features, and are used to measure the effectiveness and effect of the technology realization;

[0115] 1. The connection between key technical features and key words:

[0116] Key words: Adjectives and nouns that are identified through sentiment analysis and word vector technology and have a high similarity to known technical advantage words. Key technical features: These key words are converted into specific technical attributes, such as reliability, efficiency, ease of use, accuracy, etc.

[0117] Detailed restrictions:

[0118] Reliability: If the key words include "stability" and "reliability", the key technical characteristics may include indicators such as long-term trouble-free operation time and failure rate.

[0119] Efficiency: If "fast" or "efficient" are the key words, the characteristics may involve processing speed, resource consumption rate, etc.

[0120] Ease of use: If “simple” and “intuitive” are the key points, then features may include user interface design, ease of operation, etc.

[0121] Precision: For terms containing the words “precise” or “accurate”, the characteristics may involve the accuracy of the measurement or calculation, the range of error, etc.

[0122] 2. The connection between technical performance indicators and key words:

[0123] Key words: As the basis for defining performance indicators, guiding the selection and setting of performance indicators

[0124] Technical performance indicators: Convert key words into measurable standards, such as response time, throughput, accuracy percentage, etc.

[0125] Detailed explanation:

[0126] Response time: For the key word "fast", the performance indicator can be the average time from the system receiving a request to giving a response.

[0127] Throughput: If "high capacity" is the keyword, the performance metric may involve the amount of data or number of transactions the system can handle per unit of time.

[0128] Precision Percentage: For key words in the “precision” category, the performance metric might be the percentage of correct recognition or classification.

[0129] Resource consumption: For the term "energy saving", the performance indicator may be the average energy consumption per unit of operation.

[0130] In some optional implementations of some embodiments, the performing of technical effect analysis according to the key words and determining the key technical features and technical performance indicators related to the technical subject based on the analysis results include:

[0131] The key word analysis function of the first model is used to deeply analyze the key words provided by the user to identify key features, performance indicators or user focus related to the technical topic, wherein the key word analysis function is used to output key points of technical improvement;

[0132] By mining the semantics of key words, combined with industry standards and user needs, we can evaluate the potential value and market impact of technology improvements;

[0133] Based on the key features, performance indicators or user focus related to the technical topic, the first model is used to automatically recommend technical improvement directions;

[0134] A decision tree algorithm is used to analyze historical patent data and industry reports to assist in determining the specific goals and expected effects of technological improvements.

[0135] Taking the "intelligent robot" technology theme as an example, the implementation steps are as follows:

[0136] By analyzing the key words “autonomous navigation” and “multimodal interaction” provided by users, we identified these two features as the key to improving intelligent robot technology.

[0137] By using the decision tree algorithm to analyze historical patents, it is found that improvements in autonomous navigation technology can enhance the robot's ability to operate in complex environments, while optimization of multimodal interaction can significantly enhance user experience. Through the above analysis, specific goals for technical improvements are determined, such as "increasing the accuracy of autonomous navigation to 99%" and "realizing at least three or more multimodal interaction methods", providing clear technical effect indicators for patent writing.

[0138] In addition, if the key words are not provided, they will be supplemented according to the technical theme. That is, if the user does not provide specific key words, supplementary recommendations are required based on the established technical themes. If the user does not provide key words, the first model recommends "endurance" and "safety" as key words based on the established "intelligent robot" as the technical theme, and analyzes the potential technical improvement direction of the key words.

[0139] S500: combining the relevant information analyzed from the subject words, background words and key words to generate a final version of prompt words that matches the format and style of the science and technology text;

[0140] In some optional implementations of some embodiments, the generating of a final version of prompt words matching the format and style of a scientific and technological text by combining the relevant information analyzed from the subject words, background words and key words includes:

[0141] Based on the technical theme, background analysis, key improvement points and expected effects, a final version of the prompt words is constructed that includes keywords, technical parameters, innovations, application scenarios and target effects.

[0142] Based on the in-depth understanding of the technical topic, background analysis, establishment of technical improvement points, and evaluation of expected effects, this step aims to form a more precise and instructive set of final prompts based on the simple prompts. The final prompts will integrate the text format and style requirements to provide a clear direction for writing the final text of the second largest model.

[0143] Final prompt word generation: Combine the technical theme, background analysis, key improvement points and expected effects to construct the final prompt word including keywords, technical parameters, innovation points, application scenarios and target effects.

[0144] Taking the "intelligent robot" technology theme as an example, the implementation steps are as follows: 1. Based on the above analysis, construct the final version of the prompt words, for example: "Intelligent robot autonomous navigation multi-modal interaction improvement, enhance the ability to operate in complex environments, enhance user experience, and the expected effects include an accuracy rate of 99% and the realization of more than three interaction methods." 2. Input this final version of the prompt words into the second largest model and specify the format and style requirements of the patent text. 3. The second largest model generates a patent application text containing a detailed technical description, an explanation of the innovative points, and an explanation of the expected effects based on the final version of the prompt words and the specified requirements.

[0145] Among them, the first large model and the second large model refer to deep learning models that are based on a large pre-trained language model and are trained with a large amount of scientific and technological text data. They can generate natural language text or understand the meaning of language text. The first large model and the second large model can also be obtained using a Transformer architecture and pre-training objectives similar to the small model. It should be noted here that the first large model and the second large model can be two large models of the same type or two large models of different types.

[0146] In the second aspect, the present application proposes a method for generating scientific and technological texts, comprising:

[0147] The final version of the prompt words described in any of the above methods is input into the second largest model, processed by natural language processing technology and combined with the specific format and style requirements of scientific and technological texts to generate corresponding scientific and technological texts.

[0148] In the third aspect, the present application proposes a system for generating prompt words related to science and technology texts based on a large model, such as Figure 2 As shown, it includes: a simple prompt word classification and extraction module, a subject word application analysis module, a background word application analysis module, a key word application analysis module and a final prompt word generation module, wherein the simple prompt word classification and extraction module is used to: receive simple prompt words input by a user, analyze the simple prompt words using the first large model, and identify and extract subject words, background words and key words therefrom;

[0149] The subject word application analysis module is used to: determine the technical subject according to the subject word, and list the main technical branches, components and innovative concepts covered based on the determined technical subject;

[0150] The background word application analysis module is used to: perform technical background analysis according to the background words, and determine the technical background, innovation direction background and time background based on the analysis results;

[0151] The key word application analysis module is used to: perform technical effect analysis based on the key words, and determine the key technical characteristics and technical performance indicators related to the technical subject based on the analysis results;

[0152] The final prompt word generation module is used to generate a final version of prompt words that matches the format and style of the science and technology text by combining the relevant information analyzed from the subject words, background words and key words.

[0153] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0154] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0155] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0156] In the embodiments provided in the present disclosure, it should be understood that the disclosed apparatus / computer equipment and methods can be implemented in other ways. For example, the apparatus / computer equipment embodiments described above are merely schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection of the apparatus or unit, which may be electrical, mechanical or other forms.

[0157] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0158] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0159] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present disclosure implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, and the computer program code may be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electric carrier signals and telecommunication signals.

[0160] The above are only preferred implementations of the present invention. It should be pointed out that a number of modifications and improved technical solutions made by those skilled in the art without departing from the technical solution should also be deemed to fall within the scope of protection required by the claims.

Claims

1. A method for generating scientific and technological texts and related prompt words based on a large model, characterized in that: The following steps are involved: A simple prompt word input by a user is received, and the simple prompt word is analyzed by using the first large model to identify and extract a subject word, a background word, and a key word, which includes a subject word extraction step, a background word analysis step, and a key word determination step. The subject word extraction step is: Leveraging the text classification capabilities of the first model, simple prompt words entered by users are matched with pre-built technical domain vocabulary libraries; The word frequency-inverse document frequency algorithm is used to identify the words with the highest relevance to the technical topic among the simple prompt words as the subject words; The background word analysis steps are: Use the textual implication relationship analysis technology of the first model to identify sentences describing the background of technological development from simple prompt words; Using time expression recognition technology, extract the time background in the simple prompt words, and combine the extracted time background with the technology development background sentence to form a complete description of the technology development background; The key word determination step is: Using the sentiment analysis and keyword extraction technology of the first model, positive adjectives and nouns in the simple prompt words are identified as potential technical effects and improvement points; Using word vector technology to compare the similarity between the words in the simple prompt words and the known technical advantage words, so as to determine the key words; Determine the technical subject according to the subject words, and list the main technical branches, components and innovative concepts covered based on the determined technical subject; Perform technical background analysis based on the background words, and determine the technical background, innovation direction background and time background based on the analysis results; Analyze the technical effects according to the key words, and determine the key technical features and technical performance indicators related to the technical subject based on the analysis results; Combining the relevant information analyzed from the subject words, background words and key words to generate a final version of prompt words that matches the format and style of the scientific and technological text; The final version of the prompt words is input into the second largest model, processed by natural language processing technology and combined with the specific format and style requirements of scientific and technological texts to generate corresponding scientific and technological texts.

2. The method according to claim 1, characterized in that: The technical subject is determined according to the subject words, and the main technical branches, components and innovative concepts covered are listed based on the determined technical subject, including: After obtaining the subject terms, the deep learning algorithm of the first model is used to conduct extensive searches and analyses of relevant technical literature, patent databases, and industry reports to determine the technical topics; Using the cluster analysis function of the first model, the technologies related to the identified technical topics are classified according to functions, application fields and implementation methods, and the main technical branches are sorted out. Each technical branch is described in detail, including working principles, technical characteristics and current research progress; According to the main technical branch, each key component of the main technical branch is identified through the identification function of the first model, and the key components are listed, and each key component is elaborated in detail to explain its role and mutual relationship in the overall technical framework, wherein the identification function is used to identify the key components from the technical subject; Through the discovery function of the first model, the latest scientific research papers, technical blogs and industry trends are tracked, innovative concepts and technological trends related to technical topics are explored, how the innovative concepts are applied to technical topics is analyzed, and their impact on the industry and potential commercial value are predicted. The discovery function is used to explore innovative concepts related to technical topics from the latest materials.

3. The method according to claim 2, characterized in that: The technical background analysis is performed according to the background words, and the technical background, innovation direction background and time background are determined based on the analysis results, including: The natural language processing technology and background analysis function of the first model are used to perform semantic understanding and association analysis on background words, thereby depicting the macro-technical background of the technical theme. The analysis content includes the limitations of existing technologies, bottlenecks in industry development, and changing trends in relevant policies and market demand. The background analysis function is used to output a detailed description of the technical background. Based on background analysis and combined with the problem identification function of the first model, the potential application scenarios of the technical theme in the future are predicted, and currently unresolved or newly emerging technical problems are identified. The classification algorithm in machine learning is used to train a large amount of industry data to distinguish and extract key application scenarios and problem points. Among them, the problem identification function is used to analyze and output potential application scenarios and problems to be solved from the technical background.

4. The method according to claim 3, characterized in that: The technical effect analysis is performed according to the key words, and the key technical features and technical performance indicators related to the technical subject are determined based on the analysis results, including: The key word analysis function of the first model is used to deeply analyze the key words provided by the user to identify key features, performance indicators or user focus related to the technical topic, wherein the key word analysis function is used to output key points of technical improvement; By mining the semantics of key words, combined with industry standards and user needs, we can evaluate the potential value and market impact of technology improvements; Based on the key features, performance indicators or user focus related to the technical topic, the first model is used to automatically recommend technical improvement directions; A decision tree algorithm is used to analyze historical patent data and industry reports to assist in determining the specific goals and expected effects of technological improvements.

5. The method according to claim 1, characterized in that: The combination of the relevant information analyzed from the subject words, background words and key words to generate the final version of prompt words matching the format and style of the scientific and technological text includes: Based on the technical theme, background analysis, key improvement points and expected effects, a final version of the prompt words is constructed that includes keywords, technical parameters, innovations, application scenarios and target effects.

6. A system for generating scientific and technological texts and related prompt words based on a large model, characterized by: It includes simple prompt word classification and extraction module, subject word application analysis module, background word application analysis module, key word application analysis module, final prompt word generation module and science and technology text generation module: The simple prompt word classification and extraction module is used to receive the simple prompt words input by the user, analyze the simple prompt words using the first large model, and identify and extract the subject words, background words and key words therefrom, which includes a subject word extraction step, a background word analysis step and a key word determination step. The subject word extraction step is: Leveraging the text classification capabilities of the first model, simple prompt words entered by users are matched with pre-built technical domain vocabulary libraries; The word frequency-inverse document frequency algorithm is used to identify the words with the highest relevance to the technical topic among the simple prompt words as the subject words; The background word analysis steps are: Use the textual implication relationship analysis technology of the first model to identify sentences describing the background of technological development from simple prompt words; Using time expression recognition technology, extract the time background in the simple prompt words, and combine the extracted time background with the technology development background sentence to form a complete description of the technology development background; The key word determination step is: Using the sentiment analysis and keyword extraction technology of the first model, positive adjectives and nouns in the simple prompt words are identified as potential technical effects and improvement points; Using word vector technology to compare the similarity between the words in the simple prompt words and the known technical advantage words, so as to determine the key words; The subject word application analysis module is used to determine the technical subject according to the subject word, and list the main technical branches, components and innovative concepts covered based on the determined technical subject; The background word application analysis module is used to perform technical background analysis based on the background words, and determine the technical background, innovation direction background and time background based on the analysis results; The key word application analysis module is used to analyze the technical effects according to the key words, and determine the key technical characteristics and technical performance indicators related to the technical subject based on the analysis results; The final prompt word generation module is used to generate a final version of prompt words that matches the format and style of the science and technology text by combining the relevant information analyzed from the subject words, background words and key words; The scientific and technological text generation module is used to input the final version of the prompt words into the second largest model, process them through natural language processing technology, and combine them with the specific format and style requirements of scientific and technological texts to generate corresponding scientific and technological texts.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images