Automatic text generation method based on large model and big data platform

Abstract

The invention relates to the technical field of text data generation, in particular to an automatic text generation method based on a large model and a large data platform, and the method comprises the following steps: analyzing the membership degree between each vocabulary in each field and different clustering centers, and combining the similarity of vocabulary vectors of any two vocabularies between a source field and a target field to obtain a clustering center; determining a vocabulary correlation degree; based on the vocabulary association degree of any two clustering centers between the source domain and the target domain, and in combination with the probability distribution of any two clustering centers under different topics, determining the topic association degree; and determining the difference degree by analyzing the difference of the membership degree vectors of all the clusters between the source field and the target field, and carrying out transfer learning on the source field text data set to the target field in combination with the topic correlation degree. The invention aims to optimize a transfer learning mode and improve the quality and accuracy of the target generated text.

Description

Technical Field

[0001] The present application relates to the technical field of text data generation, and in particular to a large model-based automatic text generation method and a big data platform. Background Art

[0002] Automatic text generation is the process of automatically generating output content based on given input. Typical automatic text generation tasks include machine translation, text summarization, and conversation tasks. The large text generation model trained by deep neural networks has excellent performance in some limited fields, but when the data set or field changes, the text generation performance of the large model often drops drastically, and the neural network model requires a large amount of data to support its training. Retraining the large model in a new field requires a huge amount of time and computing resources. Therefore, the large text generation model is highly sensitive to changes in data sets and fields, making it perform poorly in cross-domain text generation.

[0003] Transfer learning is a commonly used machine learning method. By applying the task parameters obtained from training to new tasks, the knowledge learned by the model in the field is transferred to other tasks to improve the training efficiency and performance of new tasks. The more mature text automatic generation method in the industry is to fine-tune the downstream target domain tasks in the pre-trained text generation model in the source domain. In the prior art, transfer learning methods make little use of the intrinsic structure and semantic information correlation between text data in different fields. As the number of parameters of the pre-trained text generation model in the source domain increases, it is difficult to align the text features of the pre-training task and the fine-tuning task during the transfer learning process. The text generation model cannot fully utilize the knowledge learned in the pre-training stage in the source domain, thereby reducing the quality and accuracy of the target generated text. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a text automatic generation method and a big data platform based on a large model. The technical solutions adopted are as follows: In a first aspect, an embodiment of the present application provides a method for automatically generating text based on a large model, the method comprising the following steps: S1: In the big data platform, obtain text datasets in various fields and the topic-vocabulary distribution of the text datasets in various fields, and use the probability of each word under all topics to form the vocabulary vector of each word, where each field includes a source field and a target field; S2: Cluster all words in the text data set of each field, analyze the degree of membership between each word in each field and different cluster centers, determine the semantic relevance of each word in each field, and combine the similarity of the word vectors of any two words between the source field and the target field to determine the word relevance between any two words between the source field and the target field; S3: Based on the vocabulary association between any two cluster centers between the source domain and the target domain, and in combination with the probability distribution of the any two cluster centers under different topics, determine the topic association between different topics between the source domain and the target domain; S4: Based on the membership, obtaining the membership vector of each cluster in each field; determining the difference between the source field and the target field by analyzing the difference of the membership vectors of all clusters between the source field and the target field; S5: Based on the topic relevance and the difference, determine the attention weights of different topics between the source domain and the target domain, and transfer the source domain text dataset to the target domain.

[0005] Preferably, the fuzzy C-means clustering algorithm is used for clustering all vocabulary of text data sets in various fields, and the Euclidean distance between the vocabulary vector of the vocabulary and the vocabulary vector of the cluster center is used as the distance between the sample point and the cluster center in the objective function of the fuzzy C-means clustering algorithm.

[0006] Preferably, the method for determining the semantic relevance of each word in each field is: In each field, the mean membership between each word and all cluster centers is calculated. Among the memberships between each word and all cluster centers, the membership that is greater than the mean membership is obtained and recorded as the feature membership. The cumulative sum of all feature memberships is taken as the semantic relevance of each word in each field.

[0007] Preferably, the expression of the vocabulary association between any two words in the source domain and the target domain is: ; In the formula, represents the lexical association between the i-th word in the source domain and the j-th word in the target domain; Represents the similarity between the vocabulary vector of the i-th word in the source domain and the vocabulary vector of the j-th word in the target domain; represents the difference in semantic relevance between the i-th word in the source domain and the j-th word in the target domain; Indicates a preset constant greater than 0.

[0008] Preferably, the method for determining the subject relevance of different subjects between the source field and the target field is as follows: All cluster centers are recorded as central words, and the topic relevance between topic m in the source domain and topic n in the target domain is The expression is: ; In the formula, represents the lexical relevance between the p-th central word in the source domain and the q-th central word in the target domain; represents the difference between the probability of the pth central word under topic m in the source domain and the probability of the qth central word under topic n in the target domain; represents the number of all central words in the target domain; Indicates a preset constant greater than 0.

[0009] Preferably, the membership vector of each cluster in each field is composed of the membership between all words in each cluster in each field and the cluster center of the corresponding cluster.

[0010] Preferably, the expression of the difference between the source domain and the target domain is: ; In the formula, Indicates the difference between the source domain and the target domain; represents the difference between the membership vector of the hth cluster in the source domain and the membership vector of the kth cluster in the target domain; N represents the number of all clusters in the target domain; norm() represents the normalization function.

[0011] Preferably, the process of determining the attention weights of different topics between the source domain and the target domain is as follows: The attention weight between topic m in the source domain and topic n in the target domain The expression is: ; In the formula, Indicates the difference between the source domain and the target domain; represents the topic relevance between topic m in the source domain and topic n in the target domain; exp( ) represents an exponential function with a natural constant as the base; Represents the normalization function.

[0012] Preferably, the step of transferring the source domain text dataset to the target domain includes: Constructing the attention mechanism layer in the feature alignment process in transfer learning through the attention weights of all topics between the source domain and the target domain; The BERT model is used to train the source domain text dataset to obtain a pre-trained model; the source domain text dataset, the target domain text dataset and the pre-trained model are used as inputs of the transfer learning technology, wherein the attention mechanism layer is used to align the features of the text datasets in the source domain and the target domain, and the text generation model adapted to the target domain is output.

[0013] In the second aspect, an embodiment of the present application also provides a big data platform for automatic text generation based on a large model, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, it implements the steps of any one of the above-mentioned methods for automatic text generation based on a large model.

[0014] This application has at least the following beneficial effects: This application constructs word association by analyzing the similarity of the vocabulary vectors of any vocabulary between the source domain and the target domain, as well as the semantic association of the vocabulary, and enhances the model's ability to understand the semantics of vocabulary in the transfer learning process, and can better capture the association between the vocabulary in the source domain and the target domain, thereby improving the efficiency and efficiency of transfer learning; further, by synthesizing the vocabulary association of any two cluster centers between the source domain and the target domain, and combining the probability distribution of the two cluster centers under different topics, the topic association is constructed, which helps the model understand the topic relationship between the source domain and the target domain, optimizes the strategy selection in the transfer learning process, and improves the quality of text generation; further, by analyzing the difference in the membership vectors of all cluster clusters between the source domain and the target domain, the difference is constructed, which helps the model identify the difference between the source domain and the target domain, thereby adjusting the transfer learning strategy and enhancing the adaptability of the model in the target domain; finally, the topic association and difference are combined to construct the attention weight, which helps the model better align the text features of the source domain and the target domain during the transfer learning process, and improves the quality and accuracy of the target domain text generation. This application optimizes the transfer learning method by analyzing the semantic features of the text, and improves the quality and accuracy of the target generated text. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 A flowchart of the steps of a method for automatically generating text based on a large model provided in one embodiment of the present application; Figure 2 A schematic diagram of the attention weight extraction process provided for one embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of the text automatic generation method based on the big model and the big data platform proposed in the present application in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0019] The specific scheme of the automatic text generation method based on a large model and the big data platform provided by this application is described in detail below with reference to the accompanying drawings.

[0020] See also Figure 1 , which shows a flowchart of a method for automatic text generation based on a large model provided by an embodiment of the present application, the method comprising the following steps: S1: In the big data platform, obtain text datasets in various fields and the topic-vocabulary distribution of the text datasets in various fields, and use the probability of each word under all topics to form the vocabulary vector of each word, where each field includes a source field and a target field.

[0021] This embodiment uses a big data platform as a carrier, and utilizes the text data sets in different fields stored therein to implement the training of the text generation model through a big data processing framework and a deep learning framework. The topic modeling representation method of text data can effectively reduce the feature dimension of the text, thereby simplifying the subsequent processing and analysis process. However, the topic model does not adequately analyze the semantic association between words in different topics, and cannot handle the problems of "multiple words for one meaning" and "multiple meanings for one word", making it difficult to completely transfer the knowledge acquired by the model in the source field training to the target field.

[0022] In this embodiment, text datasets in various fields are obtained from a big data platform, and topic modeling is performed on the text datasets in various fields using an LDA model to obtain the topic-vocabulary distribution of the text datasets in various fields, wherein the topic-vocabulary distribution reflects the probability of different topics under a vocabulary, and therefore, further, the probability of each vocabulary under all topics is used to form a vocabulary vector for each vocabulary. In this embodiment, each field includes a source field and a target field.

[0023] The principle of the LDA model is a well-known technology, and the several-day process of using the LDA model to perform topic modeling and obtain topic-vocabulary distribution will not be repeated.

[0024] S2: Cluster all the words in the text datasets in each field, analyze the degree of membership between each word in each field and different cluster centers, determine the semantic relevance of each word in each field, and combine the similarity of the word vectors of any two words between the source field and the target field to determine the word relevance between any two words between the source field and the target field.

[0025] At present, when processing text data, transfer learning methods usually ignore the complex correlation between words in different topics in text data. This correlation is not just the direct connection between words, but covers the semantic correlation and interaction of words under multiple topics. Although the simple correlation of words, that is, the direct connection between words, can provide certain semantic information in some cases, in text data aggregation, this correlation is often not enough to capture the deep semantic structure and subtle differences between topics. In fact, the word correlation in text data carries rich semantic information, which is crucial to improving the performance of transfer learning.

[0026] Therefore, in order to better explore the vocabulary relevance between the source domain and the target domain, the fuzzy C-means clustering (FCM) algorithm is used to cluster all the words in the text data sets of each domain, and multiple clusters and the membership between each word in each domain and different cluster centers are obtained; Among them, the fuzzy C-means clustering algorithm is a well-known technology. The Euclidean distance between the vocabulary vector of the vocabulary and the vocabulary vector of the cluster center is used as the distance between the sample point and the cluster center in the objective function of the fuzzy C-means clustering algorithm. The specific principle process of the fuzzy C-means clustering algorithm and the specific calculation process of the Euclidean distance are not repeated here.

[0027] Furthermore, in each field, the mean membership between each word and all cluster centers is calculated, and among the memberships between each word and all cluster centers, the membership that is greater than the mean membership is obtained and recorded as the feature membership. The cumulative sum of all feature memberships is taken as the semantic relevance of each word in each field.

[0028] Furthermore, by analyzing the similarity of the vocabulary vectors of any two words between the source domain and the target domain, combined with the semantic relevance, the vocabulary relevance of any two words between the source domain and the target domain is determined, specifically: The lexical relevance between the i-th word in the source domain and the j-th word in the target domain The expression is: ; In the formula, Represents the similarity between the vocabulary vector of the i-th word in the source domain and the vocabulary vector of the j-th word in the target domain; represents the difference in semantic relevance between the i-th word in the source domain and the j-th word in the target domain; Indicates a constant greater than 0 to prevent the denominator from being 0. The value of is artificially set. The value of is 0.01. Under the premise of ensuring that the denominator is not 0 and does not excessively affect the calculation result, the implementer can also set it according to the specific situation. This embodiment does not impose any special restrictions.

[0029] According to the lexical association between any two words in the source domain and the target domain, it can be understood that: if the similarity between the lexical vector of the i-th word in the source domain and the lexical vector of the j-th word in the target domain is greater, the difference in semantic association between the i-th word in the source domain and the j-th word in the target domain is smaller, then the lexical association between the i-th word in the source domain and the j-th word in the target domain is greater, indicating that the association between the words in the source domain and the target domain is greater; conversely, if the similarity between the lexical vector of the i-th word in the source domain and the lexical vector of the j-th word in the target domain is smaller, the difference in semantic association between the i-th word in the source domain and the j-th word in the target domain is greater, then the lexical association between the i-th word in the source domain and the j-th word in the target domain is smaller, indicating that the association between the words in the source domain and the target domain is smaller.

[0030] S3: Based on the vocabulary association between any two cluster centers between the source domain and the target domain, and in combination with the probability distribution of the any two cluster centers under different topics, determine the topic association between different topics between the source domain and the target domain.

[0031] For ease of understanding, the words corresponding to the cluster center are recorded as central words. The central words are keywords in the context represented by the corresponding cluster clusters. The frequency of different central words in the same topic represents the relevance between the topic and the contexts corresponding to different central words. The greater the vocabulary relevance between different central words across fields, the stronger the correlation between the corresponding words they represent. Therefore, the relevance of the two central words is used as the weight to calculate the topic relevance of different topics between the source field and the target field, specifically: The topic relevance between topic m in the source domain and topic n in the target domain The expression is: ; In the formula, represents the lexical relevance between the p-th central word in the source domain and the q-th central word in the target domain; represents the difference between the probability of the pth central word under topic m in the source domain and the probability of the qth central word under topic n in the target domain; represents the number of all central words in the target domain; Indicates a constant greater than 0 to prevent the denominator from being 0. The value of is artificially set. The value of is 0.01. Under the premise of ensuring that the denominator is not 0 and does not excessively affect the calculation result, the implementer can also set it according to the specific situation. This embodiment does not impose any special restrictions.

[0032] Among them, the source domain text dataset must be greater than or equal to the target domain text dataset.

[0033] According to the topic correlation analysis of different topics between the source domain and the target domain, it is found that: if the lexical correlation between the p-th central word in the source domain and the q-th central word in the target domain is greater, the difference between the probability of the p-th central word under topic m in the source domain and the probability of the q-th central word under topic n in the target domain is smaller, then the topic correlation between topic m in the source domain and topic n in the target domain is greater, indicating that the topic correlation between the source domain and the target domain is greater, which is more conducive to text transfer; conversely, if the lexical correlation between the p-th central word in the source domain and the q-th central word in the target domain is smaller, the difference between the probability of the p-th central word under topic m in the source domain and the probability of the q-th central word under topic n in the target domain is larger, then the topic correlation between topic m in the source domain and topic n in the target domain is smaller, indicating that the topic correlation between the source domain and the target domain is smaller.

[0034] S4: Based on the membership, the membership vector of each cluster in each field is obtained; by analyzing the difference of the membership vectors of all clusters between the source field and the target field, the difference between the source field and the target field is determined.

[0035] When there are large differences in text style, vocabulary usage, etc. between the source domain and the target domain, it is difficult to align the topic features across domains, resulting in insufficient adaptability of the pre-trained text generation model in the source domain in the target domain. According to the vocabulary association and the topic association, an attention mechanism is added during the training process of transfer learning, and the alignment information of different topics across domains is automatically learned through the attention mechanism, thereby enhancing the model's use of the internal structure and semantic information of the text across domains, thereby improving the model's text generation performance. The specific process is as follows: The membership degree between all words in each cluster in each field and the cluster center of the corresponding cluster is used to form the membership vector of each cluster in each field.

[0036] The difference between the source domain and the target domain The expression is: ; In the formula, represents the difference between the membership vector of the hth cluster in the source domain and the membership vector of the kth cluster in the target domain; N represents the number of all clusters in the target domain; norm() represents the normalization function.

[0037] It should be noted that there are many methods for measuring the difference between vectors. In this embodiment, the DTW distance between the membership vector of the hth cluster in the source domain and the membership vector of the kth cluster in the target domain is used as the difference between the membership vector of the hth cluster in the source domain and the membership vector of the kth cluster in the target domain. In actual application, as other implementation methods, implementers may also adopt other methods for measuring the difference between vectors, such as Euclidean distance or Manhattan distance. This embodiment does not impose any special restrictions on the selection of methods for measuring the difference between vectors.

[0038] The calculation method of the DTW distance is a well-known technology, and its specific calculation process will not be described in detail.

[0039] It can be understood from the difference between the source domain and the target domain that if the difference between the membership vector of the hth cluster in the source domain and the membership vector of the kth cluster in the target domain is smaller, the difference between the source domain and the target domain is greater. The smaller the value, the smaller the text style difference between the source domain and the target domain, the greater the correlation, and the higher the accuracy of text migration; conversely, if the difference between the membership vector of the hth cluster in the source domain and the membership vector of the kth cluster in the target domain is greater, the difference between the source domain and the target domain is greater. The larger it is, the greater the difference in text style between the source domain and the target domain, the smaller the correlation, and the lower the accuracy of text migration.

[0040] S5: Based on the topic relevance and the difference, determine the attention weights of different topics between the source domain and the target domain, and transfer the source domain text dataset to the target domain.

[0041] The greater the topic correlation between different topics across fields, the stronger the text information correlation of the corresponding topics in transfer learning, which helps to improve the adaptability of the model to the target field. Therefore, a larger attention weight is set; the greater the difference in text style formed by the combination of different contexts across fields in the training set, the smaller the role of the model in transfer learning in the target field, and the smaller the corresponding attention weight.

[0042] Therefore, according to the topic relevance and the difference, the attention weights of different topics between the source domain and the target domain are determined. The specific process is as follows: The attention weight between topic m in the source domain and topic n in the target domain The expression is: ; In the formula, Indicates the difference between the source domain and the target domain; represents the topic relevance between topic m in the source domain and topic n in the target domain; exp( ) represents an exponential function with a natural constant as the base; Represents the normalization function.

[0043] According to the attention weights of different topics between the source domain and the target domain, it can be understood that if the difference between the source domain and the target domain is smaller, the topic correlation between topic m in the source domain and topic n in the target domain is greater, which means that the text information correlation is stronger in the transfer learning process, which helps to improve the adaptability of the model to the target domain. Therefore, the larger the attention weight between topic m in the source domain and topic n in the target domain, the larger the attention weight needs to be set to improve the accuracy of transfer learning; conversely, if the difference between the source domain and the target domain is greater, the topic correlation between topic m in the source domain and topic n in the target domain is smaller, which means that the text information correlation is weaker in the transfer learning process and the text style difference is greater. Therefore, the smaller the attention weight between topic m in the source domain and topic n in the target domain is.

[0044] Preferably, the attention weight extraction process diagram provided in this embodiment is as follows: Figure 2 shown.

[0045] Based on the topic relevance and the difference, an attention weight is obtained. Further, an attention mechanism layer in the feature alignment process in transfer learning is constructed through the attention weights of all topics between the source domain and the target domain.

[0046] The BERT model is used to train the source domain text dataset to obtain a pre-trained model; the source domain text dataset, the target domain text dataset and the pre-trained model are used as inputs of the transfer learning technology, wherein the attention mechanism layer is used to align the features of the text datasets in the source domain and the target domain, and the text generation model adapted to the target domain is output.

[0047] Among them, the attention mechanism layer in the feature alignment process in transfer learning is constructed by the attention weights of all topics between the source domain and the target domain. The BERT model and transfer learning technology are well-known technologies, and their specific principles and processes will not be repeated here.

[0048] At this point, this embodiment analyzes the semantic correlation between the source domain and the target domain, and adds an attention mechanism in the model transfer learning process to ensure accurate alignment between the text features of the source domain and the target domain, so that the model learns knowledge in the source domain from multiple perspectives, thereby improving the effect of transfer learning and further improving the performance of the text automatic generation model in text generation in the target domain.

[0049] Based on the same inventive concept as the above method, an embodiment of the present application also provides a big data platform for automatic text generation based on a large model, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of any one of the above-mentioned methods for automatic text generation based on a large model are implemented.

[0050] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0051] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0052] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for automatic text generation based on a large model, characterized in that: The method comprises the following steps: S1: In the big data platform, obtain text datasets in various fields and the topic-vocabulary distribution of the text datasets in various fields, and use the probability of each word under all topics to form the vocabulary vector of each word, where each field includes a source field and a target field; S2: Cluster all words in the text data set of each field, analyze the degree of membership between each word in each field and different cluster centers, determine the semantic relevance of each word in each field, and combine the similarity of the word vectors of any two words between the source field and the target field to determine the word relevance between any two words between the source field and the target field; S3: Based on the vocabulary association between any two cluster centers between the source domain and the target domain, and in combination with the probability distribution of the any two cluster centers under different topics, determine the topic association between different topics between the source domain and the target domain; S4: Based on the membership, obtaining the membership vector of each cluster in each field; determining the difference between the source field and the target field by analyzing the difference of the membership vectors of all clusters between the source field and the target field; S5: Based on the topic relevance and the difference, determine the attention weights of different topics between the source domain and the target domain, and transfer the source domain text dataset to the target domain.

2. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The fuzzy C-means clustering algorithm is used to cluster all vocabulary of text data sets in various fields, and the Euclidean distance between the vocabulary vector of the vocabulary and the vocabulary vector of the cluster center is used as the distance between the sample point and the cluster center in the objective function of the fuzzy C-means clustering algorithm.

3. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The method for determining the semantic relevance of each word in each field is as follows: In each field, the mean membership between each word and all cluster centers is calculated. Among the memberships between each word and all cluster centers, the membership that is greater than the mean membership is obtained and recorded as the feature membership. The cumulative sum of all feature memberships is taken as the semantic relevance of each word in each field.

4. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The expression of the vocabulary association between any two words in the source domain and the target domain is: ; In the formula, represents the lexical association between the i-th word in the source domain and the j-th word in the target domain; Represents the similarity between the vocabulary vector of the i-th word in the source domain and the vocabulary vector of the j-th word in the target domain; represents the difference in semantic relevance between the i-th word in the source domain and the j-th word in the target domain; Indicates a preset constant greater than 0.

5. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The method for determining the subject relevance of different subjects between the source domain and the target domain is as follows: All cluster centers are recorded as central words, and the topic relevance between topic m in the source domain and topic n in the target domain is The expression is: ; In the formula, represents the lexical relevance between the p-th central word in the source domain and the q-th central word in the target domain; represents the difference between the probability of the pth central word under topic m in the source domain and the probability of the qth central word under topic n in the target domain; represents the number of all central words in the target domain; Indicates a preset constant greater than 0.

6. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The membership vector of each cluster in each field is composed of the membership between all words in each cluster in each field and the cluster center of the corresponding cluster.

7. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The expression of the difference between the source domain and the target domain is: ; In the formula, Indicates the difference between the source domain and the target domain; represents the difference between the membership vector of the hth cluster in the source domain and the membership vector of the kth cluster in the target domain; N represents the number of all clusters in the target domain; norm() represents the normalization function.

8. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The process of determining the attention weights of different topics between the source domain and the target domain is as follows: The attention weight between topic m in the source domain and topic n in the target domain The expression is: ; In the formula, Indicates the difference between the source domain and the target domain; represents the topic relevance between topic m in the source domain and topic n in the target domain; exp( ) represents an exponential function with a natural constant as the base; Represents the normalization function.

9. The method for automatically generating text based on a large model as claimed in claim 1, characterized in that: The method of transferring the source domain text dataset to the target domain includes: Constructing the attention mechanism layer in the feature alignment process in transfer learning through the attention weights of all topics between the source domain and the target domain; The BERT model is used to train the source domain text dataset to obtain a pre-trained model; the source domain text dataset, the target domain text dataset and the pre-trained model are used as inputs of the transfer learning technology, wherein the attention mechanism layer is used to align the features of the text datasets in the source domain and the target domain, and the text generation model adapted to the target domain is output.

10. A big data platform for automatic text generation based on a big model, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the large model-based automatic text generation method are implemented as described in any one of claims 1-9.

Images