Self-adaptive learning oil and gas geological survey professional vocabulary translation system and method

Abstract

The invention discloses an adaptive learning oil and gas geological survey professional vocabulary translation system and method, and the method comprises the steps: constructing a professional vocabulary translation model for carrying out popular interpretation on professional vocabularies, and screening out a matched query output result from a knowledge base based on the interpretation translation content of the translation model; collecting query requests of all users, corresponding query output results and feedback suggestions of each user on the query to form a data set; classifying the data set according to the category to which the professional vocabularies in the query requests belong, integrating each category of query requests and corresponding feedback suggestions into a feedback data training set, and triggering the professional vocabulary translation model to perform adaptive learning updating when the number of samples in the feedback data training set reaches a threshold value; according to the method, the query requirement of the user can be gradually improved and better served during long-term operation, and the query accuracy and the response speed are improved.

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas geological survey, and in particular to an adaptive learning oil and gas geological survey professional vocabulary translation system and method. Background Art

[0002] Oil and gas geological survey is an important part of oil and gas exploration, involving multiple disciplines such as geology, geophysics, and geochemistry. With the development of information technology, the use of artificial intelligence and natural language processing technology for oil and gas geological survey document query has become a research hotspot. These technologies are designed to improve the efficiency and accuracy of information query and help geological survey personnel quickly obtain the required geological data and technical information.

[0003] The relevant key technologies mainly include natural language processing (NLP) technology, deep learning models, knowledge base management technology, and user feedback mechanism.

[0004] In the field of oil and gas geological survey, existing document query systems usually adopt the following implementation solutions:

[0005] 1. Direct dialogue with professional vocabulary translation model + knowledge base query: After the user enters a query request, the system directly communicates through a general language model (such as BERT, GPT, etc.), extracts the keywords entered by the user, and then searches for relevant documents from the knowledge base through keyword matching.

[0006] 2. Rule matching + keyword search: Extract keywords from the questions entered by users through predefined rules and keyword lists, and then search for relevant documents from the knowledge base through keyword matching.

[0007] The technical defects of the above document query system are: poor model applicability: the existing system is usually not updated after one-time training, and cannot self-optimize with the accumulation and change of new data. Over time, the performance of the system gradually declines and cannot maintain long-term competitive advantages. Summary of the invention

[0008] The purpose of the present invention is to provide an adaptive learning oil and gas geological survey professional vocabulary translation system and method to solve the technical problem that the prior art usually does not update after one-time training and cannot perform self-optimization with the accumulation and change of new data.

[0009] In order to solve the above technical problems, the present invention specifically provides the following technical solutions:

[0010] An adaptive learning oil and gas geological survey professional vocabulary translation system, characterized by comprising:

[0011] Professional vocabulary translation model, used to translate professional vocabulary in query content into easy-to-understand content;

[0012] A user query feedback collection module is used to record each user's feedback information on the oil and gas geological survey content, wherein each user's oil and gas geological survey content includes query content and translation output content, and the oil and gas geological survey content is classified according to the keywords in the query content;

[0013] The reverse training module forms a feedback data training set based on the classification of the query content and the feedback information corresponding to each query content, and uses the feedback data training set to adjust the professional vocabulary translation model to form an adaptive learning mechanism.

[0014] As a preferred solution of the present invention, it also includes a vocabulary translation database, which includes a vocabulary table and a relationship table. The vocabulary table stores professional vocabulary and its corresponding popular explanations. The relationship table is used to store the association relationship between different professional vocabulary, and the association relationship includes synonyms, antonyms, and hyponyms.

[0015] The professional vocabulary translation model is trained based on the vocabulary translation database until the professional vocabulary translation model can interpret the professional vocabulary translation into easily understandable content.

[0016] As a preferred solution of the present invention, a synonym library is also included, and the implementation method of creating the synonym library is as follows:

[0017] Automatically identify and extract keywords with high word frequency from the user's query content, and store each keyword and its synonymous professional vocabulary in the synonym database by marking the collected keywords;

[0018] The professional vocabulary translation model calls the synonym library to convert the keywords into matching professional vocabulary according to the keywords in the identified user query content.

[0019] As a preferred solution of the present invention, the data set of the synonym library is divided into a training set, a validation set and a test set to train the professional vocabulary translation model, and the fine-tuning accuracy of the professional vocabulary translation model for the query content is evaluated by cross-validation, and the fine-tuning optimization is continuously performed until the fine-tuning accuracy of the professional vocabulary translation model for the query content meets expectations;

[0020] When the professional vocabulary translation model is trained using the training set, the relationship table corresponding to each professional vocabulary in the synonym library is called, and the synonym group of each professional vocabulary in each synonym library is added, and the professional vocabulary translation model is trained using the synonym group of each professional vocabulary and the relationship table corresponding to each professional vocabulary to increase the data diversity of the training set;

[0021] And when the fine-tuning accuracy of the professional vocabulary translation model on the verification set no longer improves, the training is stopped to avoid overfitting.

[0022] As a preferred solution of the present invention, the basic model of the professional vocabulary translation model is a pre-trained general language model;

[0023] The general language model is fine-tuned using the vocabulary translation database to enhance the recognition and translation accuracy of the professional vocabulary translation model for professional terms in the field of oil and gas geological survey.

[0024] As a preferred solution of the present invention, the user query feedback collection module includes a user feedback log unit, which is used to save the user's query content, query output results and user feedback, and organize them into a structured format to ensure that each type of query content has corresponding feedback, and form a feedback data training set based on the classification of the query content and the feedback information corresponding to each query content, and use the feedback information of each query content as a label for the data;

[0025] The feedback classification unit automatically classifies the query content using a classification algorithm based on the professional vocabulary in the user's query content;

[0026] The data cleaning unit is used to clean the query content in the user feedback log unit, remove invalid or duplicate data, and ensure the integrity and consistency of the feedback data training set.

[0027] As a preferred solution of the present invention, the reverse training module uses the feedback data training set to adjust the professional vocabulary translation model in the following manner:

[0028] Collect user query content and feedback information of each query from the document library, expert review, and internal project data collection data, and trigger the update of the professional vocabulary translation model in real time when the data set reaches a threshold;

[0029] Using fine-tuning technology to train the professional vocabulary translation model, and after the training is completed, using the vocabulary table of the vocabulary translation database to evaluate the performance of the professional vocabulary translation model;

[0030] The parameters of the professional vocabulary translation model are adjusted based on the performance evaluation result until the performance evaluation result of the professional vocabulary translation model is better than that of the previous professional vocabulary translation model.

[0031] In addition, the present invention also provides an adaptive learning method for translating professional vocabulary for oil and gas geological survey, comprising the following steps:

[0032] Step 100: construct a professional vocabulary translation model for popular interpretation of professional vocabulary, wherein the professional vocabulary translation model translates the professional vocabulary interpretation into easy-to-understand content, and based on the interpretation and translation content of the translation model, selects query output results matching the interpretation and translation content from the knowledge base;

[0033] Step 200: collect all users' query requests, corresponding query output results, and each user's feedback on the query to form a data set;

[0034] Step 300: classify the data set according to the category to which the professional vocabulary in the query request belongs, integrate each type of query request and the corresponding feedback into a feedback data training set, and when the number of samples in the feedback data training set reaches a threshold, trigger the professional vocabulary translation model to perform adaptive learning and updating.

[0035] As a preferred solution of the present invention, in step 200, the query request includes query content, timestamp, and user ID information;

[0036] The feedback from each user on the query includes satisfaction, errors and improvement suggestions. The query request is secondary classified according to positive feedback and negative feedback, and the query request with negative feedback is used to perform adaptive learning on the professional vocabulary translation model.

[0037] As a preferred solution of the present invention, after training the professional vocabulary translation model using the feedback data training set, the performance of the professional vocabulary translation model is evaluated using the vocabulary table of the vocabulary translation database, wherein the performance indicators include accuracy, recall rate, and F1 score;

[0038] Determine whether to update the model based on the evaluation result, wherein when the performance of the trained professional vocabulary translation model is better than the old professional vocabulary translation model, the old professional vocabulary translation model is replaced; otherwise, the old professional vocabulary translation model is retained.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] (1) Adaptive learning capability: The system has adaptive learning capability and can continuously update and fine-tune the professional vocabulary interpretation model to improve its performance by collecting user query requests and their feedback. A closed-loop mechanism for data processing and feedback is established to ensure that each user interaction provides a new learning opportunity for the model. The system can gradually improve in the long run, better serve the user's query needs, and improve query accuracy and response speed.

[0041] (2) Dynamic update to improve work efficiency: Through the self-training mechanism and continuous learning framework model, ensure that the system can adapt to the latest developments in the field of oil and gas geological surveys, provide the latest translation services, and maintain the system's advancement and practicality.

[0042] (3) Improve user experience: The system supports a user feedback mechanism, which continuously improves the system’s query accuracy and response speed through user feedback to better meet user needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0044] Figure 1 A schematic diagram of a process flow of a method for translating professional vocabulary for oil and gas geological survey according to an embodiment of the present invention; DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] like Figure 1 As shown, the present invention provides an adaptive learning oil and gas geological survey professional vocabulary translation system, comprising:

[0047] Professional vocabulary translation model, used to translate professional vocabulary in query content into easy-to-understand content;

[0048] A user query feedback collection module is used to record each user's feedback information on the oil and gas geological survey content, wherein each user's oil and gas geological survey content includes query content and translation output content, and the oil and gas geological survey content is classified according to the keywords in the query content;

[0049] The reverse training module forms a feedback data training set based on the classification of query content and the feedback information corresponding to each query content, and uses the feedback data training set to adjust the professional vocabulary translation model to form an adaptive learning mechanism.

[0050] Among them, the basic model of the professional vocabulary translation model is a pre-trained general language model. When constructing the professional vocabulary translation model, the vocabulary translation database is used to train the basic general language model, and the literature in the field of oil and gas geological survey is used to fine-tune the basic model to enhance the model's recognition and translation accuracy of professional terms.

[0051] The vocabulary translation database includes a vocabulary table and a relationship table. The vocabulary table stores professional vocabulary and its corresponding popular explanations. The relationship table is used to store the associations between different professional vocabulary, including synonyms, antonyms, and hyponyms.

[0052] The professional vocabulary translation model is trained based on the vocabulary translation database until the professional vocabulary translation model can interpret the professional vocabulary translation into easily understandable content.

[0053] When constructing the vocabulary translation database, the specific method of collecting professional vocabulary is as follows: using academic search engines (such as Google Scholar, Web of Science) and professional databases (such as IEEE Xplore, SpringerLink) to search for relevant research reports, academic papers and technical manuals, and collecting a large amount of literature in the field of oil and gas geological surveys, including but not limited to professional basic training data such as research reports and white papers in the oil and gas industry. Ensure that the data covers a wide range of topics, including but not limited to geological structure, sedimentology, geochemistry, exploration technology, reservoir engineering, etc.

[0054] The data standards of the above-mentioned large amount of literature in the field of oil and gas geological survey were carried out, including professional vocabulary annotation such as "sandstone reservoir", "carbonate gas reservoir", "fault", etc., keywords, keyword diversity and context relevance, where each word or sub-word is marked as keyword (1) or non-keyword (0).

[0055] Clean the data of professional vocabulary and use regular expressions to remove punctuation marks, special characters, etc. in the text. Combine hash algorithms and set operations to detect and remove duplicate data and perform formatting to ensure text quality and avoid duplicate data affecting model training results.

[0056] Use the NLTK library and spaCy library for word segmentation to divide the text into words or phrases. Add custom word segmentation rules based on the professional vocabulary in the field of oil and gas geology, such as treating "sandstone reservoir" as a whole word.

[0057] Furthermore, when extracting professional vocabulary from the user's query content and translating the professional vocabulary into easy-to-understand content, it is necessary to use the professional vocabulary translation model to perform keyword fine-tuning training, that is, to adjust the keywords in the user's query content that are synonymous with professional terms so that the extracted keywords can match the vocabulary of the vocabulary translation database.

[0058] In order to achieve fine-tuning training of the professional vocabulary translation model for the keywords in the query content, the professional vocabulary translation model is trained using a synonym library. The implementation method of creating a synonym library is as follows:

[0059] Automatically identify and extract high-frequency keywords from user query content, annotate the collected keywords, and save each keyword and its synonymous professional vocabulary in a synonym database;

[0060] The professional vocabulary translation model calls the synonym library to convert the keywords into matching professional vocabulary based on the keywords in the identified user query content.

[0061] Use natural language processing technology (such as named entity recognition (NER) technology to identify proper nouns such as names of people, places, and organizations, TF-IDF, TextRank and other algorithms to extract keywords from documents, and use topic models such as LDA (Latent Dirichlet Allocation) to identify the subject and keywords of documents, etc.) to automatically identify and extract frequently occurring professional terms and keywords from user query records. Combined with expert knowledge, manually collected questions are annotated to identify the keywords and professional terms, and a vocabulary containing common terms and their synonyms is manually constructed to expand a term into multiple related terms. For example, word embedding models (such as Word2Vec, GloVe, etc.) are used to automatically expand the synonym library.

[0062] For the keywords in the user's query, we analyze their usage in different contexts, and classify and replace words with similar meanings but different expressions. For example, "geological structure" can be fine-tuned to "structural characteristics", "fault analysis" or "fold research"; "oil and gas reservoir type" can be fine-tuned to "sandstone reservoir", "carbonate gas reservoir" or "unconventional oil and gas reservoir"; "resource assessment" can be fine-tuned to "reserve calculation", "economic evaluation" or "development potential analysis". By building a synonym library or using word embedding models (such as Word2Vec, GloVe, etc.) to achieve semantic similarity matching between words, the fine-tuning of keywords can be completed.

[0063] When training a professional vocabulary translation model, the dataset of the synonym library is divided into a training set, a validation set, and a test set to train the professional vocabulary translation model. The fine-tuning accuracy of the professional vocabulary translation model for the query content is evaluated through cross-validation, and adjustments and optimizations are continuously made until the fine-tuning accuracy of the professional vocabulary translation model for the query content meets expectations.

[0064] When the professional vocabulary translation model is trained using the training set, the relationship table corresponding to each professional vocabulary in the synonym library is called, and the synonym group of each professional vocabulary in each synonym library is added. The professional vocabulary translation model is trained using the synonym group of each professional vocabulary and the relationship table corresponding to each professional vocabulary to increase the data diversity of the training set.

[0065] And stop training when the fine-tuning accuracy of the professional vocabulary translation model on the validation set no longer improves to avoid overfitting.

[0066] When fine-tuning the keywords in the query content using the professional vocabulary translation model in this embodiment, a data set of 1,000 oil and gas geological reports is used, and the training set (700 samples) is randomly divided into K subsets (usually K = 5 or 10), and each subset is used as a validation set in turn, and the remaining K-1 subsets are used as training sets, and K training and validation are performed. The test results show an average accuracy rate of 0.854, an average precision rate of 0.872, an average recall rate of 0.834, and an average F1 value of 0.854.

[0067] Furthermore, the user query feedback collection module includes a user feedback log unit, which is used to save the user's query content, query output results and user feedback, and organize them into a structured format to ensure that each type of query content has corresponding feedback. A feedback data training set is formed based on the classification of the query content and the feedback information corresponding to each query content, and the feedback information of each query content is used as a label for the data.

[0068] The feedback classification unit uses a classification algorithm to automatically classify the query content based on the professional vocabulary in the user's query content.

[0069] The data cleaning unit is used to clean the query content in the user feedback log unit, remove invalid or duplicate data, and ensure the integrity and consistency of the feedback data training set.

[0070] Collect user query requests, query output results, and user feedback from the logs. Organize the collected data into a structured format to ensure that each query request has corresponding feedback. Remove invalid or duplicate data to ensure data quality.

[0071] Use classification algorithms (such as support vector machines, random forests, deep learning models, etc.) to automatically classify query requests in order to better manage and utilize the data. Manual review of automatic classification results is performed to ensure the accuracy of classification. For example, query requests can be classified into categories such as "geological feature analysis", "reservoir evaluation", "drilling engineering", and "production management".

[0072] The user feedback corresponding to each query request is integrated into the training data set to form labeled training samples. The integrated data is cleaned to remove invalid or duplicate data to ensure the integrity and consistency of the data set. The number of samples in each category is regularly counted to ensure the balance of the data.

[0073] Set the trigger condition. When the number of samples in the training data set reaches the threshold, the model's adaptive learning update is triggered. For example, when the number of samples in a certain category reaches 1,000, the model's adaptive learning update is triggered.

[0074] The reverse training module uses the feedback data training set to adjust the professional vocabulary translation model in the following way:

[0075] Collect user query content and feedback information for each query from the literature database, expert review, and internal project data collection. When the data set reaches the threshold, the professional vocabulary translation model is updated in real time.

[0076] The professional vocabulary translation model is trained using fine-tuning technology. After the training is completed, the vocabulary table of the vocabulary translation database is used to evaluate the performance of the professional vocabulary translation model.

[0077] Based on the performance evaluation result, the parameters of the professional vocabulary translation model are adjusted until the performance evaluation result of the professional vocabulary translation model is better than that of the previous professional vocabulary translation model.

[0078] Use fine-tuning techniques to train the oil and gas geological survey model, such as using pre-trained models such as BERT and Bi-LSTM. After training, use the validation set to evaluate the model's performance, including indicators such as accuracy, recall, and F1 score. Finally, based on the evaluation results, decide whether to update the model. If the new model performs better than the old model, replace the old model; otherwise, keep the old model.

[0079] In summary, this embodiment provides an adaptive learning oil and gas geological survey professional vocabulary translation model, including a general language model, which can interpret professional vocabulary in the field of oil and gas geological survey; a user query feedback collection module, which is used to record each user's feedback information on the query translation content, wherein each user's query translation content includes query content and translation output content, and the query translation content is classified according to the keywords in the query content; a training data set is formed based on the classification of the query content and the feedback information corresponding to each query content, and the general language model is adjusted using the training data set to form an adaptive learning mechanism.

[0080] In addition, the present invention also provides an adaptive learning method for an adaptive learning oil and gas geological survey professional vocabulary translation system, comprising the following steps:

[0081] Step 100: construct a professional vocabulary translation model for popular interpretation of professional vocabulary, the professional vocabulary translation model translates the professional vocabulary interpretation into easy-to-understand content, and based on the interpretation and translation content of the translation model, select the query output results matching it from the knowledge base;

[0082] Step 200: collect all users' query requests, corresponding query output results, and each user's feedback on the query to form a data set;

[0083] In step 200, the query request includes query content, timestamp, and user ID information;

[0084] Each user's feedback on the query includes satisfaction, errors and improvement suggestions. The query requests are secondary classified into positive feedback and negative feedback, and the query requests with negative feedback are used to perform adaptive learning on the professional vocabulary translation model.

[0085] Step 300: classify the data set according to the category to which the professional vocabulary in the query request belongs, integrate each type of query request and the corresponding feedback into the feedback data training set, and when the number of samples in the feedback data training set reaches a threshold, trigger the professional vocabulary translation model to perform adaptive learning and updating.

[0086] After training the professional vocabulary translation model using the feedback data training set, the performance of the professional vocabulary translation model is evaluated using the vocabulary table of the vocabulary translation database, where the performance indicators include precision, recall, and F1 score;

[0087] Determine whether to update the model based on the evaluation result, wherein when the performance of the trained professional vocabulary translation model is better than the old professional vocabulary translation model, the old professional vocabulary translation model is replaced; otherwise, the old professional vocabulary translation model is retained.

[0088] In the above-mentioned adaptive learning oil and gas geological survey professional vocabulary translation model, after one month of adaptive learning, the system's query accuracy increased from 90% to 96%, an increase of 6 percentage points. The response time was reduced from 2.4 seconds to 1.7 seconds, an increase of 29%.

[0089] Therefore, the oil and gas geological survey professional vocabulary translation system provided by this embodiment has adaptive learning capabilities, and can continuously update and fine-tune the professional vocabulary interpretation model to improve its performance by collecting user query requests and their feedback. A closed-loop mechanism for data processing and feedback is established to ensure that each user interaction provides a new learning opportunity for the model. The system can gradually improve in long-term operation, better serve the user's query needs, and improve query accuracy and response speed.

[0090] Through the self-training mechanism and continuous learning framework model, the system can adapt to the latest developments in the field of oil and gas geological surveys, provide the latest translation services, and maintain the system's advancement and practicality.

[0091] It also supports a user feedback mechanism, which continuously improves through user feedback, thereby improving the system's query accuracy and response speed to better meet user needs.

[0092] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.

Claims

1. An adaptive learning oil and gas geological survey professional vocabulary translation system, characterized in that: include: Professional vocabulary translation model, used to translate professional vocabulary in query content into easy-to-understand content; A user query feedback collection module is used to record each user's feedback information on the oil and gas geological survey content, wherein each user's oil and gas geological survey content includes query content and translation output content, and the oil and gas geological survey content is classified according to the keywords in the query content; The reverse training module forms a feedback data training set based on the classification of the query content and the feedback information corresponding to each query content, and uses the feedback data training set to adjust the professional vocabulary translation model to form an adaptive learning mechanism.

2. The adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 1 is characterized in that: It also includes a vocabulary translation database, which includes a vocabulary table and a relationship table. The vocabulary table stores professional vocabulary and its corresponding popular explanations. The relationship table is used to store the association relationship between different professional vocabulary, and the association relationship includes synonyms, antonyms, and hyponyms. The professional vocabulary translation model is trained based on the vocabulary translation database until the professional vocabulary translation model can interpret the professional vocabulary translation into easily understandable content.

3. The adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 2 is characterized in that: It also includes a synonym library, and the implementation method of creating the synonym library is as follows: Automatically identify and extract keywords with high word frequency from the user's query content, and store each keyword and its synonymous professional vocabulary in the synonym database by marking the collected keywords; The professional vocabulary translation model calls the synonym library to convert the keywords into matching professional vocabulary according to the keywords in the identified user query content.

4. The adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 3 is characterized in that: The data set of the synonym library is divided into a training set, a validation set and a test set to train the professional vocabulary translation model, and the fine-tuning accuracy of the professional vocabulary translation model for the query content is evaluated by cross-validation, and the fine-tuning accuracy of the professional vocabulary translation model for the query content is continuously adjusted and optimized until the fine-tuning accuracy of the professional vocabulary translation model for the query content meets expectations; When the professional vocabulary translation model is trained using the training set, the relationship table corresponding to each professional vocabulary in the synonym library is called, and the synonym group of each professional vocabulary in each synonym library is added, and the professional vocabulary translation model is trained using the synonym group of each professional vocabulary and the relationship table corresponding to each professional vocabulary to increase the data diversity of the training set; And when the fine-tuning accuracy of the professional vocabulary translation model on the verification set no longer improves, the training is stopped to avoid overfitting.

5. The adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 1 is characterized in that: The basic model of the professional vocabulary translation model is a pre-trained general language model; The general language model is fine-tuned using the vocabulary translation database to enhance the recognition and translation accuracy of the professional vocabulary translation model for professional terms in the field of oil and gas geological survey.

6. The adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 1 is characterized in that: The user query feedback collection module includes a user feedback log unit, which is used to save the user's query content, query output results and user feedback, and organize them into a structured format to ensure that each type of query content has corresponding feedback. Based on the classification of the query content and the feedback information corresponding to each query content, a feedback data training set is formed, and the feedback information of each query content is used as a label for the data; The feedback classification unit automatically classifies the query content using a classification algorithm based on the professional vocabulary in the user's query content; The data cleaning unit is used to clean the query content in the user feedback log unit, remove invalid or duplicate data, and ensure the integrity and consistency of the feedback data training set.

7. The adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 1 is characterized in that: The reverse training module uses the feedback data training set to adjust the professional vocabulary translation model in the following manner: Collect user query content and feedback information of each query from the document library, expert review, and internal project data collection data, and trigger the update of the professional vocabulary translation model in real time when the data set reaches a threshold; Using fine-tuning technology to train the professional vocabulary translation model, and after the training is completed, using the vocabulary table of the vocabulary translation database to evaluate the performance of the professional vocabulary translation model; The parameters of the professional vocabulary translation model are adjusted based on the performance evaluation result until the performance evaluation result of the professional vocabulary translation model is better than that of the previous professional vocabulary translation model.

8. An adaptive learning method for a professional vocabulary translation system for oil and gas geological survey based on the adaptive learning described in any one of claims 2 to 7, characterized in that: The following steps are involved: Step 100: construct a professional vocabulary translation model for popular interpretation of professional vocabulary, wherein the professional vocabulary translation model translates the professional vocabulary interpretation into easy-to-understand content, and based on the interpretation and translation content of the translation model, selects query output results matching the interpretation and translation content from the knowledge base; Step 200: collect all users' query requests, corresponding query output results, and each user's feedback on the query to form a data set; Step 300: classify the data set according to the category to which the professional vocabulary in the query request belongs, integrate each type of query request and the corresponding feedback into a feedback data training set, and when the number of samples in the feedback data training set reaches a threshold, trigger the professional vocabulary translation model to perform adaptive learning and updating.

9. The adaptive learning method of the adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 8, characterized in that: In the step 200, the query request includes query content, timestamp, and user ID information; The feedback from each user on the query includes satisfaction, errors and improvement suggestions. The query request is secondary classified according to positive feedback and negative feedback, and the query request with negative feedback is used to perform adaptive learning on the professional vocabulary translation model.

10. The adaptive learning method of the adaptive learning oil and gas geological survey professional vocabulary translation system according to claim 8, characterized in that: After training the professional vocabulary translation model using the feedback data training set, the performance of the professional vocabulary translation model is evaluated using the vocabulary table of the vocabulary translation database, wherein the performance indicators include accuracy, recall rate, and F1 score; Determine whether to update the model based on the evaluation result, wherein when the performance of the trained professional vocabulary translation model is better than the old professional vocabulary translation model, the old professional vocabulary translation model is replaced; otherwise, the old professional vocabulary translation model is retained.

Images